conference programme book - European Survey Research ...

3Local Organiser‘s WelcomeDear Colleagues,We both are thrilled to welcome you in Lausanne for the 4th conference of the European Survey ResearchAssociation (ESRA). The number of participants exceeded all expectations and we thank you for the greatinterest to come to Lausanne for this big event. With such a large number of presentations, we are proud tooffer you a program that is of high quality, diverse, with contributions from Europe and beyond that covervarious disciplines.FORS, the Swiss Centre of Expertise in the Social Sciences, is the local organizing institution. It has a strongfocus on methodological research as well as survey data collection and dissemination. FORS is involved in theinternational projects European Social Survey (ESS) and International Social Survey Programme (ISSP) and representsSwitzerland on the Council of European Social Science Data Archives (CESSDA). In addition, the SwissHousehold Panel of FORS is part of the Cross National Equivalent File (CNEF). As a national centre of expertise,FORS also serves as a liaison office between Swiss and international research in the social sciences.We received a lot of help with the preparation for this big event. First of all, we are grateful to the Universityof Lausanne for offering their beautiful lakeside facilities as a location for the conference, and for the supportwith all the logistics we received from Unibat and Unicom. From ESRA we thank, of course, the conferencechair Patrick Sturgis and Daniel Oberski for all their help with the website and ESRA database. We are very gratefulto Kaisa Puustinen (University of Southampton), who was willing to organize the exhibition and managethe all other sponsoring activities. We extend our thanks also to Gosia Turner (University of Southampton) forher help with correspondence and registration.From FORS we are much obliged to Denise Bloch for her assistance with all administrative matters in Lausanne,Corine Bolle for her help with financial issues, and Isabelle Renschler for her advice on public relations.We thank Stephanie Kernich (University of Zurich) for her work on the layout of the program book.Finally, we thank the Swiss National Science Foundation, the city of Lausanne and the canton of Vaud for theirfinancial support to this edition of the ESRA conference. Of course we are also grateful for the contributionsof all our sponsers.We hope that you will enjoy the conference and the unique sites this city and region have to offer!Marieke Voorpostel and Kathrin KissauThe local organisers from FORS

5A Welcome from the President of ESRADear ParticipantsI am looking forward, with great anticipation, to the fourth conference of the European Survey Research Association(ESRA). The first three conferences were already a big success but I think that this one in Lausannewill be the best so far. We started in 2005 in Barcelona with 200 papers. In Prague there were 300 paperspresented, in Warsaw 400 and now we jump ahead to nearly 600 presentations. This increase makes it clearthat the creation of ESRA was necessary in order to meet the demand for dissemination of survey research inEurope. It seems that there is a strong and growing need for this type of forum, which brings together surveyresearchers, methodologists, and users of survey research.There has been a discussion within the board of ESRA about the size of these meetings: should we keep itsmall or not. The problem is that it is very difficult to specify criteria for selection of the best papers. It isalso preferable, in many ways, for ESRA to be an open organization which gives a wide variety of people atall career stages, the opportunity to present their work in a friendly and welcoming environment. So far wehave been very fortunate that we have received very few weak papers at the submission stage. We hope andexpect this feature of our biennial meeting to continue in Lausanne.I would also like to mention the other major activity of ESRA: the journal Survey Research Methods (SRM).Like the conference, our journal has turned out to be a big success, with 60 paper downloads made from thewebsite every day – quite an achievement for a journal that is in only its fifth year of publication. We hopethat the journal contributes to improving the quality of your work and of survey research in general. I wouldlike to congratulate Rainer Schnell, Peter Lynn and the editorial board of SRM for their excellent achievementin establishing SRM as a key journal in the field in so short a space of time.I would like to finish this short note by paying tribute to the people who have realized this ESRA conference forus all. First, I must mention Patrick Sturgis, the Conference Chair, and his team who have organized the conferenceand compiled the scientific programme. I would also like to extend my gratitude to FORS (the SwissCentre of Expertise in the Social Sciences), particularly Peter Farago, Kathrin Kissau and Marieke Voorpostel,for the excellent job they have done in handling the unanticipated 30% increase in participants this year. I amalso grateful to the University of Lausanne, for providing such a wonderful venue for our conference. Togetherthe conference chair and the local committee have put together a varied and stimulated programme in thebeautiful location of Lausanne and I hope you will join me in congratulating them for their effort.Finally, I would like to thank all participants of the ESRA conference for their contributions in presenting papersand coordinating sessions. I wish you an interesting and enjoyable conference.Willem SarisPresident of ESRA

9Exhibition: There are a number of organisations that are presenting their field of expertise in the exhibitionarea in the hallway connecting the Amphipôle and Amphimax buildings. The exhibition will be open duringthe conference from 9:00 to 17:30 and on Friday until 12:00. Exhibitors are:- The National Centre for Research Methods NCRM- FORS, the Swiss Centre of Expertise in the Social Sciences- SAGE Publications- The German Socio Economic Panel Study (SOEP)- cApStAn Linguistic Quality Control- Economic Development Initiatives (EDI) Limited- German Data Forum (RatSWD)- University of Duisburg-Essen- John Wiley PublishersLayout of conference building Amphipôle

10Layout of conference building Amphimax (level 3, ground floor when viewed from the lake side)Layout of conference building Amphimax (level 4)

12TNS is a world leading provider of customised political, social and economic research services. Wecover elections and are delivering insight to assist decision‐makers in a wide range of social policyissues ‐ education, health, social services, environment, employment, parenting and child protection,public transport, justice, immigration and community integration.Our clients include public sector and government bodies, political parties, media, non‐governmentalorganisations, major international and national institutions, including universities and scientificestablishments.Each year we conduct millions of interviews to help our clients understand how the public is thinkingor reacting to major issues on a national or international level.The Global P&S CommunityTNS Political & Social has more than 500 dedicated politicaland social researchers working across 40+ countriesenabling us to better understand complex multinational andmulticultural contexts.Each year, we conduct millions of interviews to help ourclients understand and track public opinion on a local andinternational level.Specialist insightsWe have specialist expertise in the following key areas:Behaviour Change and SocialMarketingWe assist governments and NGOs to plan,implement, and evaluate a wide range ofBehaviour Change and Social Marketingprogrammes worldwide.Political and electoral researchOur political teams have an unmatchedrecord of accurate polling and votingintention results in many countries“The Eurobarometer helps European leaders to shape theirpolicies. It’s an important tool to get European citizens toexpress their voice and opinions regularly.”Strategic Qualitative researchWe assist governments and politicalorganisations to develop and test policy, toset political strategy, to develop and testcommunication.Programme evaluationWe use customised, pragmatic approachesto assess the impact and effectiveness ofgovernment programmes worldwide.For more information, please visitwww.tnsglobal.comInternational researchWe use cutting edge and uniquecoordination tools developed over the last35 years of working with internationalorganizations like the European Institutions(we run the Eurobarometer) and the WorldBank.

13Local InformationImportant telephone numbersPolice: 117; Ambulance: 144; Tourist Information: 0216137373; Taxi: 0844814814Exchange ratesSfr 100 = EUR 83,85 / USD 117,64 / GBP 73,00 (on June 16, 2011)LausanneLausanne is situated on the shores of Lake Geneva (French: Lac Léman) with a view of the Alps and the Juramountain range. It lies in the middle of the Lavaux wine region, which is one of the UNESCO World HeritageSites. Lausanne has two main sightseeing areas: The Upper Town, or Haute Ville, runs uphill from the mainrailroad station and includes the medieval old town. Interesting sites to visit are the cathedral, the St-Françoischurch, Château Saint-Maire (a former Episcopal palace), and the Tour de l‘Ale (a watchtower dating back to1340 that was once part of the city walls). Flon, a trendy warehouse district turned nightlife quarter, is just tothe west of the old town and the Place de l‘Europe. The Lower Town, or Haute Basse, more commonly knownas Ouchy (the name of the fishing village that once stood on the shoreline), is a district of parks, lakeside promenades,elegant 19th century resort hotels and a handful of museums such as the world-famous OlympicMuseum Lausanne.Travelling in LausanneLausanne has a very good public transport system with buses and 2 metro lines. Day tickets for the publictransport system are included in the visitor’s tax that guests need to pay in all hotels in Lausanne. Please askfor these if you do not receive them upon your arrival. If you need to buy a ticket at a bus or metro stop it isuseful to have coins ready as most machines do not accept bank notes.At the website of the Swiss Federal Railways SBB website you can search for all public transport connectionsfrom train stations in many European countries to Switzerland as well as train and bus connections withinSwitzerland.It is best to use the website of the Swiss Federal Railways SBB (http://www.sbb.ch/en/index.htm) to searchfor your connections when travelling in Lausanne. At this website you can enter city names, specific bus ortrain stops as well as addresses and the best connections from the closest public transport points will beretrieved.

14Area map of Lausanne with the metro linesNational Centre for Research Methods (NCRM)...................................................................................................................................................................................Promoting research methods in the social sciencesNCRM was set up in 2004 to improve thestandards of research methods across thesocial science community in the UnitedKingdom.NCRM conducts research and training inquantitative, qualitative and mixed methods.We publish working paper series andmethodological reviews, which are freelyavailable online through NCRM EPrints inhttp://eprints.ncrm.ac.ukWhat does NCRM do?• Methodological research• Training courses and events• Research Methods Festival, 2-5 July 2012 atSt Catherine’s College, Oxford, UKOrganisation - Hub & Nodes• Coordinating Hub at the University ofSouthampton• 19 Nodes since 2005• Phase 1-3 Nodes hosted by 12 universitiesacross the UK• Professor Patrick Sturgis is the Director ofNCRMWeb. http://www.ncrm.ac.uk Email. info@ncrm.ac.uk Twitter @NCRMUK

16SOEP is funded by the Federal Government (BMBF)and the State Governments via the Leibniz GemeinschaftSocio-Economic PanelSOEP | DIW BerlinMohrenstraße 58D-10117 Berlin+49 -30-8 97 89-292soemail@diw.dehttp://www.diw.de/soepThe Socio-Economic Panel –A Portrait of Changes in German SocietySOEP is a representative longitudinal survey of more than 20,000individuals in over 10,000 households in Germany that providesthe basis for a wide range of novel scientific analyses.Features:• individual longitudinal data surveyed annually since 1984• data on household composition (adults and children), livingsituation• possibilities for regional comparison• oversampling of immigrants and high-income households• SOEP data are integrated into internationally comparablepanel data sets such as the CNEF (Cross National EquivalentFile), which contains comparable panel data for Australia,Germany, Great Britain, Switzerland, and the United StatesTopics include• personality traits• birth, childhood, and youth• labor market participation and occupational mobility• biography and intergenerational mobility• social participation and time allocation• physical and mental health• housing and regional mobilityThe SOEP data are available to researchers in• income dynamicsGermany and abroad in SPSS, SAS, and STATA• life satisfactionformats for immediate use. Extensive documentationin English and German is available • survey methodologyonline.For details please refer to our website: www.diw.de/gsoep

17Social programmeWelcome receptionAll conference participants are warmly invited to the welcomereception of the ESRA conference on Monday, July18th 2011 at 18:00. Drinks and finger food will be served onlevel 3 of the Amphimax building on the campus of the Universityof Lausanne. A few words of welcome will be spokenby a representative of ESRA as well as a spokesmen of thecity of Lausanne. Registration for this event is not necessary,everyone is welcome!The registration and information desk will be open beforeand during the welcome reception.Conference dinner-cruiseThe ESRA conference dinner-cruise will take place on Wednesday evening, July 20th 2011, on Lac Léman.Departing from Ouchy, the scenic port of Lausanne, the Valais will take us on a three hour cruise across thelake providing us with splendid views of the Alps, the famous Lavaux vineyards and beautiful lakeshore townssuch as Montreux and Evian. On board the Valais we will enjoy a gourmet dinner with the best wines of theregion. The 2011 ESRA “Early Career Researcher Award” and the prize for “Outstanding Service to EuropeanSurvey Research” will also be awarded during the dinner.Directions: From the conference venue, take the metro (M1) back to the city centre, get off at the final stopFlon. Here change to the other metro (M2) going down to the lake, direction Ouchy. Get off at the final stop(Ouchy). After leaving the metro station, cross the street and go in the direction of the water front. You findthe harbour on your left. Also see the map of Lausanne for an overview.Embarking: 19:30; Departure: 19:50This special event is made possible by the generous support of Westat. Westat is a professional servicesorganization, headquartered in Maryland, USA, with an established reputation for quality research and abroad range of capabilities in statistical surveys, program evaluation, technical assistance, epidemiologicstudies, clinical trials, and information technology.Westat has conducted successful research studies in a wide range of subject areas: health, education, socialservices, transportation, housing, energy, the environment, human services, military human resources, andscience and technology. Westat is also a leader in designing and administering customer and employeesatisfaction surveys and providing consulting services and marketing research to meet client requirements.Westat is a full-service research corporation.

18Programme overview!!"#$"%&&'(#)'")*'+(Monday, 18. July,#-.%/0(12(345/(6*&'( ( 7##&( !%$'(189::;1?#"@(A#4"B'(C9(,45@*;5')'5(&#.'5B(D#"(B4")'/(.%@%(( 818( E(( >?#"@(A#4"B'(CC9(>4")'/(-#-"'BF#-B'9("'.4A@*#-0(G*%B(%-.(A#&F%"%G*5*@/! 81H! E(129::;1I9::( J'5A#&'("'A'F@*#-( ?%55( 1'BB*#-(-%&'( 7##&( !%$'(I9::;I9=:( K#-D'"'-A'(#F'-*-$( =H1(

19!6*&'( >'BB*#-(-%&'( 7##&( !%$'(1E9::;17R(S'-'"%5(RBB'&G5/( 021! =!!J'.-'B.%/0(N:(345/(6*&'( >'BB*#-(-%&'( 7##&( !%$'(I9::;1:9=:( O%/(N(;(B'BB*#-(B5#@(8( ( E=(( X)57&B&-).&#'7!&'!@9:;6

20!6*&'( >'BB*#-(-%&'( 7##&( !%$'(119::;1N9=:( O%/(N(;(B'BB*#-(B5#@(H( (

!!22!6*&'( >'BB*#-(-%&'( 7##&( !%$'(1E9::;16%OF)76*!@9:;6

23Day 1Tuesday 19 July1.1 Combining and Enhancing data ITo be held on July 19, 2011 from: 11:00 to 12:30, in room 315.1.Coordinated by: Mark Elliot - University of Manchester, United Kingdom1.1.1 Generang Employer-Employee-Data Using Populaon Surveys: Drop Out-Stages and Data QualityC. Gerhards 1 , A. Meyermann 1 , S. Liebig 2 , S. Edler 11 Bielefeld University, Germany; 2 Department of Sociology, Bielefeld University, GermanyThere is a growing stock of linked-employer-employee data in economics and social sciences. As most of thesedata rely on official stascs the main problem is, however, that they only offer restricted informaon eitheron the side of employees and their households or the employer. One way to overcome these restricons isto generate employer-employee-data using populaon surveys as a starng point and adding informaon onemployers stemming from an organizaon survey. But this ”employee-first-dual-survey” approach is relatedto a number of methodological problems, which are addressed in this paper. The empirical basis is a datasetwhich links the German general social survey (ALLBUS) from 2008 with an organizaonal survey conducted in2009. Employed ALLBUS-respondents were asked to provide the name and address of their employer. Theresulng list of employers made up the sample for the organizaonal survey...1.1.2 New developments in MTB, a record linkage toolbox for the social sciencesR. Schnell 1 , T. Bachteler 1 , J. Reiher 11 University of Duisburg-Essen, GermanyRecord linkage, i.e. bringing together data files from different sources, meanwhile is a common task in surveyresearch. As long as the data files contain correct merging keys (e.g. names or idenficaon numbers) theprocedure is rather trivial. However, if the merging keys are error prone, resort to specialized soware isinevitable.1.1.3 Synthec approaches to data linkageM. Elliot 1 , J. Reiter 21 University of Manchester, United Kingdom; 2 Duke University, United StatesOrthodox approaches to linking data aempt connect overlapping informaon sources through key or commoninformaon; record linkage is paradigmac here with K datasets of records being linked from key variables(which may be either formal or informal idenfiers). Such approaches, however hit problems when K>2.Crucially, at present, there are no well developed non-heurisc methods for resolving data-data divergancebetween the datasets. Here we consider different approach. Rather than viewing each of the K datasets asa structure which needs to be linked in to create a super-dataset, we consider each as a source of evidence

24 TUESDAY 19 JULYto be utalised in combinaon with synthec models to populate a new dataset. We consider various formsof data linkage and evaluate how well the approach we propose might fit the requirements of that form. Aspecificaon of the emprical tests required to asses the approach is then given...1.1.4 You Can Match My Data! Biasing Effect in the Use of Linked Administrave and Survey DataS. Bender 1 , M. Huber 2 , A. Schmucker 21 Research Data Centre (FDZ) of the Federal Employment Agency at the Instute for Employment Research,Germany; 2 Instute for Employment Research, GermanyMost quantave longitudinal research in social sciences is done with survey data. Surveys suffer from nonresponsein many ways, for example, coverage errors, unit and item-non-response or arion. In parcularanswers to retrospecve quesons in surveys oen imply gaps or incomplete details of remembered episodes.Furthermore retrospecve survey data oen do not cover a very long period. In order to correct for these errorsadministrave data can be linked to survey data. Though the administrave data have drawbacks too – e.g.small number of variables or me lag – they contain valid and exact informaon. By linking survey data withadministrave data, the data quality can be improved by creang a dataset that balances the disadvantages ofthe administrave and survey data using the advantages of these two different types of data...1.2 Challenges in health examinaon surveysTo be held on July 19, 2011 from: 11:00 to 12:30, in room 321.Coordinated by: Hanna Tolonen - Naonal Instute for Health and Welfare (THL), Finland1.2.1 Self-reported and measured height and weight in adolescents: biases and social gradientS. Legleye 3 , S. Spilka 1 , F. Beck 2 , R. Guignard 21 OFDT, France; 2 INPES, France; 3 INED, FranceHeight and weight are simple but highly useful data for describing the health status of the populaon, especiallycombined into the Body Mass Index (BMI). There have been quite few validaons of these reports inadolescent populaon. Nevertheless, during adolescence, the body can change in a very short a period of meand adolescents are not always aware of it and willing to measure it. Do they report their height and weightaccurately when it is collected by self-administered quesonnaire? Are there biases according to gender andfamily socioeconomic status? Knowing the difference between reported and measured BMI could allow differenalcorrecons of the self-reported data in order to provide reliable prevalences of obesity in adolescentsurveys.1.2.2 Standardizaon of health examinaon surveys in EuropeH. Tolonen 1 , P. Koponen 1 , J. Heldal 2 , K. Kuulasmaa 11 Naonal Instute for Health and Welfare (THL), Finland; 2 Stascs Norway, NorwayHealth examinaon surveys (HES) are populaon surveys with physical measurements, biological samples andquesonnaire items. For many health indicators, they are the most reliable and oen the only data sourceto monitor prevalence and trends. Obtaining comparable informaon between surveys requires standardiza-on of the survey protocol.nHESs have been carried out in Europe since the 1960’s without European levelstandardizaon. The European Health Examinaon Survey (EHES) Pilot project was started in 2009. It aims toprepare a standardized survey protocol for naonal HESs and conduct pilot surveys in 13 countries. The pilotsurveys include a minimum of 200 parcipants per country. For the full-size naonal HES, a sample of at least4000 people aged 25-64 years per country is recommended. The EHES Manual has the standardized surveyprotocol, which provides detailed guidelines for the enre survey process...

TUESDAY 19 JULY 251.2.3 The Italian Health Examinaon Survey: Time Trends of CVD Risk FactorsL. Palmieri 10 , D. Vanuzzo 1 , C. Lo Noce 10 , C. Donfrancesco 10 , F. Vancheri 2 , L. Iacoviello 3 , C. Goldoni 4 , C. Caserta 5 ,A. Lopizzo 6 , N. Meloni 7 , M. Gaone 8 , A. Boccanelli 9 , F. Dima 10 , P. De Sancs Caiola 10 , P. Ciccarelli 10 , S. Vannucchi 10 ,S. Giampaoli 101 Centro per la Prevenzione Cardiovascolare, Azienda Socio-Sanitaria 4 Medio Friuli and Agenzia Region, Italy;2 Azienda Ospedaliera S. Elia, Caltanissea, Italy; 3 Università Caolica, Campobasso, Italy; 4 Diparmento diSanità Pubblica AUSL, Modena, Italy; 5 Associazione Calabrese di Epatologia, Reggio Calabria, Italy; 6 Unità Operavadi Cardiologia Medica, Ospedale San Carlo, Potenza, Italy; 7 General Praconer, Nuoro, Italy; 8 FondazioneS. Maugeri, Veruno, Novara, Italy; 9 Ospedale S. Giovanni Addolorata, Rome, Italy; 10 Istuto Superiore di Sanità,ItalyIntroducon: Health Examinaon Survey (HES) is appropriate to assess prevalence of chronic diseases anddistribuon of main risk factors in the general populaon. HES validity depends on sample representaveness,parcipaon rate, and use of standardized methodologies. Italian periodic HES-Osservatorio EpidemiologicoCardiovascolare preliminary results for 2008-11 compared with those from HES 1998 are here presented.1.2.4 Health Survey for England: Standardising ProtocolsC. Robinson 11 Naonal Centre for Social Research - NatCen, United KingdomThe Health Survey for England (HSE) is a series of annual surveys commissioned by The NHS Informaon Centrefor health and social care involving both health interview and health examinaon elements. The HSE isdesigned to provide high quality data representave of the general populaon living in private households inEngland, including adults and children. It is carried out by the Naonal Centre for Social Research (NatCen)and the Department of Epidemiology at University College London (UCL). The HSE consists of an interview ofparcipants’ health and health related behaviours. Parcipants’ height and weight measurements are alsotaken during the interview. The interview is followed by a visit from a qualified nurse to collect further measurements,biological samples and details of any medicaons taken.1.3 What Do Survey Researchers Want from Data Archives these Days? ITo be held on July 19, 2011 from: 11:00 to 12:30, in room 319.Coordinated by: Peter Granda - University of Michigan – Inter-university Consorum for Polical and SocialResearch (ICPSR), United States1.3.1 Social Science Queson Database and Research ToolsA. Cousteaux 2 , X. Schepler 1 , L. Lesnard 21 Réseau Quetelet, France; 2 Sciences Po, FranceThe increase in the number of surveys archived and disseminated raises new challenges for data archives.Indeed, without new tools, the growth of the number of surveys makes it increasingly difficult for users toidenfy surveys that are relevant for them. Without new tools, combining different surveys with similar ques-ons to conduct comparave research is also increasingly a daunng task. Survey designers who wish to reusequesons asked in previous surveys also need tools to find similar quesons.1.3.2 Geng to know our users: a survey of researchers in SwitzerlandB. Kleiner 1 , E. Ferrez 1 , M. Bichsel 1 , F. Lorétan 11 FORS - Swiss Foundaon for Research in Social Sciences, SwitzerlandAs a relavely new instuon, FORS aims to develop and maintain es with social sciensts in Switzerlandwho form its client base. One key method for achieving this is to engage researchers periodically about their

26 TUESDAY 19 JULYdata needs and interests. Knowing how researchers work and what they need allows us to beer customiseour services, especially as the research culture and data service landscape shi connuously to keep up withtechnological developments and new capacies.1.3.3 Serving Data Users of the Naonal Longitudinal Surveys of YouthP. Baker 11 Ohio State University, United StatesThe U.S. Naonal Longitudinal Surveys have spanned many decades of change in data management and disseminaon.From main frame to web-based data download, the NLS has responded to the needs of mulplegeneraons of data users. This paper describes the integrated system used by CHRR at the Ohio State Universityto document, manage, and archive the Naonal Longitudinal Surveys of Youth (NLSY), with a focus onthe NLSY79, a naonal, intercohort survey started in 1979 that extends across two generaons. CHRR hasdeveloped a user interface equipped with search and retrieval soware that allows users to peruse the topicalcontent of a given cohort, select variables of interest, create extract files, run descripve reports, and generatecontrol files to read the data...1.3.4 German BIBB-Research Data Centre and User NeedsD. Rohrbach-Schmidt 1 , H. Alda 1 , A. Friedrich 11 Federal Instute for Vocaonal Educaon and Training (BIBB), GermanyBased on a wide range of surveys on educaon, employment, further training and the transions betweenthem, the Research Data Centre of the Federal Instute for Vocaonal Educaon and Training (BIBB-FDZ) offersseveral services for researchers: -Standard access to firm- and individual-level data on skill aainment andits use -Documentaon of these data sets, i.e. a descripon of their central characteriscs, main issues andvariables, data collecon, anonymizaon, weighng and recoding etc. -Advisory service on data choice, accessand handling of the data, research potenal, scope and validity of the data -Supply of several data tools, e.g.standard measures and classificaons in the fields of educaon, occupaons, industries and regions, data forremote data access, references to relevant publicaons. - Assistance and support for vising researchers atthe safe centre at BIBB-FDZ in Bonn...1.4 Sensive Quesons and Social Desirability Bias: Theorecal Perspecves andData Collecon Strategies ITo be held on July 19, 2011 from: 11:00 to 12:30, in room 412.Coordinated by:• Ivar Krumpal - Universität Leipzig, Germany• Ben Jann - University of Bern, Switzerland1.4.1 Measuring social desirability effect on reported turnout using ”list experiment design” techniqueM. Comsa 2 , C. Postelnicu 11 ”Babes-Bolyai” University, Romania; 2 Babes-Bolyai University of Cluj Napoca, RomaniaRegardless of the cultural area in which it is performed or of the survey type, electoral research shows thatself-reports in surveys over-esmate voter turnout. Over-reporng can be the (combined) result of manyfactors: social desirability bias, memory related errors, acquiescence, and unrepresentave sample. Socialdesirability seems to be one of the most influencing factors. In order to reduce social desirability effect onself-reported turnout many techniques are used: indirect queson, self-administered quesonnaire, counterbiasing,weighng the responses based on a social desirability scale, randomized response, list experimentdesign (or item / unmatched count technique), etc. In this paper we focus on the ”list experiment design”

TUESDAY 19 JULY 27technique and we are using it for esmang social desirability bias related to self-reported turnout about2009 European Parliamentary elecons in Romania...1.4.2 Elicing illicit work. Item Count and Randomized Response Technique put to the test.A. Kirchner 2 , I. Krumpal 1 , M. Trappmann 2 , H. von Hermanni 11 Universität Leipzig, Germany; 2 Instute for Employment Research, GermanyWe address an ongoing debate how to assess sensive topics in telephone surveys. Examining three exisngmethods and implemenng one new method, we developed a module to measure illicit work and tested thisin two CATI studies (both conducted in 2010). In an experimental seng, we compare a double-list implementaonof the Item Count Technique (ICT) with direct quesoning as well as a forced-response implementaonof the Randomized Response Technique (RRT) with direct quesoning. In the first study (ICT; n=1.603), respondentswere selected from the German general populaon. In the second study (RRT; n=3.211), respondentsof two specific populaons were sampled from a register: employed persons and those qualifying for basicincome support in Germany, i.e. people depending on state transfer payments...1.4.3 Answering Sensive Quesons in Face-to-Face-Interviews using the Randomized Response- and ItemCount-Technique. Results from a Validaon SurveyF. Wolter 1 , P. Preisendörfer 1 , A. Skarbek-Kozietulska 11 Department of Sociology, Johannes Gutenberg-University Mainz, GermanyTo overcome the problem of misreporng on sensive quesons in surveys, the randomized response-technique(RRT) and the item count-technique (ICT) have been proposed. The idea behind both techniques is to limit respondents’incenves to misreport by a complete anonymizaon of the interview situaon.However, research has found mixed evidence on the performance of these techniques compared to conven-onal direct quesoning (DQ). Most importantly, there is a lack of validaon studies which are able to comparea validated ”true value” to the answers given in interviews.1.4.4 Sensive Queson Techniques in Online Surveys: An Experimental Comparison of Different ImplementaonsA. Diekmann 2 , M. Höglinger 2 , B. Jann 11 University of Bern, Switzerland; 2 ETH, SwitzerlandThe successful implementaon of special techniques for asking sensive quesons (such as the RandomizedResponse Technique) in self-administered online surveys poses several new challenges. A first crucial issue ishow to make respondents understand the techniques’ instrucons and appreciate the provided privacy protec-on. As no interviewer is present to explain the procedure and answer possible quesons, the risk of break-offor noncompliance with the instrucons is high if the procedure is not easily comprehensible. For the RandomizedResponse Technique a second issue is finding a suitable randomizing device. Oen used devices such asdices, coins, or banknotes may not be at a respondent’s immediate disposion and also require a mode shi,which increases non-compliance and non-response. Randomizaon based on innocuous quesons has thedisadvantage that usually there is only a limited set of suitable quesons...1.5 Data Quality in Special Populaon Surveys ITo be held on July 19, 2011 from: 11:00 to 12:30, in room 318.Coordinated by:• Susanne Vogl - Katholische Universität Eichstä-Ingolstadt, Germany• Marek Fuchs - Darmstadt University of Technology, Germany

TUESDAY 19 JULY 31During the last two decades the Netherlands instute for Social Research has accumulated a lot of knowledgeand experience in conducng surveys among non-western ethnic minories in the Netherlands. So far thesesurveys have always been unimode face-to-face surveys, but for the 2010 survey (SIM2010) an experiment hasbeen set up to compare this design with a sequenal mixed-mode survey.1.8 Using Paradata to Improve Survey Data Quality ITo be held on July 19, 2011 from: 11:00 to 12:30, in room 414.Coordinated by:• Oliver Lipps - FORS - Swiss Foundaon for Research in Social Sciences, Switzerland• Volker Stocké - University of Bamberg, Germany• Annelies Blom - Survex - Survey Methods Consulng, Germany1.8.1 Ethical Dilemmas in Dealing with Web ParadataM. Couper 1 , E. Singer 11 University of Michigan, United StatesAt the last ESRA conference, we presented the results of a vignee-based study in the Netherlands exploringhow best to inform respondents about the capture of paradata in Web surveys. We found that any disclosureof paradata capture reduced willingness to parcipate.1.8.2 Brevity is the Soul of Wit! The Effect of Quesonnaire Length on Item NonresponseV. Ludwig 1 , U. Krieger 2 , K. Pforr 11 University of Mannheim, Germany; 2 SHARE MEA University of Mannheim, GermanyLiming the number of quesons is an important task of quesonnaire design and also a problem well-knownto praconers. This paper looks at one major consequence of posing addional quesons: increasing itemnonresponse during the interview.We assume that respondents’ movaon and cognive resources decrease over the course of a long, demandinginterview. Respondents may then ease the growing burden of the interview by not formulang wellthought responses but a ”don’t know” answer instead. Therefore, the probability of giving invalid answersshould increase over the course of an interview.1.8.3 Improving Contact Times in Panel SurveysO. Lipps 11 FORS - Swiss Foundaon for Research in Social Sciences, SwitzerlandAlthough informave about opmal calling mes, the amount of explained variance in obtaining contact in theliterature is small. Exisng studies suffer from four shortcomings:• nonrandomized assignment of interviewersto calls, • lack of including household characteriscs beyond those available from census or interviewerobservaons, • lack of including unobserved household heterogeneity; and1.8.4 Usefulness of paradata in cross-naonal surveys: evaluaon of dwelling and neighborhood characteriscsH. Matsuo 1 , G. Loosveldt 11 Katholieke Universiteit Leuven, BelgiumIncreasing aenon is given to paradata in cross-naonal research. This is because paradata serves to monitorthe data collecon procedure and interviewer performance, collects detailed informaon on contact proceduresand some informaon on refusers. In addion, paradata collects dwelling and neighborhood charac-

32 TUESDAY 19 JULYteriscs among all sample units, but so far, less aenon is given. The purpose of this paper is three folded.In the first place, concepts of dwelling and neighborhood characteriscs are revisited and discussed. In viewof reviewing the operaonalizaon and measurement of these concepts and variables, quality of paradata(ESS Round 4 contact file) is assessed. The ulmate purpose of this paper is to idenfy list of acon-orientedrecommendaons: idenfy gaps between concepts and operaonalizaon, list area of intervenons on theexecuon of fieldwork acvies and interviewer training...1.9 Recent Developments in Modelling ArionTo be held on July 19, 2011 from: 11:00 to 12:30, in room 415.Coordinated by:• Roger Penn - Lancaster University, United Kingdom• Damon Berridge - Lancaster University, United Kingdom1.9.1 Changing Household Dynamics: Relaons Between Husbands, Wives and Children in Britain since1991Y. Chen 11 Lancaster University, United KingdomThe proposed paper will explore the effects of internal household dynamics upon labour force parcipaonof women and household income in Britain since 1991. The paper will model the effects of changing internalstructures of the household upon individual outcomes such as women’s labour force parcipaon, as well ascollecve outcomes in the form of aggregate household income. The labour force parcipaon outcomes willbe treated as ordinal in nature, and household income will be grouped into bands. The analyses will controlfor secondary explanatory variables which will include gender, marital status, age, educaonal aainment,socio-economic background and geographical region. The model will also incorporate birth of children andorientaon to gender roles. Inially, a separate model will be fied to each outcome (i.e. female labour forceparcipaon and household income)...1.9.2 Changing Atudes to Gender Roles in BritainR. Penn 11 Lancaster University, United KingdomThe proposed paper will examine changing atudes to gender roles in Britain between 1991 and 2007. It willpresent the results of mulvariate modelling of the factors that determine whether or not respondents have’tradionalisc’ or ’egalitarian’ views towards gender roles using data from the Brish Household Panel Study.The proposed paper will explore two different responses to quesons concerning atudes to gender. The firstis ’a husband’s job is to earn money; a wife’s job is to look aer the home and family’. The second is ’childrenneed a father to be as closely involved in their upbringing as the mother’.1.9.3 A Joint Model for Ordered Response and Binary Dropout Processes in the Social SciencesD. Berridge 11 Lancaster University, United KingdomData collected as part of large naonal longitudinal surveys such as the Brish Household Panel Study (BHPS)rounely include responses which comprise ordered categories. These can include Likert items or more generalordinal responses. Such longitudinal data present social sciensts with a range of stascal problems includingresidual heterogeneity, arion/dropout and state dependence.In this paper, we propose an approach which allows us to address each of these problems:

TUESDAY 19 JULY 331.9.4 A Model for Trivariate Ordered Response Data in the Social SciencesD. STOTT 1 , D. Berridge 11 Lancaster University, United KingdomData collected as part of large naonal longitudinal surveys such as the Brish Household Panel Study (BHPS)rounely include mulple Likert items which comprise ordered categories. For example, the BHPS includes arange of Likert items which ask respondents about their atudes towards gender roles. Three such items are:Both the husband and wife should contribute to the household income, a husband’s job is to earn money; awife’s job is to look aer the home and family, children need a father to be as closely involved1.9.5 Dropout in Web-based studies: MethodologyU. Reips 11 (1) University of Deusto, (2) IKERBASQUE, Basque Foundaon for Science, SpainThe talk will summarize more of a decade of Web research on dropout in Web-based studies, including Websurveys and Web experiments. Techniques of dropout management that once were generated from experienceor were derived theorecally have since been studied empirically. Such techniques include the seriousnesscheck, the warm-up technique, the high hurdle technique, the use of dropout to detect movaonalconfounding, and how the dropout can be modelled. The talk will explain these techniques, their empiricalbasis, and their relave importance in dropout management.A second poron of the talk will be devoted to using dropout as a dependent variable. Results from a series ofexperiments on factors influencing dropout will be presented.1.10 Nonresponse ITo be held on July 19, 2011 from: 11:00 to 12:30, in room 410.Coordinated by: Dominique Joye - University of Lausanne, Switzerland1.10.1 Determinants of Parcipaon in the German Naonal Health Survey Analyses from the ”GermanHealth Interview and Examinaon Survey for Adults” DEGSP. Kamtsiuris 1 , A. Goesswald 1 , M. Thamm 11 Robert Koch Instute, GermanyBackground: The willingness to parcipate in interview and examinaon surveys depends on different factors.In order to opmize the process of movang persons to parcipate and to analyse the representavity ofcollected data, it is of major interest to idenfy factors influencing the parcipaon rate. Data of the processof movang randomly selected persons to parcipate in the naonal health survey will be analysed. Methods:Since November 2008 the Robert Koch Instute is connuing the naonal health survey of 1998 with the”German Health Interview and Examinaon Survey for Adults” (DEGS). Within 3 years 7500 adults will be examinedcomprehensively in 180 sample points. Factors influencing the willingness to parcipate are idenfiedby mulple logisc regressions...1.10.2 Idenfying Sources of Non-response Bias: A Mixed-Mode Health Survey of a Low-income, CulturallyDiverse PopulaonD. McAlpine 1 , T. Beebe 2 , J. Kemmick Pintor 11 University of Minnesota School of Public Health, United States; 2 Mayo Clinic - Dept. of Health Sciences Research,United StatesLower response rates in surveys has resulted in increased efforts to encourage parcipaon, including creaveincenves, mixed-mode designs, refusal conversaons and more aggressive aempts to reach potenal respondents.While such efforts appear to increase response rates, lile is known about whether they reduce

34 TUESDAY 19 JULYnon-response bias. Understanding non-response bias requires informaon about the enre sampling frame.Even when this informaon is available, researchers have commonly grouped the sample into completes andnon-completes, which may mask differences between types of non-respondents. In contrast, this study examinestotal non-response bias, and bias associated with specific reasons for non-parcipaon.Data come froma mixed-mode survey (mail and telephone) of enrollees in a public health care program (RR 44.3%)...1.10.3 Experiences from 40 Years of the Cologne High School PanelK. Birkelbach 1 , A. Grauenhorst 21 University of Cologne, Research Instute for Sociology, Germany; 2 Universität zu Köln, Germany(CHiSP: 1969/70, 1985, 1996/97, 2010) We will present experiences from a longitudinal study of former Germanhigh-school students (German ”Gymnasium”) who have been interviewed four mes between ages 15and 56 and discuss strategies used to reduce panel mortality.1.11 Value and atude changeTo be held on July 19, 2011 from: 11:00 to 12:30, in room 413.Coordinated by: Guy Elcheroth - University of Lausanne, Switzerland1.11.1 Dynamics of social phenomena: Composite measures as a soluon to comparave stascal measuresK. Prevodnik 2 , V. Dolničar 2 , V. Vehovar 11 University of Ljublana, Slovenia; 2 University of Ljubljana, Faculty of Social Sciences, SloveniaSocial surveys are increasingly becoming indispensable resources for decisions on policies and acons. Veryoen, however, we also observe phenomena thru me, where interpretaons are very vulnerable towardsmethodological treatment. Usually, the empirical analyses focus only on basic comparisons of absolute (e.g.10% difference) or relave change (e.g. 2% increase), while somemes also the me distance is also involved(e.g. 5 year delay).1.11.2 Welfare atudes in a changing Europe: a class and gender perspecveJ. Ferreira de Almeida 1 , R. Brites 1 , A. Torres 11 Instuto Superior do Ciências do Trabalho e da Empresa - ISCTE, PortugalWell-being associated with a posive perspecve is a quality of life subjecve indicator. Knowledge aboutEuropean cizen’s well-being percepon becomes possible thru ESS round 4 data analysis. We have access,then, not only to a powerful instrument for quality of life comparave analysis between countries but also it ispossible to contribute for policy makers informed decisions on this issue. Subjecve well-being study increasingrelevance is due to the impossibility of reducing relevant social analysis to tradional economic measurementindicators, as for instance GDP per capita.Therefore, the aim of this paper is to map well-being atudes acrossEurope. Our main contribuon, however, is to illustrate how social class and gender do differenate individuals’welfare atudes.1.11.3 Pseudo-opinions in public opinion researchK. Reuband 11 University of Duesseldorf, GermanySurvey respondents oen give an answer when they are not able to give one. This has been amply studied sincePhilip Converse seminal paper on ”non atudes” (1964). There is less knowledge however about how manypeople present themselves as knowledgable on issues that do not exist. Drawing on local face-to face-surveys

TUESDAY 19 JULY 35in Germany an assessment is done of the prevalence of pseudo-opinion in the populaon and the factors thatmake an impact. Hereby also some of the studies on pseudo-opinions in the USA by George Bishop (2005) arereplicated.1.12 Cross-, Mul- and Transnaonal SurveysTo be held on July 19, 2011 from: 11:00 to 12:30, in room 340.Coordinated by: Marieke Voorpostel - FORS - Swiss Foundaon for Research in Social Sciences, Switzerland1.12.1 A mul-naonal study on the associaon between depressive symptoms, gender equity and educaonJ. Wartna 1 , J. Smits 2 , H. Tobi 31 Wageningen University and research Center, Netherlands; 2 Radboud Universiteit Nijmegen, Netherlands;3 University of Wageningen, NetherlandsDepression is a common mental disorder that can occur in persons of all genders, ages and backgrounds. In2004, depression was the 3rd leading cause of burden of disease worldwide and the projecon for 2030 is thatdepression will rise to the 1st leading cause (World Health Organizaon 2008). Using the World Health Survey2002 (WHS) of the World Health Organisaon, this study aims to invesgate self-reported depressive episodesworldwide. The focus of the study was on the relaon between depression and gender equity and educaonlevel, taking usual risk factors for depression (e.g. sex and age) as well as the regional or naonal context intoaccount. In addion, some methodological issues will be addressed.1.12.2 The effects of caring on self-assessed health status: A cross naonal comparisonP. Arnsberger 2 , F. Li 2 , U. Lynch 11 Queens University Belfast, United Kingdom; 2 University of Hawaii, United StatesThe purpose of this study was to compare the self assessed health status (SAHS) of carers of older adultsacross selected regions of three countries: California in the US (N= 1496), the Beijing and Shanghai areas ofChina (N= 485), and Northern Ireland in the UK (N=252). The study used secondary data sets collected ineach country within the past decade, and extracted data specific to carers of older adults. Equivalent variablevalues and their meanings (including both the outcome and independent variables) were determined acrossthe three countries by researchers from each country. A merged analyc data set was then analyzed to isolatesignificant predictors of SAHS using an adapted meta-analysis technique as well as ordinal probit analysis totest for possible cross naonal effects. Results across both types of analyses were then compared...1.12.3 Analysing Labour Market Effects of Cross-Border Outsourcing through EU Enlargement – A Transna-onal Survey in the Czech Republic and GermanyN. Litzel 1 , V. Hecht 1 , M. Moritz 1 , J. Schäffler 11 Instute for Employment Research (IAB), GermanyEconomic integraon typically goes along with a disintegraon of producon through outsourcing and offshoring.The consequences especially for the labour market are an issue of ongoing debate. The countries oforigin fear job losses. However, a range of models and studies show that the producvity gains lead to jobgrowth at the domesc plants. Other topics discussed concern the impact on different qualificaon groups,their wage levels, job volality and task structure in companies both in the country of origin and in the targetcountry. The discussion is even more eager where high-wage countries share a common border with low-wagecountries.

36 TUESDAY 19 JULY1.13 Human Values ITo be held on July 19, 2011 from: 14:00 to 15:30, in room 415.Coordinated by:• Eldad Davidov - University of Zürich, Switzerland• Peter Schmidt - University of Giessen, Germany• Constanze Beierlein - GESIS - Leibniz Instute for the Social Sciences, Germany1.13.1 Congruence and Performance of Three Value Concepts in Social Research: A Comparison of Inglehart’s,Welzel’s and Schwartz’s Approach in one SurveyT. Beckers 3 , P. Siegers 1 , A. Kuntz 21 University of Köln, Germany; 2 University of Cologne, Germany; 3 University of Düsseldorf, GermanyResearch in values has gained increasing aenon in cross-cultural psychology and comparave social research.The availability of large scale comparave surveys provides manifold opportunies for researchers to study theeffects of values on polical, social, and moral atudes. Two value concepts have become dominant in thefield of social research: (1) Schwartz’s theory of basic human values which, through the Portraits of ValuesQuesonnaire (PVQ), are included in the European Social Survey. (2) Inglehart’s postmaterialism and Welzel’sextension to the self-expression values which are part of the World Values Survey/European Values Study. Twoquesons are relevant to advance research in values: (1) Is it possible to convert the different concepts andmeasurements of values from one to another (i.e...1.13.2 Schwartz’s Theory on Human Values and Trust in Instuons: A Mullevel Test of Cross-CountryEquivalenceD. Morselli 3 , D. Spini 2 , T. Devos 11 San Diego State University, United States; 2 University of Lausanne, Switzerland; 3 ISS, Université Lausanne,SwitzerlandResearch based on Schwartz’s theory of human values can be organized according to three main focuses: Researchershave either (1) tested the cross-cultural invariance of the value structure, (2) studied how valuepriories relate to a variety of social atudes, or (3) examined societal differences in value priories. The aimof the present research was to integrate these three aims using a mullevel methodology. More precisely, ourgoal was to demonstrate that structural equaon modeling associated to mullevel regressions represents apowerful tool to examine innovave cross-cultural research quesons. Two studies on the relaonship betweenvalue types and trust in instuons were carried out. Study 1 compared results of a series of mullevelMulple Indicators and Mulple Causes (MIMIC) models with random slopes on four internaonal datasets...1.13.3 Deviant value structures – random or systemac?M. Janik 1 , W. Bilsky 11 University of Münster, GermanyRecent theory-based MDS analyses of values data from the European Social Survey (ESS, rounds 1-3) widelycorroborate Schwartz´s structural model of human values (Bilsky, Janik& Schwartz, 2010). However, data of some countries showed minor deviaons from the hypothesized valuestructure. While possibly originang from random fluctuaons, these deviaons may also reflect methodologicalarfacts or culture-specific paerns when observed repeatedly. Therefore, we analyzed the recentlypublished values data of ESS round 4 and focused on the value structures of those ESS-countries, for whichdeviaons were found in our prior analyses. In this paper, the results of these analyses are outlined, andrecurring structural deviaons are discussed in the light of compeng explanaons.

TUESDAY 19 JULY 371.13.4 Tesng the Circumplex Structure of Human Values: A Meta-Analycal Structural Equaon ModelingApproachH. Steinmetz 1 , R. Isidor 11 University of Giessen, GermanySchwartz’ theory of human values has found widespread interest in the social sciences. One central part ofthe theory is that the 10 proposed value types (i.e., achievement, power, self-direcon, hedonism, smulaon,benevolence, universalism, conservasm, security, and tradion) form a circumplex structure that reflects thecongruence versus conflicts of the respecve goals which are associated with each value type. The presentstudy applies a meta-analycal structural equaon modeling approach to test the circumplex structure. Moreover,the study explores in how far the circumplex structure varies with the used samples (e.g., culture) andmethodological characteriscs of the studies (e.g., measure). The meta-analysis comprised 318 matrices withthe correlaons among the 10 value types gathered from 86 studies as well as the European Social Survey(overall N = 251,239)...1.14 The development and validaon of psychological short scales and their benefitsfor survey research ITo be held on July 19, 2011 from: 14:00 to 15:30, in room 413.Coordinated by:• Beatrice Rammstedt - GESIS - Leibniz Instute for the Social Sciences, Germany• Jürgen Schupp - SOEP - German Instute for Economic Research (DIW Berlin), Germany1.14.1 Short Assessment of the Big Five: Robust Across Survey Methods Except Telephone InterviewingD. John 1 , F. Lang 4 , O. Lüdtke 5 , J. Schupp 2 , G. Wagner 31 Instute of Psychogerontology, Erlangen, Germany; 2 SOEP - German Instute for Economic Research (DIWBerlin), Germany; 3 DIW Berlin, Germany; 4 Instute of Psychogerontology, Germany; 5 HU Berlin, GermanyWe examined measurement invariance and age-related robustness of a short 15-item Big Five Inventory (BFI-S) of personality dimensions, which is well suited for applicaons in large-scale muldisciplinary surveys. TheBFI-S was assessed in three different interviewing condions: computer-assisted or paper-assisted face-to-faceinterviewing (PAPI, CAPI), computer-assisted telephone interviewing (CATI), and a self-administered quesonnaire.Randomized probability samples from a large-scale German panel survey and a related probability telephonestudy were used in order to test method effects on self-report measures of personality characteriscsacross early, middle, and late adulthood. Exploratory structural equaon modeling (ESEM) was used in orderto test for measurement invariance of the Five-Factor Model of personality trait domains across differentassessment methods...1.14.2 The Development of a Short Scale Assessing Locus of ControlA. Kovaleva 1 , C. Beierlein 1 , C. Kemper 1 , B. Rammstedt 11 GESIS - Leibniz Instute for the Social Sciences, GermanyLocus of control defines a personal belief about whether outcomes of behavior are determined by one’s aconsor by forces outside of one’s control. This personality trait appears to influence human behavior across a widespectrum of social situaons. Due to its direct link to educaon, common health, overall life sasfacon etc.it is interesng not only for psychological but also for sociological research quesons.1.14.3 Measuring the construct of Opmism-Pessimism with single item indicatorsC. Kemper 1 , A. Kovaleva 1 , C. Beierlein 1 , B. Rammstedt 1

38 TUESDAY 19 JULY1 GESIS - Leibniz Instute for the Social Sciences, GermanyDeveloping one item measures of reflecve psychological constructs with sufficient psychometric quality forsurvey research may be considered a crical research endeavor. By using the construct of Opmism-Pessimismas an example we demonstrate that this approach is possible and worthwhile. Opmism and Pessimism aredefined as generalized posive or negave expectancies concerning future events. Usually, both are measuredby a set of many items. As most surveys operate under severe me and/or monetary constraints we developedtwo single item indicators, one item to measure Opmism and one item for Pessimism.1.14.4 Assessing Schwartz’ global human values dimensions using a short scale for survey researchC. Beierlein 2 , C. Kemper 2 , A. Kovaleva 2 , B. Rammstedt 2 , S. Schwartz 11 Hebrew University of Jerusalem, Israel; 2 GESIS - Leibniz Instute for the Social Sciences, GermanySurvey researchers have become increasingly interested in basic values in recent years as evidence that the10 human values proposed by Schwartz predict various polical and social atudes and behaviors across cultures.The ten values are usually assessed with the 40 item Portrait Value Quesonnaire (PVQ; Schwartz etal., 2001) or the 57 item SVS (Schwartz, 2006). However, social surveys oen require shorter, valid, and reliableinstruments due to limited resources of me and money. In order to construct such a short scale, thenumber of constructs assessed has to be reduced. Therefore, instead of assessing all ten values, we selectedPVQ-items that serve as good indicators for the four global value dimensions (Schwartz and Boehnke, 2004):Self-Enhancement, Self-Transcendence, Openness to change, Conservaon. We measure these four poles withone item from each of the two to three values that combine to form that pole...1.15 The process of quesonnaire design in a cross-naonal perspecveTo be held on July 19, 2011 from: 14:00 to 15:30, in room 340.Coordinated by:• Evi Scholz - GESIS - Leibniz Instute for the Social Sciences, Germany• Jessica Forn - GESIS - Leibniz Instute for the Social Sciences, Germany1.15.1 The ISSP Model of Mulnaonal Quesonnaire DesignN. Lewin-Epstein 11 Tel Aviv University, IsraelAs the methodology of populaon surveys establishes itself in a growing number of countries, mulnaonalcomparave surveys are becoming the primary mode of comparave research. The presentaon will discussthe ISSP which is an ongoing internaonal collaboraon currently encompassing instuons from 46 countries,all commied to comparave survey research. Member instuons of the ISSP annually field an agreed uponsurvey module agreed upon by majority vote and have been doing so for 25 years. The ISSP collaboraon israther unique in adopng a very democrac and parcipatory model for its development of survey topics andquesonnaire design. It is of interest, therefore, to introduce to a wide audience the principles and processthat adopted by the ISSP and to point out its strengths and weaknesses to potenal users and collaborators...1.15.2 Challenges in comparave quesonnaire design: Illustraons from the European Social SurveyR. Fitzgerald 11 City University London, United KingdomA ’principle of equivalence’ or comparability underpins all high quality surveys. It demands that all membersof the relevant populaon have an equal (or at least a known, non-zero) probability of selecon, and that thequesons should have equivalent meaning for respondents. Such input equivalence is difficult to achieve in

TUESDAY 19 JULY 39a single-naon study but it poses further crucial difficules in comparave research with addional obstaclessuch as differing languages, cultures, and ’survey climates’.1.15.3 The CSES Quesonnaire Design ProcessD. Howell 1 , J. Forn 21 University of Michigan, United States; 2 GESIS - Leibniz Instute for the Social Sciences, GermanyThe Comparave Study of Electoral Systems (CSES; www.cses.org) is a cross-naonal research project withcollaborators from over sixty naons. Each five years, a new CSES quesonnaire module is developed by anelected Planning Commiee, aer considering suggesons and feedback from the project’s collaborator baseand user community of thousands. Quesonnaire design in a comparave framework presents a number ofchallenges not faced by researchers that are focused on single country designs. Involving so many stakeholdersalso presents special challenges. In this presentaon, the quesonnaire design process for the CSES willbe described, changes and improvements in the process over me will be idenfied, and the impact of theapproach on comparability and data quality will be evaluated.1.15.4 European Values Study – how to ensure comparability in an all European Survey over 30 yearsR. Luijkx 1 , L. Halman 11 Tilburg University, NetherlandsThe European Values Study (www.europeanvaluesstudy.eu) is a large-scale, cross-naonal, and longitudinalsurvey research program that started in 1981 and had in 2008 its 4th wave. Starng with 14 countries in thefirst wave, in 2008 the survey was held in all European countries. The focus is on a broad range of values and ahigh comparability is ensured both between countries and over me. How this was achieved, will be sketchedin this presentaon. Points of aenon will be the way the quesonnaire was developed and how a highqualityoriginal language documentaon can facilitate the quality of the field quesonnaires (more than 70 forthe 2008 survey). EVS is a collaboraon of EVS at Tilburg University, GESIS Data Archive for Social Sciences,CEPS Luxembourg, and research teams in the EVS member countries.1.15.5 Developing the SHARE Survey Instrument – An Iterave Process in a Mul-Disciplinary, Mul-Actor,Internaonal EnvironmentF. Malter 1 , C. Diemand 21 Mannheim Research Instute for the Economics of Ageing (MEA), Germany; 2 SHARE, MEA, University ofMannheim, GermanyThe Survey on Health, Aging and Rerement in Europe (SHARE) is a biennial longitudinal study of the populaonage 50 and older across 20 countries in Europe. To invesgate economic, social and health aspects ofageing, SHARE implements an ex-ante harmonized CAPI instrument and has integrated into the CAPI a SampleManagement System (SMS) that allows handling and documentaon of the enre contact protocol. TheSMS is also extensively used to monitor progress of fieldwork. An adjunct PAPI drop-off quesonnaire complementsSHARE’s tool set to accommodate country-specific needs and administer items that are sensive tothe face-to-face-mode.1.16 Surveying children and young people ITo be held on July 19, 2011 from: 14:00 to 15:30, in room 321.Coordinated by:• Lisa Calderwood - Centre for Longitudinal Studies, Instute of Educaon, University of London, UnitedKingdom• Kate Smith - Centre for Longitudinal Studies, Instute of Educaon, University of London, United Kingdom

40 TUESDAY 19 JULY1.16.1 Cognive Tesng of Web Quesonnaires with Children and Young People: Assessments of VisualDesign and Technical FeatureJ. D’Ardenne 11 Naonal Centre for Social Research - NatCen, United KingdomThere is a growing demand for research conducted with children and young people, this is driven by the factthat proxy informaon given by parents or teachers on children’s behalf is inaccurate and that children areincreasingly seen as actors in their own right (Sco; 1997). There is some evidence that young people prefercomputer based quesonnaires to paper ones (Mangunkusomo et al, 2005) and that computerised surveysconducted in schools are more cost efficient than paper based equivalents (Van Haum and de Leeuw, 1999).Despite this lile advice is available on how to design self-compleon quesonnaires, computerised or otherwise,for children and young people. This presentaon will share results from the cognive tesng of theAcve Young People Survey (AYPS); a web survey conducted in primary and secondary schools across Wales...1.16.2 Kids’ Life and Times: Using the internet to collect survey data from childrenK. Lloyd 11 Queen’s University Belfast, United KingdomKids’ Life and Times (KLT) is an annual survey of 10 and 11 year old children in Northern Ireland which beganin 2008. KLT is innovave in two main ways. Firstly, it invites all Primary 7-aged children in Northern Ireland(approximately 24,000) to parcipate. Secondly, it is an animated online survey carried out in school. As KLTis now entering its fourth year, the presentaon will reflect on its successes and failures. It will discuss thelessons learned for researchers interested in using this mode of administraon to collect data from children,as well as the impact of including standardised quesonnaires designed for pen and paper within an onlineanimated survey of this type. It will also provide a live demonstraon of the online version of the survey anddiscuss its development in the pilot year in 2008 through to the preparaon and delivery of the 2011 survey...1.16.3 Tesng quesons on a large-scale schools omnibus panel for the fih wave of the Millennium CohortStudyL. Calderwood 2 , L. Clements 1 , K. Smith 2 , E. Wallace 11 Ipsos MORI, United Kingdom; 2 Centre for Longitudinal Studies, Instute of Educaon, University of London,United KingdomIt is standard pracce on most large-scale quantave surveys to carry out field pilots to evaluate new data colleconinstruments. These are oen supplemented by cognive interviewing, parcularly for new quesons.While these queson tesng methods are undoubtedly beneficial, they are oen relavely small scale andlack the sample sizes sufficient to carry out quantave evaluaons of quesons or to test different versionsof quesons.1.16.4 Interviewing NLSY79 Children and Young Adults: Design Elements and Response in a LongitudinalIntercohort SurveyP. Baker 11 Ohio State University, United StatesThis paper reports on the design and content of the longitudinal surveys of the Children of the Naonal LongitudinalSurveys of Youth/1979 cohort (NLSY79). Elements of the NLSY79 Child surveys have been adapted andincorporated into other major U.S. surveys as well as naonal child cohort studies in Canada, Great Britain,mainland Europe, East Asia, and the Middle East, offering opportunies for comparave analyses. The paperdetails the informaon available over the life course for the NLSY79 children, delineang the modificaons inmodes of administraon as well as the content of quesons asked as children age up into adolescence and

TUESDAY 19 JULY 41young adulthood. The paper charts the intercohort parallels that are available in the NLSY79 survey for health,atudes, and behaviors that can be followed for mothers and their children...1.17 Public trust and instuonal legimacy: Cross-naonal analyses ITo be held on July 19, 2011 from: 14:00 to 15:30, in room 303.Coordinated by: Jonathan P. Jackson - London School of Economics, United Kingdom1.17.1 Explaining the fall in cizens’ trust in the ECB during the global financial crisisL. Stracca 1 , M. Ehrmann 11 European Central Bank, GermanyThe 2007-09 global financial crisis has been accompanied by a marked fall in the public trust in the ECB asmeasured by the European Commission’s Eurobarometer survey. This stands in contrast with the commonpercepon of central banks in general, and the ECB in parcular, having played a fundamental role in prevenngthe financial crisis from developing into a full-blown Great Depression. As a maer of fact, the fall in trust inthe ECB can be rather well explained based on previous, pre-crisis regularies. We find evidence that the fall intrust reflected the macroeconomic deterioraon, a more generalised fall in the trust in European instuons inthe wake of the crisis as well as the severity of the banking sector’s problems, to which the ECB was somehowassociated even though the ECB does not have direct supervisory responsibilies...1.17.2 Trust in instuons and protest parcipaon: a comparison of established and postcommunist EUdemocraciesA. Kirbis 1 , S. Flere 1 , M. Tavčar Krajnc 1 , R. Klanjsek 1 , M. Lavric 1 , B. Musil 11 University of Maribor, SloveniaIn the recent decades, protest parcipaon has emerged as one of the most widely accepted and praccedform of cizen engagement in the West. Many students of democracy believe that protest parcipaon is criticalfor a well-funconing and stable democracy and research suggests that protest parcipaon is one of themain characteriscs of a democrac public. Studies have also shown that the frequency of protest parcipa-on is on the increase in established democracies, while protest engagement in postcommunist democraciesis in decline. On the other hand, there has been a general decline in trust in public instuon in both old andnew democracies. The aim of our research is to analyze the role that trust in public instuons plays in protestpotenal in western and postcommunist EU democracies...1.17.3 Developing European Indicators of Trust in Criminal JusceJ. Jackson 1 , B. Bradford 1 , J. Kuha 1 , S. Stares 1 , S. Widdop 2 , R. Fitzgerald 21 London School of Economics, United Kingdom; 2 City University London, United KingdomA social indicators approach to trust in jusce recognises that the police and criminal courts need public supportand instuonal legimacy if they are to operate effecvely and fairly. We present in this paper theconceptual and methodological tools to devise indicators of trust and legimacy across Europe. First, we outlinethe conceptual roadmap for a comparave European analysis of trust in jusce. Second, we describe thedevelopment process of a 45-item module in Round 5 of the European Social Survey that fields the survey indicators.Third, we present the findings from the quantave pilong of the indicators in the UK and Bulgariaand document the final wording of the measures...1.17.4 Race, parsanship, and polical trust in the United States from 1958 to 2008R. Wilkes 11 University of Brish Columbia, Canada

42 TUESDAY 19 JULYThis paper examines the interrelaonships between race, parsanship and polical trust over a 50-year periodin the United States. An ongoing uncertainty as to the nature of the effects of race and parsanship stems fromthe type of data used in studies of the socio-demographic bases of polical trust. Most studies only considerone or two me points, making it difficult to consider the role of the larger context. If trust is lower amongBlacks and/or Democrats in a given year this may be because the members of these groups have lower levelsof trust in the polical system. However, if, rather than reflecng trust in the system, trust reflects atudestowards current polical authories, then trust levels could depend on whether the party in power in a givenyear was Republican or Democrat...1.18 Mixed-Mode Surveys: A Total Survey Error Perspecve IITo be held on July 19, 2011 from: 14:00 to 15:30, in room 315.Coordinated by:• Don A. Dillman - Washington State University, United States• Edith De Leeuw - University of Utrecht, Netherlands1.18.1 Is a Mixed-Mode Approach to Reduce Arion in a Longitudinal Mental Health Study (NESDA) aGood Idea?A. Hoogendoorn 2 , F. Lamers 2 , C. Hoekstra 1 , B. Penninx 2 , J. Smit 31 Vrije Universiteit Amsterdam, Netherlands; 2 GGZ inGeest Instute of Psychiatry and Mental Health, Netherlands;3 Vrije Universiteit Medical Center Amsterdam, NetherlandsIn longitudinal studies arion may cause problems, since a selecve drop out of respondents will bias researchresults. In mental health studies it is well known that respondents (paents) who suffer more severely frompsychiatric disorders usually drop out earlier from the study. To minimize these problems, it is recommendedto make efforts to keep the arion as low as possible. In the NESDA study, the respondents who refusedto visit the interview site (clinic) were either offered an interview at home or a telephone interview, thuscreang a mixed-mode design. With the introducon of the mixed mode design we reduced arions rateswith about 10%. But since a reducon in arion does not necessarily reduce bias, we will not only evaluatethe response rate but also the response selecvity. From the perspecve of a researcher it is interesng toknow the implicaons of the mixed-mode design for their research results...1.18.2 Comparison of personal interviewing and self-administered quesonnaires: Effect on representa-veness and prevalence of selected health indicatorsA. Christensen 1 , O. Ekholm 1 , K. Juel 11 Naonal Instute of Public Health, DenmarkThe Danish Naonal Instute of Public Health, University of Southern Denmark has carried out naonal representavehealth interview surveys among the adult Danish populaon (16 years or older) in 1987, 1994, 2000and 2005. The main purpose of the surveys is to describe the status and trends in health and morbidity. Thedata have been collected via face-to-face interviews at the respondent’s home. However, in 2010 data werecollected via self-administered quesonnaires. In order to assess the comparability of the survey with the formersurveys, a small face-to-face interview survey (n=1,500) was carried out in addion to the main survey(n=25,000). The same quesonnaire was used in both surveys and the surveys were carried out in the sameme period. Both surveys were based on mutually exclusive random samples...1.18.3 The effects of mixed mode designs on simple and complex analysesP. Marn 1 , P. Lynn 21 City University London, United Kingdom; 2 University of Essex, United Kingdom

TUESDAY 19 JULY 43Survey researchers, who must choose between mixed mode and single mode surveys, currently have limitedevidence regarding implicaons for measurement and esmaon. While many studies have invesgated measurementeffects associated with the main survey modes, and the overall effect of conducng a survey in onemode rather than another, few studies assess the overall effect of a mixed mode design versus a single mode.1.18.4 Changing research methods in Ukraine: CATI or Mixed-Mode Surveys?V. Panioo 2 , N. Kharchenko 11 Kiev Internaonal Snstute of Sociology, Ukraine; 2 Kiev Internaonal Instute of Sociology, UkraineAccording to the last available industry survey, ESOMAR_Global-Market-Research-2009, face-to-face interview method has first given way to CATI, and later to on-linesurveys. In 2009, total amount of interview orders has made up about 12%; CATI – 18%; while online surveys –20%. Concerning Ukraine, face-to-face interviews are connue to dominate (over 50%), while phone surveystake about 30%, and online surveys just start to develop.1.19 Combining and Enhancing data IITo be held on July 19, 2011 from: 14:00 to 15:30, in room 315.1.Coordinated by: Mark Elliot - University of Manchester, United Kingdom1.19.1 Who is care-giving? Assessing the quanty of people who are engaged in voluntary care workT. Mika 1 , C. Czaplicki 21 German Pension Fund - Research Date Centre, Germany; 2 Mannheim Research Instute for the Economics ofAging, GermanySurvey data are increasingly oen linked with administrave data in order to improve on the one hand thedata quality and on the other hand the informaon content. One example is the linkage of the third waveof the Survey of Health, Ageing and Rerement in Europe (SHARELIFE) with longitudinal administrave datafrom the German Pension Insurance (SHARE-RV). Both sources contain informaon on care-giving yet gatherit in different ways. SHARE provides informaon about voluntary care in a broad sense, measuring whetherrespondents provided somebody praccal household help (cleaning, gardening, shopping) or personal care(washing, dressing). The longitudinal data of the pension fund contain informaon on care-giving acviesonly in the restricted concepon of the compulsory care insurance (personal care like washing or dressing)...1.19.2 An empirical evaluaon of methods for privacy-preserving string comparison in record linkageT. Bachteler 1 , R. Schnell 1 , J. Reiher 11 University of Duisburg-Essen, GermanyDue to the frequency of spelling and typographical errors, in praccal applicaons record linkage algorithmshave to use string similarity funcons. In many legal contexts idenfiers such as names have to be encryptedbefore a record linkage can be aempted. Therefore, algorithms for compung string similarity funcons withencrypted idenfiers are essenal for approximate string matching in private record linkage.1.19.3 Small area esmaon combining informaon from several sourcesJ. Kim 1 , S. Kim 21 Iowa State University, United States; 2 Stascal Research Instute, Stascs Korea, Korea, SouthSmall area esmaon improves the direct esmators using auxiliary variables. The auxiliary variables are oenobtained from Census or from other external sources. A stascal model is used to link the direct esmatorand the auxiliary variable. Other external sources can include other independent surveys and results fromother administrave data.

44 TUESDAY 19 JULYWe consider the area-level model approach to small area esmaon when there are several source of auxiliaryinformaon. A measurement error model can be used to incorporate the informaon from the census orfrom other independent surveys. Propensity score adjustment method can be used to account for the undercoveragein the administrave data. Using a generalized least squares method, the proposed method can beexpressed as a composite esmator. Esmaon of the mean squared error will be discussed. The proposedmethod is discussed1.19.4 The empirical test of fault-tolerant record linkage using encrypted self-generated codesA. Pöge 11 Bielefeld University, GermanyIn panel studies with sensible topics it is oen wishful to link respondents across panel waves using encryptedself-generated codes due to data privacy protecon. These codes are mostly generated by using part of theanswers to a number of quesons concerning personal me-stable aributes. E. g. first or last leers of theanswers (e. g. surnames etc.) are used (see e. g. Grube, Morgan und Kearney 1989, S. 159). Data linkage isthen done using these codes which somemes are encrypted addionally aerwards. With this method thecodes have to meet several requirements such as the following: - respondents have to answer consistentlyat each me point of measurement otherwise the codes of different panel waves are not idencal - recordlinkage has to be fault-tolerant because the last point is oen violated so that one has to deal with inaccuratecodes belonging together (see e. g. Galan, Siliquini, Cuomo et al...1.20 What Do Survey Researchers Want from Data Archives these Days? IITo be held on July 19, 2011 from: 14:00 to 15:30, in room 319.Coordinated by: Peter Granda - University of Michigan – Inter-university Consorum for Polical and SocialResearch (ICPSR), United States1.20.1 An integrated resource discovery landscape: Many routes in, many final desnaons?J. Kneeshaw 11 UK Data Archive, United KingdomUsers of the UK’s Economic and Social Data Service (ESDS) (www.esds.ac.uk) are interested in a variety ofsurvey resources: study-level informaon; individual survey collecons; survey datasets; fieldwork documents;variables and their derivaons; quesons and their roung; parcular modules/scales/instruments; and so on.These resources oen share metadata and can be linked but, tradionally, they haven’t been: users searchingfor studies/surveys/datasets use one catalogue; users searching for variables use another; users searching fordocuments/modules/quesons use yet another...1.20.2 Purposing your survey: archives as a market regulator, or how can archives connect supply anddemand?A. Katsanidou 2 , L. Horton 11 UK Data Archive, United Kingdom; 2 GESIS - Leibniz Instute for the Social Sciences, GermanyWhat do researchers need from archives? This queson covers two different types of researchers that encounterarchives. There are creators of data and then the re-users of data. These two groups have differentneeds but the archive is asked to mediate as a middle point or the connecng ligament between these twoaudiences. An effecve role for an archive is to support data creators in producing high quality data, metadataand documentaon collecon to facilitate the wide communies of data reuse and to achieve mulpurposereuse and value for public investment in research. This paper will discuss the opmizaon of such a trainingprocedure for primary data creators and the benefits they can extract from such training.

TUESDAY 19 JULY 451.20.3 What is the best way to distribute large cumulave datasets with mul-naonal data?H. Midtsæter 1 , H. Orten 11 Norwegian Social Science Data Services (NSD), NorwayRepeve mul-country surveys face a challenge in how to best distribute their data collecons over me.With each repeang round of the survey, the cumulave dataset expands and will eventually become toolarge to work with and distribute efficiently.1.20.4 Secure Data Service: an improved access to disclosive dataR. Ahami 11 UK Data Archive, United KingdomThe Secure Data Service is a secure environment funded by the UK’s ESRC to provide researcher access todisclosive microdata from remote locaons. The operaon is legally framed by the Stascs Act, 2007 whichallows access to the confidenal data for stascal purposes. This short paper introduces this new UK DataArchive service and proposed specificaons, as well as discussing the challenges facing data service providers.We expect that the infrastructure will meet the requirements of the data security model. The paper will alsohighlight the potenal issues in the operaon of such a service which we believe will be an exemplar for asecure remote access pracce.Keywords: Remote access, Data security, Citrix technology, Secure Data Service1.21 Using Paradata to Improve Survey Data Quality IITo be held on July 19, 2011 from: 14:00 to 15:30, in room 414.Coordinated by:• Annelies Blom - Survex - Survey Methods Consulng, Germany• Oliver Lipps - FORS - Swiss Foundaon for Research in Social Sciences, Switzerland• Volker Stocké - University of Bamberg, Germany1.21.1 Effects of revisits and ’so’ refusal conversion on data quality. A first analysis of data from the 5thEuropean Working Condions SurveyG. van Houten 11 Eurofound, IrelandThe European Foundaon for the Improvement of Living and Working Condions (Eurofound) carries out threerecurring Europe-wide surveys: the European Working Condions Survey (EWCS), the European Quality of LifeSurvey (EQLS), and the European Company Survey (ECS). Most recently, in 2010, the 5th wave of the EWCSwas implemented. As part of our efforts to produce high quality data, we collect a range of variables on thecontact sheets interviewers fill out at every aempt to make contact with the (potenal) respondents.This paper is aimed at gaining insight in the impact of increasing the number of revisits, as well as the conversionof inial ’so’ refusals, on the composion and therefore the quality of the sample of respondents included inthe survey.1.21.2 Proposed Indicators and Measures to Assess Interviewer Performance in CATI SurveysF. Laflamme 11 Stascs Canada, CanadaInterviewer Performance, defined in this paper as the ability of an interviewer to contact and convince respondents,is generally assessed by call centres in using descripve measures such as the number of completed

46 TUESDAY 19 JULYinterviews, the number of completed interviews per hour, etc. Other more comprehensive performance indicatorssuch as the cooperaon rate at first contact and Net Contribuon to Performance Index have beendeveloped over the past few years. However many factors might impact interviewers’ performance in a centralizedcall centre environment. In addion to the interviewer’s characteriscs and environmental factors,the type and porolio of cases called, the effort already put into these cases, the me the call is made andthe general producvity of the survey at the moment at which the call is made are some of these potenalinfluencing factors...1.21.3 The Ulity of Incorporang Paradata in Mulple Imputaon Models for Item NonresponseR. Young 2 , T. Sanders 1 , R. Schofield 11 Pennsylvania State University Survey Research Center , United States; 2 The Pennsylvania State University,United StatesIncorporang informaon about quesonnaire design features, survey implementaon, and interviewer characteriscsinto post-hoc adjustments for missing data has the potenal to help reduce nonresponse bias. Inthis paper we consider whether, and to what extent, adding paradata as auxiliary informaon in mulpleimputaon models of item nonresponse can improve the ulity of the procedure. We use data from the ConsumerHealth Survey (n = 5,577), which was collected through computer-assisted telephone interviews between2007 and 2008 and is naonally representave of the United States populaon. The survey quesonsincluded health topics such as diabetes, blood pressure, heart disease, asthma, and depression, as well as demographiccharacteriscs. Paradata about the quesonnaire design include: queson order, queson length,and queson difficulty...1.21.4 Agree or Disagree? Cognive processes in answering contrasve survey quesons.N. Kamoen 2 , B. Holleman 2 , H. van den Bergh 11 University of Amsterdam, Netherlands; 2 University of Utrecht, NetherlandsRespondents are more likely to disagree with negave quesons (This arcle is bad. Yes/No), than to agreewith posive ones (This arcle is good. Yes/No). Taking a cognive perspecve, two possible causes can bedisnguished for this answering difference. One: posive and negave quesons measure different atudes,and therefore differ in their validity. Two: contrasve quesons measure the same underlying atude, but thisatude is expressed differently to the response scale because of the wording: even though response oponslike yes and no are straight opposites, the answer yes to a posive queson is not equivalent to a no-answerto a negave queson. This implies that contrasve quesons are equally valid.1.22 Sensive Quesons and Social Desirability Bias: Theorecal Perspecves andData Collecon Strategies IITo be held on July 19, 2011 from: 14:00 to 15:30, in room 412.Coordinated by:• Ben Jann - University of Bern, Switzerland• Ivar Krumpal - Universität Leipzig, Germany1.22.1 Reducing Overreporng in the Voter Turnout QuesonE. Zeglovits 1 , S. Kritzinger 11 University of Vienna, AustriaInformaon on voter turnout is crucial when studying electoral behaviour in liberal democracies. However,electoral researchers are confronted with the problem of ’overreporng’, i.e. respondents report turnout butdid actually not vote. The social desirability of parcipang in an elecon is regarded as one of the explanaon

TUESDAY 19 JULY 47for this phenomenon. There have been a few aempts – most of them in the US context – to reduce overreporng(1) by introducing new ways of queson wording, and (2) by using list experiments (or item counttechniques). However, these more sophiscated forms have neither been tested in the European mulpartycontext, nor translated into other languages. In this paper, first, we aim at filling this gap by tesng the tools inan European, non-English context (namely Austria), and second, we test an addional new queson wordingto obtain an even more valid queson on turnout...1.22.2 Social desirability in prejudice research: Results from three empirical studiesS. Gosen 2 , S. Thörner 2 , P. Schmidt 1 , J. Leibold 31 University of Giessen, Germany; 2 DFG research training school ”Group Focused Enmity”, Germany; 3 Univerisityof Göngen, GermanyIn the context of a 10 year project in Germany, named ”group focused enmity” we try to find out the tendencyof respondents to answer social desirable. In this presentaon we want to report results of three empiricalstudies performed in 2009, 2010 and 2011 in Germany.1.22.3 An Approach to Understanding the Underreporng Tendency in the Item Count TechniqueT. Tsuchiya 2 , Y. Hirai 11 Tokyo Metropolitan University, Japan; 2 The Instute of Stascal Mathemacs, JapanThe item count technique, which is used to invesgate issues such as highly private or socially undesirablebehaviours, requests respondents to answer solely the number of applicable items from among a list of itemsincluding the target key behaviour. Although the item count technique is expected to elicit more truthful answersthan the convenonal direct quesoning technique, the esmates obtained from this technique regardingthe key behaviours are oen smaller than those yielded by the convenonal direct quesoning technique,even when the target items are sensive to the respondents. In this presentaon, we first show examplesof such failure cases conducted via face-to-face interviews for invesgang both sensive and non-sensivebehaviours among Japanese people...1.22.4 Interview Privacy and Social Desirability Effects: The World Mental Health Survey ExperienceZ. Mneimneh 1 , B. Pennell 11 University of Michigan, United StatesTo encourage accurate responses, many surveys that address sensive topics require that the interview takeplace in a private seng so that the interviewer is the only individual to hear the respondent’s answers. Maintainingsuch privacy is always a challenge but becomes more so in a cross-naonal study. This presentaonfocuses on the invesgaon of interview privacy measures across different cultures in a cross-naonal survey,the World Mental Health Survey Iniave (WMH). The WMH has conducted naonally or regionally representavesurveys in 30 countries around the globe.1.23 Data Quality in Special Populaon Surveys IITo be held on July 19, 2011 from: 14:00 to 15:30, in room 318.Coordinated by:• Marek Fuchs - Darmstadt University of Technology, Germany• Susanne Vogl - Katholische Universität Eichstä-Ingolstadt, Germany1.23.1 What protocol best improves the parcipaon rate to surveys for physicians in France?S. Legleye 2 , N. Razafindratsima 3 , C. Moreau 1 , A. Bohet 1 , N. Bajos 11 Ined-Inserm, France; 2 INED, France; 3 Instut Naonal d’Etudes Démographiques, France

48 TUESDAY 19 JULYThe parcipaon of physicians working in private offices in research studies is tradionally very low in France.In order to improve the parcipaon rate, we conducted a methodological randomised trial to assess thecost/effecveness of different enrolment protocols for use in a future survey exploring healthcare provider’satudes, knowledge and pracces regarding sexual and reproducve health.1.23.2 Implementaon of a Pracce Guideline for the Pain Management of Residents in Nursing Homes –tesng the effecveness by a cluster randomized trialS. Ellert 1 , A. Budnick 1 , M. Kölzsch 2 , I. Wulff 1 , S. Kalinowski 1 , R. Kreutz 2 , D. Dräger 11 Instut für Medizinische Soziologie der Charité, Germany; 2 Instut für Klinische Pharmakologie und Toxikologieder Charité, GermanyIntroducon: The intenon of the PAiN II study is to develop and implement a ”Pracce Guideline for theAppropriate Pain Management” for nursing home staff and general praconers. The second aim is to evaluatethe effecveness of this guideline. Methods: The design of this study will be a cluster randomized trial andit is conducted in eight nursing homes (NH) in Berlin and Brandenburg. In the intervenon group (four NH)the guideline will be implemented by trainings delivered to nursing home staff and praconers three mesduring the trial.1.23.3 Why telephone surveys are the best choice to conduct a quantave representave firm surveyI. Krause 11 University of Bielefeld , GermanyFirm representaves are not as easy to reach by telephone, by mail, or by leer as private persons. Usuallythe communicaon path is a selecve one. The interviewers have to pass gatekeepers to reach the personof choice, and it is impossible to contact these populaon personally without an appointment. Furthermorea smaller populaon and high research interest lead to increasing problems of accessibility. So firm representavesshould be considered as a special populaon. But what is the best method to carry out a surveyin these special populaon to get a representave sample? The proposed presentaon wants to discuss in acomparave manner the different survey methods that can be used in a firm survey...1.24 Nonresponse IITo be held on July 19, 2011 from: 14:00 to 15:30, in room 410.Coordinated by: Dominique Joye - University of Lausanne, Switzerland1.24.1 Anonymous and non-anonymous mail surveys in the general populaon. Their impact on responserates and response effects.K. Reuband 11 University of Duesseldorf, GermanyAccording to common pracce in mail surveys quesonnaires have a number on their first page in order tomake reminders of non-respondents more economical. However the number might be seen as a sign for lackof anonymity by respondents and affects their responses. No studies have been yet done and shown whatthe effects of these differences in methodology are.Bases on a split-design (with and without number on thequesonaire in four surveys of the general populaon in three German cies (in West and East Germany)it is analyzed whether effect on response rates and responses exists, in which subgroups and under whatcondions. The surveys were done by Dillman design with up to three reminders and contained non-sensiveand sensive quesons. The samples were drawn from the city register of residents 18 years and older.

TUESDAY 19 JULY 491.24.2 Comparison of parcipants and non-parcipants to the ORISCAV-LUX populaon-based study oncardiovascular risk factors in LuxembourgA. Alkerwi 1 , N. Sauvageot 1 , S. Couffignal 1 , A. Albert 2 , M. Lair 1 , M. Guillaume 21 Centre de recherche Public en Santé, Luxembourg; 2 ULG, BelgiumPoor response is a major concern in public health surveys. In a populaon-based ORISCAV-LUX study carriedout in Grand-Duchy of Luxembourg to assess the cardiovascular risk factors, the non-response rate was notnegligible. The aims of the present work were: 1) to invesgate the representaveness of study sample to thegeneral populaon, and 2) to compare the known demographic and cardiovascular health-related profiles ofparcipants and non-parcipants.1.24.3 Why do respondents drop-out from online surveys? Results from follow-up surveys in the GermanLongitudinal Elecon Study (GLES).J. Rossmann 3 , J. Blumensel 1 , M. Steinbrecher 21 MZES, University of Mannheim , Germany; 2 University of Mannheim, Germany; 3 GESIS - Leibniz Instute forthe Social Sciences, GermanyHigh drop-out rates are considered a major shortcoming of web surveys and considerably threaten data qualityif drop-out is systemac rather than random. However, despite growing scholarly aenon the knowledge onsurvey drop-out is sll fraconal. Previous research mainly addresses the impact of survey design, quesonwording, and characteriscs of the respondents on survey drop-out via ex-post stascal methods. The researchpresented here is innovave in that the respondents are asked directly about the reasons for droppingout, the interview situaon, psychological predisposions, as well as a reduced number of quesons on poli-cs and polical atudes in a series of follow-up surveys conducted subsequently to three consecuve onlinesurveys of the GLES. These follow-up surveys, featuring more than 300 interviews with drop-outs, allow for anenhanced understanding of the complex processes underlying the phenomenon...1.25 Collecng Physical Measure and Biomarker Data in Surveys ITo be held on July 19, 2011 from: 16:00 to 17:30, in room 318.Coordinated by:• Barbara Schaan - University of Mannheim, Germany• Mary Beth Ofstedal - University of Michigan, United States• Heidi Guyer - University of Michigan, United States1.25.1 Scaling it up: In-home biosocial data collecon on Understanding Society - the UK Household LongitudinalSurveyK. Deepchand 1 , A. Conolly 21 Independent Research Consultant, United Kingdom; 2 Naonal Centre for Social Research - NatCen, UnitedKingdomUnderstanding Society, led by the Instute for Social and Economic Research, is an ESRC funded study of thesocio economic circumstances of 100,000 individuals within 40,000 households across the UK. The study waslaunched in 2008 and since 2010, standard survey data has been augmented with physical measurements andbiological samples collected in home.1.25.2 Collecng saliva samples for DNA extracon from children and parents on the fih wave of theMillennium Cohort StudyL. Calderwood 2 , J. Heather 2 , N. Rose 1 , A. Thompson 1

50 TUESDAY 19 JULY1 Ipsos MORI, United Kingdom; 2 Centre for Longitudinal Studies, Instute of Educaon, University of London,United KingdomThe Millennium Cohort Study (MCS) is large-scale social survey following over 19,000 children born in the UKin 2000/1. So far there have been four waves of the study at 9 months, 3 years, 5 years and 7 years which haveincluded physical measurements, funconal tesng and collecon of biological samples by survey interviewersin a home seng.1.25.3 Challenges in recruing parcipants in a populaon-based survey including 24-hour urine collec-on: example from the Swiss Survey on Salt intakeA. Chappuis 1 , N. Glatz 2 , P. Vuisner 3 , V. Forni 2 , F. Paccaud 3 , M. Burnier 2 , M. Bochud 31 Instute of social and prevenve medicine, Hospital University Center, Lausanne, Switzerland; 2 NephrologyDivision, Hospital University Center (CHUV), Lausanne, Switzerland; 3 Instute of social and prevenve medicine,Hospital University Center (CHUV), Lausanne, SwitzerlandBackground and aim: We conducted a naonal populaon-based examinaon survey in nine Swiss centres,which aimed at esmang dietary salt intake using 24-hour urine collecon in the populaon aged 15 yearsand over. The survey started in 2009 and is sll ongoing. We present data of the canton of Vaud to illustrate thepraccal challenges encountered during the recruitment process. Methods: We chose a two-stage samplingstrategy similar to the one used during the Swiss Health interview Surveys. We sent an informaon leer to arandom sample of households from the Swisscom directory (fixed lines) and subsequently contacted them byphone. Aer having defined the household composion, we randomly selected a single individual to parcipatein the survey, within 8 predefined age and sex strata. The study included 2 hospital visits and a 24-hoururine collecon.Results: A total of 1729 households were contacted, 34...1.25.4 Change in Handgrip Strength and Overall Sasfacon with Life as Predictors of MortalityJ. Schupp 1 , J. Ambrasat 2 , G. Wagner 31 SOEP - German Instute for Economic Research (DIW Berlin), Germany; 2 Free University, Berlin, Germany;3 DIW Berlin, GermanySelf-reported measures of respondents’ health status are treated as weak measures of the objecve healthstatus of respondents. On the other hand in most surveys health status measured by self-reports and a lotof insights about socio-economic characteriscs of the local environment on health status are based on selfreportedhealth. Thus a validaon of self-reports is of interest for the research community in public health.In 2006 and firstly repeated in 2008 handgrip strength measurement was part of the German Socio-EconomicPanel (SOEP) a widly analyzed annual German household panel study. The handgrip data can be comparedwith other health and well-being indicators from the survey.We will discuss examplary comparisons of the predicve power of subjecve and objecve health indicators.1.26 Assessing the Quality of Survey Data ITo be held on July 19, 2011 from: 16:00 to 17:30, in room 410.Coordinated by: Jörg Blasius - University of Bonn, Germany1.26.1 The measurement of anomie and self-esteem in ESPAD quesonnaireS. KALOGERAKI 11 DEPARTMENT OF SOCIOLOGY, UNIVERSITY OF CRETE, GreeceThe European School Survey Project on Alcohol and Other Drugs (ESPAD) has become an important sourceof informaon in licit and illicit drug use among European adolescents. The main raonale of the projectis to monitor trends in substance use among students in Europe and to compare trends between countries

TUESDAY 19 JULY 51and between groups of countries. The paper crically examines the quality of specific survey items of theproject’s quesonnaire measuring anomie and self-esteem. These items are idenfied with important sourcesof measurement errors associated with the formulaon of the quesons and the response scales applied. Itis advocated that the baery form of agree/disagree negave and affirmave statements yields response biasassociated with acquiescence issues and cognive procedures, thus eliminates the quality, i.e., the reliabilityand validity of the measures under study...1.26.2 Fague in Payment Diaries – Empirical Evidence from GermanyT. Schmidt 11 Deutsche Bundesbank, GermanyIn this paper we analyze whether the recording behaviour of consumers keeping a payment diary changesover the diary period. Using data from a large study on the payment behaviour of German consumers wefind that individuals tend to report a higher number of transacons on the first day of the diary period than onsubsequent days. Contrary to exisng literature we also find that the number of small cash payments recordeddoes not decrease during the one-week diary period. Our findings indicate that shorter diaries may be enoughto reflect adequately the payment behaviour of all consumers. However, longer diaries also have their merits,especially when it comes to analysing subgroups of payment types or rare events.1.26.3 Follow-Ups and Data Quality in Mail Surveys on Sensive TopicsA. Skarbek-Kozietulska 1 , P. Preisendörfer 1 , F. Wolter 11 Department of Sociology, Johannes Gutenberg-University Mainz, GermanySince Dillman’s (1978) Total Design Method (TDM) it is a well-known fact that follow-ups increase the responserates in mail surveys. But do they increase also the quality of survey data? The most suitable way to answer thisqueson is conducng a validaon study. A mail survey which allows such a validaon has been successfullyrealized at the University of Mainz in Germany. The study is part of a broader research project ”Asking SensiveQuesons”, supported by the German Science Foundaon (DFG).1.26.4 Opmal Appropriateness Measurement in the context of the European Social SurveyI. Lamprianou 11 University of Cyprus, CyprusOpmal Appropriateness Measurement (OAM) is a general stascal method for the idenficaon of personswhose aggregated responses on a scale, survey quesonnaire or examinaon test might not be a valid indicatorof their true latent atude, trait or ability (Drasgow, Levine, and Zickar, 1996; Levine and Drasgow, 1988). Ithas long been known that some people may aempt to give socially desirable responses to some quesons,may answer randomly some quesons or may even copy their responses from another person. Idenfyingthese persons is meaningful because their responses may distort our results and because it is interesng toidenfy sub-groups which are more likely to give unreliable responses to surveys1.27 The Civil Society Index as a tool for cross-naonal comparisons. Methodologicalissues and substanve applicaonsTo be held on July 19, 2011 from: 16:00 to 17:30, in room 340.Coordinated by:• Michael Hoelscher - University of Heidelberg, Germany• Helmut Anheier - University of Heidelberg, Germany

52 TUESDAY 19 JULY1.27.1 A typology of state-civil society relaonships based on the raonales for the support of CSOs testedwith the Civil Society Index data-setW. Doerner 11 University of Siena, ItalyMost of the models of state-civil society relaonships see the specific situaons determined mainly by thestate’s acvies, structures and actors1. This paper builds on the assumpon that different understanding ofthe role of civil society organizaons at least co-determined, if not precede the formaon of the establishedand formalized relaons. The analysis thus shis the focus from the ’official’ manifestaons of the relaonshipsto their societal condions.1.27.2 Threshold aggregaon as a new approach to the CIVICUS Civil Society Index (CSI) composionO. Kononykhina 1 , T. Anderson 11 CIVICUS:World Alliance for Cizen Parcipaon, South AfricaThe CIVICUS Civil Society Index (CSI) project is one of the most complex of several exisng internaonal researchesmeasuring civil society and the nonprofit sector. However, the aggregang process that creates thefinal empirical illustraon, the CSI Diamond, is based on a linear aggregaon funcon that cannot reflect thediscrepancies in scores in a parcular sub-dimension or dimension. By enabling a high score of one indicatorto compensate for a low score of another (or vice versa), there is a danger of these dimensional and subdimensionalaverages both covering up weaknesses of civil society that the disaggregated data revealed ornegang the strengths. In order to resolve this problem, this paper will test a threshold aggregaon method,an algebraic data analysis technique for construcng indexes, as an alternave for CSI analysis...1.27.3 Confirmatory Factor Analysis of the CSI variable structureL. Fioramon 2 , M. Hoelscher 11 University of Heidelberg, Germany; 2 Centre for Social Investment, University of Heidelberg, GermanyThe Civil Society Index (CSI) has affirmed itself as one of the most comprehensive crossnaonal assessments ofcivil society’s features and components. Yet, limited stascal analysis has been conducted on the mul-levelstructure of the CSI methodology and its internal dimensions. This paper aims to review the consistency ofthe CSI internal structure by tesng it through confirmatory factor analysis of the quantave results of the2008-2010 implementaon phase.1.27.4 Values and acon in the context of civil societyT. Laux 11 Heidelberg University, GermanyThe cross-naonal study of the state of civil society is hindered by the lack of comparable data. Up ll now themost prominent source for cross-naonal comparisons has been the World Value Survey (WVS), which focuseson micro level data to evaluate social and polical parcipaon in different forms. Now the data of the CIVICUSproject offers a more inclusive and differenated perspecve on the state of civil society in different countries.The CIVICUS project avoids a normave bias of civil society, thus there is no a priori selecon of ’good’ and ’bad’civil society (Anheier 2004). This offers the opportunity of using a more value-free or analycal perspecve,whereas most exisng studies are orientated towards a ’western model’ of civil society and exclude otherpossible alternaves (Welzel 1999; Wiarda 2003)...1.27.5 Mapping the value related dimension of civil society: Classical survey approaches in comparisonwith the CIVICUS Civil Society Index.A. Labigne 1

TUESDAY 19 JULY 531 Freie Universität Berlin, GermanyComparave research on Civil Society oen focuses on organizaons as the unit of analysis. Voluntary nonstate,non-market associaons are seen as an indicator for the extensity and intensity of a more or less ”civil”society. However, fundamental quesons about civil sociees demand a complementaon in the relevant unitsof analysis. Therefore the presentaon will not address the comparison of nonprofit organizaon density bycountry, but discuss the following research queson: How can the normave dimension of civil society becompared empirically and what role can the CIVICUS Civil Society Index (CSI) play thereby?1.28 The role and benefit of structured metadata in survey research ITo be held on July 19, 2011 from: 16:00 to 17:30, in room 414.Coordinated by: Joachim Wackerow - GESIS - Leibniz Instute for the Social Sciences, Germany1.28.1 Overview of the Data Documentaon Iniave (DDI) Metadata Standard: Goals and BenefitsM. Vardigan 11 ICPSR, United StatesMetadata are costly to produce but essenal to effecve data analysis. Thus, it makes good sense to createmetadata that are (1) machine-aconable with the capability to drive data collecon processes, and (2)reusable across the enre data life cycle.1.28.2 Benefits of Structured DDI Metadata across the Data Lifecycle: The STARDAT Project at the GESISData ArchiveM. Linne 1 , E. Brislinger 1 , W. Zenk-Möltgen 11 GESIS - Leibniz Instute for the Social Sciences, GermanyHigh quality data and documentaon of survey datasets is needed by social sciensts during the research process.The Data Documentaon Iniave (DDI) uses the idea of the data life cycle across the stages of discoveryand planning, inial data collecon, data preparaon and analysis, publicaon and sharing, and long-termmanagement. The GESIS Data Archive already uses several tools to create standardized documentaon onstudy and on dataset level to fulfil the needs of the scienfic community.Current examples for the benefits of structured metadata are the quesonnaire development of the 4th waveof the European Values Study, which has been supported by the documentaon of the 3rd wave, and thesupport for enhanced publicaons, which enables researchers to indicate the datasets they have used in theirpublicaons.1.28.3 The Use of Structured Survey Instrument Metadata throughout the Data LifecycleS. Hansen 11 University of Michigan, United StatesComputer-assisted survey soware packages such as Blaise store queson-level metadata that are associatedwith the survey data collected. The Michigan Quesonnaire Documentaon System (MQDS) extracts thosemetadata from a Blaise metadata file and outputs them to a DDI3 compliant XML file. Once those metadataare available, they may be used and reused throughout the data lifecycle to document and compare versionsof the survey instrument, document preliminary and cleaned datasets, produce codebooks, and to harmonizeand disseminate data.This presentaon will follow the flow of a study’s queson-level metadata from a Blaise survey instrumentvia MQDS, and follow their and reuse throughout the data lifecycle phases of instrument design, tesng, datacollecon, data processing, and data disseminaon.

54 TUESDAY 19 JULY1.28.4 DDI and the Lifecycle of Longitudinal SurveysL. Hoyle 1 , J. Wackerow 21 Instute for Policy& Social Research, Univ. of Kansas, United States; 2 GESIS - Leibniz Instute for the Social Sciences, GermanyFrom the very beginning of a longitudinal survey, management of metadata is crucial. Structuring metadatain a standardized form greatly increases its ulity, making metadata driven processes possible. Soware usingmetadata may generate survey instruments and samples, administer an online survey, and produce tablesand graphs of summary stascs. Good metadata in structured form makes higher quality documentaonmore accessible and can aid in retrieval of selected data. Quality metadata in structured form should facilitatereplicaon of data collecon and analyses across waves of a longitudinal survey. Replicaon of all or part of asurvey in other places or mes will depend on accessible metadata.1.29 Mixed-Mode Surveys: A Total Survey Error Perspecve IIITo be held on July 19, 2011 from: 16:00 to 17:30, in room 315.Coordinated by:• Don A. Dillman - Washington State University, United States• Edith De Leeuw - University of Utrecht, Netherlands1.29.1 Flexibility of Web Surveys: Probing ’do-not-know’ over the Phone and on the WebJ. Hox 1 , E. De Leeuw 11 University of Utrecht, NetherlandsIn interview surveys, usually a ’do-not-know’-opon is not explicitly offered to a respondent, but interviewerscan accept it. It is good general pracce to train interviewers in using a probe aer an inial ’do-not-know’ toreduce item-nonresponse.In web surveys designers are hesitant to offer an explicit do-not-know opon for fear of encouraging respondentsto choose this opon as a quick answer. One the other hand, not accepng do-not-know, and issuing anerror message insisng on an answer may lead to either irritaon and more break-offs or to guessing and lessvalid answers.1.29.2 A Comparison of CAPI and PAPI through a Randomized Field ExperimentJ. De Weerdt 11 Economic Development Iniaves (EDI), TanzaniaThis paper reports on a randomized survey experiment among 1840 households, designed to compare penand-paperinterviewing (PAPI) to computer-assisted personal interviewing (CAPI). We find that PAPI data containa large number of errors, which can be avoided in CAPI. We show that error counts are not randomlydistributed across the sample, but are correlated with household characteriscs, potenally introducing samplebias in analysis if dubious observaons need to be dropped. We demonstrate a tendency for the mean andspread of total measured consumpon to be higher on paper compared to CAPI, translang into significantlylower measured poverty, higher measured inequality and higher income elascity esmates...1.29.3 Deployment of a Mixed-mode Data Collecon Strategy Does Not Reduce Nonresponse Bias in aGeneral Populaon Health SurveyT. Beebe 2 , J. Ziegenfuss 2 , D. McAlpine 1 , S. Jenkins 2 , L. Haas 2 , M. Davern 31 University of Minnesota School of Public Health, United States; 2 Mayo Clinic - Dept. of Health Sciences Research,United States; 3 NORC - University of Chicago, United States

TUESDAY 19 JULY 55Objecve: To assess nonresponse bias in a mixed-mode (mail and telephone), general populaon health survey.Data sources: Secondary analysis of linked survey sample frame and administrave health data thatincluded survey response disposions and informaon about demographics, medical and surgical diagnoses,comorbidies, and health care ulizaon for the full sample eligible to complete a general populaon healthsurvey. Study Design: The survey was administered by mail, with a telephone follow-up for nonresponders.Characteriscs of respondents (N=3220) were compared to the total eligible sample (N = 6716). We also comparedrespondents who completed the survey aer the 1st and 2nd mail contacts and those who completedthe interview by telephone to the total eligible sample in bivariate and mulvariate models...1.29.4 Assessing the Quality of Data by Mode in a Mixed Mode SurveyD. Griffin 11 U.S. Census Bureau, United StatesAs mixed mode surveys become a more common approach to survey-taking, quality assessments must expandto look at the unique error properes of each data collecon mode. The choice to implement a mixed modesurvey must acknowledge that measures of overall survey quality can mask potenal issues at the mode level.1.30 Combining and Enhancing data IIITo be held on July 19, 2011 from: 16:00 to 17:30, in room 315.1.Coordinated by: Mark Elliot - University of Manchester, United Kingdom1.30.1 Combining different types of data in educaonal research. Data organizaon issuesT. Khavenson 1 , Y. Tyumeneva 21 State University - Higher School of Economics, Russia; 2 State University Higher School of Economics, RussiaThe paper describes the part of the research project that studies a dynamics of students’ academic achievementswith the relaon to their country’s social context. The data from 7 waves of educaonal surveys TIMSS(IEA) and PISA (OECD) has been used in the research. Several post-soviet union and East-European countrieswere taken as these are the countries of our area of interest.In our report we concentrate on the methodological issues about database organizaon rather than on theresults’ descripon. Our data is divided into 3 levels. By moving from the type 1 data to the type 3 comparabilityof each characterisc is decreasing but at the same me the explaining power is increasing which happenedto be a very important thing for drawing and interpreng the findings from research. We can consider datatype 1 like quantave one, and the 2nd and 3d types as the qualitave ones.1.30.2 Tracking decisions and informaonal strategies of university students: A combinaon of data fromvarious sourcesV. Lang 2 , M. Han 1 , S. Hillmert 11 Universität Tübingen, Germany; 2 University of Tübingen, GermanyThe combinaon of data from different modes of collecon – online surveys, expert interviews, administra-ve sources – has specific potenals for assessing and enhancing data quality. We explore possibilies andeffects of such combinaons in the context of ongoing research on biographical decisions and informaonalstrategies of university students (cf. ”ScienceCampus Tübingen”; www.wissenschascampus-tuebingen.de).A crical example is the assessment of selecon processes in various fields of study. Administrave data givesa quantave overview of such selecon by providing standardized indicators. Structured interviews withcourse advisors inform us about the criteria of selecon and their praccal implementaon. Data from (online)surveys delivers structured informaon on the level of the research units, in our example students withcharacteriscs relevant for (self-)selecon...

56 TUESDAY 19 JULY1.30.3 Creang arficial student cohorts by stascal matching of disnct educaonal data sourcesD. Becker 11 Cologne Graduate School in Management, Economics and Social Sciences, GermanyMost studies in educaonal research are in the fortunate posion to be able either to rely on pure individualleveldata that originate from the same source, or to merge an exisng individual-level file with informaonon the level of the schools, school districts, etc. However, only lile is known how to proceed if a parcularresearch queson cannot be answered based on one single person-level data file.On the other hand, stemming from the general idea of missing data handling (Rubin 1976), stascians havedeveloped methods to combine informaon from two disnct individual-level data files via the approach of’stascal matching’ (Rubin 1986); but as regards educaonal research, this technique has been praccallydisregarded (a notable excepon is Schubert/Becker 2010).1.31 What Do Survey Researchers Want from Data Archives these Days? IIITo be held on July 19, 2011 from: 16:00 to 17:30, in room 319.Coordinated by: Peter Granda - University of Michigan – Inter-university Consorum for Polical and SocialResearch (ICPSR), United States1.31.1 Data Archives and the Data Lifecycle: The case of organizaonal dataM. Diewald 2 , S. Liebig 11 Department of Sociology, Bielefeld University, Germany; 2 Department of Sociology, Bielefeld University ,GermanyData on organizaons and their members are geng more and more important in the social sciences as theyhelp to understand a wide variety of social phenomena. The rising demand cannot sasfied by usually ”thin”,process-generated data from official stascs alone, but should profit from survey-based data and case studiesconducted by researchers as well, who usually seek to get denser informaon on organizaons and/or theirmembers. Such studies are more theory-driven and guided by substanal research quesons. However, mostof these studies are cross-seconal and are limited to single branches or even single organizaons. Moreover,there is a lack of standardizaon of measuring theorecal constructs. Contrary to individual or householdrelated empirical research, there is no tradion of using standard instruments for informaon retrieval in organizaonresearch...1.31.2 Interacve Exploraon Tools for Geospaal Data on the WebA. Cornilleau 1 , A. Schaal 11 Sciences Po, FranceWith archives of geospaal data becoming larger and more varied, data exploraon and analysis is oen agreater challenge than data access. Unfortunately, geospaal analysis tools are generally expensive, and henceare frequently not at a user’s disposal. To address the problem of accessibility, web based visualizaon toolshave become more common. However, just as interacve online technologies progress, so do the tools thatrely on these technologies. As such, most geospaal analysis tools on the web are lacking certain desiredfeatures that might be found in their more advanced desktop counterparts.1.31.3 Flexible services and intensive care. Challenges for the future of data archiving in GermanyO. Waeler 11 GESIS - Leibniz Instute for the Social Sciences, GermanyOver the past decades the number of empirical studies in the Social Sciences has resulted in, what has some-mes been called, a ”data deluge”: more specialized research in various fields, more large-scale projects,

TUESDAY 19 JULY 57release of formerly disclosed stascal data, expansion of cross-cultural research and so forth. Thus, the characterof the data archive is changing, too, and the surge demands new ways both technical and organizaonalto cope with these challenges.From the view of the Data Archive for Social Sciences (GESIS) the presentaon focuses on four topics:1.31.4 Development of a Social Research Infrastructure in Central Europe: The Case of the Czech RepublicY. Leonyeva 1 , T. Cizek 1 , M. Vavra 21 Instute of Sociology, Academy of Sciences of the Czech Republic, Czech Republic; 2 Instute of Sociology,Czech Academy of Sciences, Czech RepublicThis presentaon will focus on social data services and data archiving in the Czech Republic using the experienceof Czech Social Science Data Archive (CSDA). The authors will provide an overview of the development ofCSDA from its beginnings as a small department within a naonal sociological research instuon to its currentposion where it faces important challenges in transforming itself into an effecve nodewithin the Council of European Social Science Data Archives (CESSDA). The main aim of this presentaon isto share CSDA’s experiences and to discuss the challenges this young CESSDA member faces in (a) promongsecondary data analysis and data quality research, and (b) how best to serve its users within the social sciencesin the Czech Republic.1.32 Human Values IITo be held on July 19, 2011 from: 16:00 to 17:30, in room 415.Coordinated by:• Constanze Beierlein - GESIS - Leibniz Instute for the Social Sciences, Germany• Eldad Davidov - University of Zürich, Switzerland• Peter Schmidt - University of Giessen, Germany1.32.1 Naonal Instuonal Structures and Self-Enhancement Values: A Mullevel Approach Using theEuropean Social SurveyD. Köthemann 11 Bielefeld Graduate School in History and Sociology, Bielefeld University, GermanyContemporary market-sociees can be characterised by a ”culture of compeon” (Coleman 1987). Compe-on is ”a builder of character, a test of personal worth, and a powerful smulus to individual achievementthat ulmately produces the maximum economic value for society as a whole.” (ibid.) From this point of viewliving in a market-society has a strong impact on individual value priories. But as Messner/ Rosenfeld (1994)point out, the influence of market-mechanism on individuals differs between market-sociees and depend onnaonal instuonal arrangements. Konty (2005) makes the idea more concrete when he agues that the moreinfluence economic social instuons gain the greater the people’s importance of self-enhancement-values.Other social instuons like the family or the polity can counterbalance these effects of societal economicinstuons...1.32.2 Social Capital, Values and Trust: A Conceptual and Analycal ClarificaonU. Häfliger 11 University of Zürich, SwitzerlandTrust does not feature in the values or movaon literature as a determinant of social capital, except in somecases as a personal trait (John and Srivastava 1999). In the social capital literature, however, shared norms arediscussed as a key ingredient in building trust but no link is made to the values literature (e.g., Uslaner 2002).

58 TUESDAY 19 JULYAt the same me, trust is oen considered to constute social capital as well as influencing the formaonof social es thus leading to conceptual obscurity (Tarrow 1996, Portes 1998). The link between trust andstructural social capital (social relaons as memberships in organizaons or networks, volunteering, informalrelaons), however, has been shown to be rather tenuous (Newton 2001; Uslaner 2000; Glaeser et al. 2002;Durlauf 2002, Schwartz 2007)...1.32.3 Shalom Schwartz’ personal basic human values circular model from top-down and down-top perspecvesI. Tart 11 Tallinn University, EstoniaSchwartz Value Survey (SVS) and its counterpart Portrait Value Quesonnaire (PVQ) had served well to researchhuman values landscape and its correlates into broad social and polical acvies fields in different culturalenvironments. Now we have opportunity to test the model on European Social Survey (ESS) four rounds (2002-2008) 21-item PVQ data (190753 respondents reduced to 176309 aer adjusng to criteria of not more than5 missing& 16 repeated values per respondent). Response style for each respondent is adjusted by using centred personalvalues which can be effecvely calculated at any aggregate level. Using the Proxscal algorithm of muldimensionalscaling (MDS) the top-down model of the whole massive has done in two- and three-dimensionalversions. It proves that all ten types of Schwartzean personal values are disnct and effecve...1.32.4 Value similarity among grandparents, parents, and adolescents: the stereotype effectD. Barni 1 , S. Ranieri 11 Athenaeum Center for Family Studies and Research, Catholic University of Milan, ItalyThe connuity of values between generaons is an important goal of value transmission and it is crucial forthe funconing of society (Vedder et al., 2009). Value similarity can originate from parental influence on theirchildren’s priories as well as from the common socio-cultural environment in which parents and children live.There are several studies concerning the parental influence on children’s values (see, Knafo& Schwartz, 2009), whereas lile is known about the impact of the common context on value similarity. Accordingto Kenny and Acitelli (1994), the typical cultural response, called ”stereotype effect”, must necessarilybe considered to establish whether and to what extent the similarity is the result of the (unique and specific)relaon between one parent and his/her offspring rather than the result of shared cultural value climate...1.32.5 Tesng a New Approach for Operaonalizaon of the Basic Value ModelL. Lilleoja 11 Tallinn University, EstoniaThe Basic Value theory of Shalom Schwartz has a central place in the contemporary value studies. Despite ofthe extensive use of his model, there is an on-going discussion about number of its discrete factors. MDSbasedanalyses have mainly suggested model with 10 value types, but studies using CFA have mostly referredto 7-type soluons (with some overlapping factors).Knoppen and Saris have described an alternave CFA approach, which allows a more precise differenaon ofvalue types and solves a high correlaon problem between some of them (Knoppen& Saris 2010). In their upcoming paper (in cooperaon with Schwartz), this method is tested with 40-item PVQon two samples of German students (N₁=395; N₂=321), which leads to a new, more detailed, 19-value typemodel.1.33 Public trust and instuonal legimacy: Cross-naonal analyses IITo be held on July 19, 2011 from: 16:00 to 17:30, in room 303.

TUESDAY 19 JULY 59Coordinated by: Jonathan P. Jackson - London School of Economics, United Kingdom1.33.1 Making Comparisons Meaningful: Conceptual and Measurement Issues of Polical Trust AtudeD. Poznyak 11 Katholieke Universiteit Leuven, BelgiumStudies in public opinion oen examine the variaon in polical and social atudes over me using longitudinalor me-series designs. Most researchers however do not test whether the latent constructs can in factbe meaningfully compared – that is whether their meaning and measurement remains the same across theenre range of available data.1.33.2 Public Opinion Concerns on Data Privacy in Eurobarometer Surveys.F. Odella 11 Università degli Studi di Trento, ItalyIn 1996 the European Union produced a comprehensive directory for establishing privacy rights and independentnaonal agencies to control for the implementaon of the new requirements by public and private actors;this came aer a long debate at the European commission and aer several country members had launchedtheir own legislave framework. EU instuons paid much aenon to the establishment of detailed andcomplex procedural requirements for protecon and ensuring cizens’ privacy but the situaon tends to bevery different inside country members. To describe how public opinion changed aer the implementaonof the 1996 direcve this paper uses Eurobarometer data concerning invesgaons on privacy opinions andstascal data...1.33.3 Determinants of Values and Atudes toward Society and Civic Systems in Lan-American YoungPeople: An Internaonal Comparave AnalysisC. LOPERA 1 , A. Uzaheta 21 Colombian Instute for Educaonal Evaluaon (ICFES ), Colombia; 2 Colombian Instute for Educaonal Evaluaon(ICFES), ColombiaThroughout recent decades, the growing impact of globalizaon and modernizaon of economic systems, theexternal threats to civil society and its liberes, the increasing importance of non-governmental organizaons,as well as the younger generaon´s limited interest and engagement in public life and polics, are all factorsthat have contributed to an innovave reflecon on the meaning of cizenship and especially about the waysin which young people are prepared and, consequently, capable of assuming their roles as cizens (Schulz, W.,Fraillon, J., Ainley, J., Losito, B.,& Kerr, D., 2008)...1.34 Sensive Quesons and Social Desirability Bias: Theorecal Perspecves andData Collecon Strategies IIITo be held on July 19, 2011 from: 16:00 to 17:30, in room 412.Coordinated by: Daniel Oberski - Universitat Pompeu Fabra, Spain1.34.1 The Use of Split Sample Technique to Dissociate Atudes towards White Collar & Blue Collar MigrantWorkersA. Diop 2 , D. Al-Emadi 2 , D. Howell 11 University of Michigan, United States; 2 Social and Economic Survey Research Instute (SESRI), Qatar University,Qatar

60 TUESDAY 19 JULYAccording to the labor market compeon theory, low-skilled naonals in developed countries are more likelyto have an-immigrant atudes because of fears about labor market compeon, especially in mes of economicdownturns. Other studies also point to educaon as the main driving force behind naonals’ posiveatudes towards migrant workers. Most of the studies trying to explain naonals’ atudes towards migrantworkers are based on surveys which ask quesons about immigrants regardless of their skills levels. Under thisformat, we do not know which of the two groups of migrant workers (high-skilled or low-skilled) respondentsrefer to when answering the survey quesons. Consequently, it is difficult to dissociate naonals’ atudestowards the various sub-groups of migrant workers...1.34.2 Does survey mode affect propensity to report perceived racial and ethnic discriminaon in healthcare? Findings from a culturally diverse sampleJ. Kemmick Pintor 1 , D. McAlpine 1 , T. Beebe 21 University of Minnesota School of Public Health, United States; 2 Mayo Clinic - Dept. of Health Sciences Research,United StatesObjecve: To examine the effect of survey mode (mail vs. telephone) on the likelihood of reporng healthcare-related discriminaon based on race or skin color in a culturally diverse sample.Methods: Data come from a mixed-mode, mail and telephone survey of adult enrollees in public health careprograms (N=2194) in a Midwest U.S. state. The survey oversampled diverse cultural groups; the final samplerepresented the American Indian, Somali, Hmong, African American, Lano, and European-American popula-ons. Perceived discriminaon was measured as experience of discriminaon due to race or skin color duringhealth care visits in the past year. Significant mode effects were tested controlling for differences betweenrespondents who completed the survey by phone or mail (age, gender, educaon, primary language, and employment,marital and health status).1.34.3 Measures for Desirability Beliefs about Atudes toward Foreigners and Their Predicve Validityfor Social Desirability BiasV. Stocké 11 University of Bamberg, GermanyHow respondents perceive answers to be evaluated must be regarded as an important determinant for thestrength and direcon of social desirability bias. Despite of the importance of these desirability beliefs thequeson how to measure them has received lile aenon. In the case of connuous traits, as for example theatude toward foreigners, different measures for these beliefs are possible. These are on the one hand ’onepointmeasures’, capturing either the ancipated evaluaon of an extremely posive or an extremely negaveatude answer. The validity of both measures presupposes the neutrality assumpon to be empirically given.Accordingly, posive or negave evaluaons of the respecve trait to represent a valid measure for the strengthand direcon of incenves for socially desirable response behavior, assumes the complementary trait to beneutrally evaluated...1.35 Surveying children and young people IITo be held on July 19, 2011 from: 16:00 to 17:30, in room 321.Coordinated by:• Lisa Calderwood - Centre for Longitudinal Studies, Instute of Educaon, University of London, UnitedKingdom• Kate Smith - Centre for Longitudinal Studies, Instute of Educaon, University of London, United Kingdom

TUESDAY 19 JULY 611.35.1 How Surveying Teenagers About their Licit and Illicit Drug Use?S. Legleye 3 , S. Spilka 1 , O. Le Nézet 1 , G. Vivier 4 , E. Markou 4 , F. Beck 21 OFDT, France; 2 INPES, France; 3 INED, France; 4 Instut Naonal d’Etudes Démographiques, FranceSince 2004, different surveys on drug use have been conducted among French adolescents aged 17 living inParis, thanks to a partnership between the City of Paris and the French monitoring centre for drug and drugaddicon.In 2004, the ESCAPAD survey interviewed a large representave sample teenagers living in Paris (n=1,552,age=17) with self-administered quesonnaires (anonymity and confidenality being guaranteed). It showedthat adolescents from the working-class districts reported less licit (tobacco and alcohol) and illicit drug usethan those from the wealthier areas. Different hypotheses were formulated to interpret these findings, includingsocial support, material resources, school engagement, percepon of dangerousness and addicve powerof drugs, culture and religion, lack of trust in the survey, lack of self-confidence.1.35.2 Combining mul-informant, quantave and qualitave survey data to esmate the prevalence ofmental disorders among childrenH. Meltzer 11 University of Leicester, United KingdomThe first naonal survey on the mental health of children and young people aged 5-15 in Great Britain wascarried out in 1999. This was repeated in 2004 with a longitudinal component in 2007. Overall, 24,000 assessmentswere made. There were many methodological challenges in conducng these surveys. The first mostsalient issue was who should be interviewed: the children themselves, a parent and a teacher. We had toconsider at what age children could be interviewed, which parent to interview and the most difficult decisionwhich teacher to contact. Apart from a maximum of three sources of quantave data from the structuredquesonnaires, qualitave data were also collected – respondents were prompted to describe, enlarge on orembellish their responses to pre-coded quesons. As expected we got extensive conflicng and contradictoryinformaon from each informant...1.35.3 Are young children highly movated respondents? The willingness of six-year-olds to cooperate indevelopmental testsB. Becker 11 GESIS – Leibniz Instute for the Social Sciences , GermanyMany large-scale surveys that focus on children and youths conduct standardized psychological tests of variouskinds (e.g., assessments of children’s intelligence, academic achievement, development, etc.). The quality ofthis kind of data probably depends on how movated the children are to solve the test tasks.This paper analyses six-year-old children’s willingness to cooperate in developmental tests that are part of asurvey conducted at the respondents’ homes. Child characteriscs, family/parent characteriscs and interviewercharacteriscs are considered as determinants of children’s movaon during the tests.1.35.4 Non-Response and Arion in the PSID Child Development Supplement and their Relaonship withChildren’s Test ScoresN. Sastry 11 University of Michigan, United StatesThe PSID Child Development Supplement (CDS) was begun in 1997 among a sample of 0–12 year old childrenin sample households, with up to two children selected in each family. Two follow-up waves were fielded in2002–03 and 2007–08. The CDS data support a wide range of analyses; a large fracon of these analyses focuson children’s test scores in reading and mathemacs which represent gold-standard measures of academicachievement (and one of the costliest components of the study). Informaon from auxiliary respondents—the

62 TUESDAY 19 JULYchild’s primary caregiver (PCG) and school teacher—are also relevant for examining children’s achievement aswell as other key outcomes. In this paper, we look at the relaonship between test scores for both childrenand PCGs, on one hand, and sample arion and non-response on the other hand...1.36 The development and validaon of psychological short scales and their benefitsfor survey research IITo be held on July 19, 2011 from: 16:00 to 17:30, in room 413.Coordinated by:• Beatrice Rammstedt - GESIS - Leibniz Instute for the Social Sciences, Germany• Jürgen Schupp - SOEP - German Instute for Economic Research (DIW Berlin), Germany1.36.1 Response Space as the New Froner: How Number of Items, Response Categories, and Sample SizeAffect Aggregate EsmatesR. Thomas 1 , J. Bremer 21 ICF Internaonal, United States; 2 Compete Internaonal, United StatesIn psychometrics, increasing the number of items to measure a specific concept has been shown to improvethe stability of measurement according to the Spearman-Brown formula. Thomas (1999) showed that thenumber of items and response categories can be jointly described as a response space and that increasingthe gradaons within the response space improves response stability, reaching an asymptote of reliability.However, many survey projects are not interested in individual classificaon per se but in populaon esmates(e.g. are males less depressed than females?). Since most studies draw samples of the populaons of interest, amajor concern has been how increasing sample size improves precision of esmates of populaon parameters,following the Law of Large Numbers (Bernoulli, 1713). Lile prior research has be done on the intersecon ofindividual measurement stability and sample measurement stability...1.36.2 A Mul-Method Approach to ForgivenessS. Fücker 2 , C. von Scheve 2 , A. Merkl 1 , M. Park 21 Charite Berlin, Germany; 2 Freie Universität Berlin, GermanyThis contribuon gives insights into ongoing research invesgang the nature, condions, and consequences offorgiveness using a mul-method approach and close cooperaon between the social and behavioral sciences.Interpersonal transgression and wrongdoing are ubiquitous incidents in human social interacon. One of themost prevalent responses to transgression is retaliaon. Though effecve in deterring future harm, retaliaonis a prime precursor of prolonged conflict. In contrast, forgiveness is a response to transgression aiming atlong-term conflict resoluon and the re-establishment of social relaonships.1.36.3 The Parent-reported Pediatric Perceived Cognive Funcon Item Bank (pedsPCF) and Its Applicaon– Computerized Adapve Tesng (CAT)J. Lai 1 , D. Cella 1 , Z. Bu 1 , F. Zelko 2 , S. Goldman 21 Northwestern University, United States; 2 Children’s Memorial Hospital in Chicago, United StatesOBJECTIVE. This paper reports the development and psychometric properes of a parent-reported pediatricperceived cognive funcon item bank (pedsPCF) and the evaluaon of the comparability between scoresproduced by the full-length pedsPCF and pedsPCF-based CAT tesng.

63Day 2Wednesday 20 July2.1 Falsificaons in Survey Data: Prevenon and DeteconTo be held on July 20, 2011 from: 09:00 to 10:30, in room 315.1.Coordinated by:• Natalja Menold - GESIS - Leibniz Instute for the Social Sciences, Germany• Peter Winker - Justus-Liebig-University Giessen, Center for internaonal Development and EnvironmentalResearch (Z, Germany• Gesine Güllner - University of Bonn, Germany• Nina Storfinger - Justus-Liebig-University Giessen, Germany2.1.1 Applicaon of surveys in calculang chess ELO rangsR. Andrei 11 Naonal Instute of Stascs Romania, RomaniaIn chess there is a rang(ELO) indicang the players’ strength.The Internaonal Chess Federaon (FIDE) uses analgorythm to calculate the ELO rang. Some players or some observers of the chess phenomenon aren’t happywith the way it is calculated now and have launched a campaign to change the method of determining ELOrang. To achieve this goal, they need a survey to demonstrate the approval of the chess community regardingtheir proposal. My opinion is there is some need of changing the way ELO rang is calculated, as there aresome cases in which it doesn’t actually reflect the performances of a player. The stascal series demonstratethat in the last years, there is a trend of concentraon of the rang around certain values(for example 2400 or2700)...2.1.2 The Failure of the Half-Open Interval Missed Housing Unit ProcedureS. Eckman 2 , C. O’Muircheartaigh 11 University of Chicago, United States; 2 Instute for Employment Research, GermanyThe half-open interval procedure seems to offer an inexpensive method of reducing undercoverage in housingunit frames during data collecon: units missing from the frame are linked to covered units and given a chanceof selecon. The procedure is commonly used in U.S. face-to-face surveys. This paper provides details onhow the procedure should work, and then discusses problems both in principle and in pracce. Reporng theresults of an experiment in three U.S. cies, we show that interviewers do not carry out the procedure correctly,leading to uncorrected undercoverage and the introducon of overcoverage. On the basis of this evidence, weconclude that the half-open interval procedure should not be considered a cure for undercoverage and shouldbe used only with great care and training.

64 WEDNESDAY 20 JULY2.1.3 Idenficaon of quesons and quesonnaire characteriscs that favor the detecon of falsificaonsin surveysN. Menold 2 , G. Güllner 11 University of Bonn, Germany; 2 GESIS - Leibniz Instute for the Social Sciences, GermanyIn this paper we present first results of a research project (funded by the DFG), which deals with ex-post detec-on of falsified data in face-to-face surveys. As a first step in the project we conducted two explorave studiesto idenfy, which aributes of quesonnaires would be useful to idenfy interviewers producing falsified data.During this step exisng survey data are compared with falsified data to detect quesonnaire elements, whichappear to be sensible to falsificaon. The falsified data is collected as following: a subsample of an exisngdataset of a survey (German General Social Survey ALLBUS) is drawn; for this subsample, false data are producedby people parcipang in the study, based on several socio demographic characteriscs of persons inthe subsample supplied to the falsifiers. The produced ”falsificaons” are then compared with exisng, realsurvey data...2.1.4 Indicator based idenficaon of falsificaons in survey dataN. Storfinger 1 , P. Winker 21 Justus-Liebig-University Giessen, Germany; 2 Justus-Liebig-University Giessen, Center for internaonal Developmentand Environmental Research (Z, GermanyData quality in face-to-face interviews might be affected by interviewers’ irregular behavior like deviaon fromthe prescribed interviewing procedures. If this is done consciously it might be called cheang or interviewerfalsificaon. Based on the movaon of such behavior we develop a mulvariate stascal method for ex-postidenficaon of falsificaons in survey data. In order to assess each interviewer, we compute some specific ”indicesof cheang”, e.g. the amount of unanswered quesons, and implement a cluster analysis based on theseindices. Using this procedure we try to split the interviewers into two groups, correct and possibly cheangones. The performance of this method is then assessed referring to the fracon of correctly assigned interviewers.One must note that in the examined data the cheang interviewers are known beforehand, so thatwe are able to validate the clustering process immediately...2.2 Mixed-mode surveys: quality, costs, administraon and opmizaonTo be held on July 20, 2011 from: 09:00 to 10:30, in room 315.Coordinated by:• Vasja Vehovar - University of Ljublana, Slovenia• Nejc Berzelak - University of Ljubljana, Slovenia2.2.1 Impact of the mode of data collecon on the quality of survey quesons depending on respondents’characteriscsM. Revilla 11 Universitat Pompeu Fabra, SpainBecause of costs and me reducon, the Internet is more and more used to conduct surveys. However, movingfrom tradional modes of data collecon to the Internet may threaten the comparability of the data (longitudinallyor across groups) if the mode of data collecon has an impact on the way of answering the quesons ofthe respondents. In previous research, Revilla and Saris (2010) find similar average quality for several surveyquesons when asked in a face-to-face interview or when asked in a Web quesonnaire...

WEDNESDAY 20 JULY 652.2.2 A mixed mode pilot on consumer barometerP. Taskinen 1 , M. Simpanen 11 Stascs Finland, FinlandIn Finland, consumer (barometer) survey is widely recognised as an important tool for forecasng consumerbehaviour as well as cyclical movements. Stascs Finland’s interviewers conduct the telephone interviewsmonthly in accordance with the harmonised EU data collecon method. Survey suffers from increasing nonresponserate. Stascs Finland has set up a web survey pilot project concerning the EU Consumer Survey tobe carried out in March 2011. The main task of the project is to find out: - What mixed mode data collec-on requires from praccal arrangements, such as quesonnaire design, - What costs and/or cost savings thecombined collecon brings about,- What effects the combined collecon has on the quality and results of thesurvey.Special aenon has been given to sample design in the preparatory work. The sample size of the pilotproject is 4,000 (which is more than that of the basic survey, 2,200)...2.2.3 Opmizaon of dual frame telephone survey designsA. Slavec 1 , V. Vehovar 21 University of Ljubljana, Slovenia; 2 University of Ljublana, SloveniaDual frames of fixed and mobile phone numbers are an aracve alternave to tradional fixed phone surveyswhich suffer from under-coverage due to the growing share of mobile-only households. However, adding themobile phone frame can bring some operaonal problems and raise costs. To make best advantage of thismixed mode design we try to determine the opmal mixture parameter of the two frames: what share of themobile subsample is sufficient to reduce error without substanally raising costs? First, we divide the targetpopulaon into five strata according to phone use domains determined by possession of availability rates ofboth phone services. Then, we develop an opmizaon procedure that minimizes the product of error andcosts giving four analycal soluons. This is applied to a 2008 Flash Eurobarometer survey where dual framesampling was used for eight countries...2.2.4 Separaon of selecon bias and mode effect in mixed-mode survey – Applicaon to the face-to-faceand web Lyon household travel surveyC. BAYART 2 , P. BONNEL 11 LET, ENTPE, France; 2 Laboratoire de Sciences Actuarielle et Financière, Université Lyon 1, FranceHousehold travel surveys response rates are decreasing. Efforts are made to increase response rate for tradionalsurvey by improving the quesonnaire, reducing respondent burden, increasing reminders… Even ifresults are generally posive, it is in most cases not sufficient. Weighng aims at reducing the impact of nonresponse, but it is always necessary to postulate that people with some socio-demographic characteriscs whodo not respond to a survey have the same behaviour than people with the same socio-demographic characteriscswho respond. But evidence seems to indicate that it is not always the case for travel. To reduce this biasof non-response, we have realised a web survey in parallel of the 2006 household travel survey conducted inface to face in Lyon. The idea was to propose to households who refuse to respond in face to face or was notreachable aer a certain number of aempts to respond by the web...2.3 New perspecves on queson design, translaon and adaptaon. (Formertle: Scripted quesonnaires versus their oral implementaon)To be held on July 20, 2011 from: 09:00 to 10:30, in room 303.Coordinated by:• Janet Harkness - University of Nebraska, Lincoln, United States• Brian Kleiner - FORS - Swiss Foundaon for Research in Social Sciences, Switzerland

66 WEDNESDAY 20 JULY• Beth-Ellen Pennell - University of Michigan, United States• Isabelle Renschler - FORS - Swiss Foundaon for Research in Social Sciences, Switzerland2.3.1 Adaptaon and measurement comparability in mullingual surveysJ. Harkness 11 University of Nebraska, Lincoln, United StatesThis paper sets out to provide a general overview of adaptaon needs and challenges in mullingual surveyresearch.2.3.2 Survey Quesonnaire Adaptaon across ModesS. Hansen 1 , B. Pennell 11 University of Michigan, United StatesIdeally, standardizaon of quesonnaires, meant to reduce measurement error, would involve standardizaonof mode (e.g., telephone or face-to-face and interviewer- or self-administered) and less variaon in other typesof error. The choice of modes becomes more complex in cross-naonal survey research. There may be widevariaon across naons in sample coverage, populaon access, nonresponse trends, literacy levels, and socialdesirability, as well as technical infrastructures, labor availability, and cost structures.2.3.3 Survey translaon in countries with mulple and non-standardised languagesE. Ersanilli 11 University of Oxford, United KingdomThere is a growing body of research on best pracces in survey translaon procedures, such as back-translaon,parallel translaon and TRAPD. Most of the studies on translaon procedures focus on research in Westerncountries. Less is known about translaon pracce and problems in surveys in developing and transion countries.These countries do not only tend to have a higher degree of linguisc diversity (in African almost everycountry has several populaon groups with different mother tongues), but some of the naonal and minoritylanguages do not have a fully standardized wrien form. This poses specific challenges to the translaonprocess.2.4 Quality Assurance in Cross-Naonal SurveysTo be held on July 20, 2011 from: 09:00 to 10:30, in room 319.Coordinated by: Gijs van Houten - Eurofound, Ireland2.4.1 Centralisaon vs. Harmonisaon, Streamlining mulnaonal survey operaons in a single projectinfrastructureA. Illyes 1 , P. Huszk 1 , G. Hideg 11 Gallup, BelgiumThe presentaon describes Gallup’s integrated mul-country survey architecture that centralises project implementaonto the fullest possible extent. This infrastructure, serving the Flash Eurobarometer as well asother presgeous mulnaonal studies, has been created in an XML environment using standard DDI, so thatevery instrument of the survey planning, preparaon, implementaon and assessment runs in a centralisedenvironment, with specific language ’masks’ where necessary. That is, all relevant documents, tools, processesand protocols are created centrally, and are automacally distributed with the appropriate languageor country flag to the naonal partners via web-based applicaons. Under the presented infrastructure, eachmajor data domain of survey implementaon (sampling, quesonnaire, translaon, data capture/scripng,automac eding, weighng, basic aggregaons)...

WEDNESDAY 20 JULY 672.4.2 Data Collecon Quality Assurance in Cross-Naonal Surveys: The Example of the ESSV. Halbherr 1 , A. Koch 1 , A. Scheuer 11 GESIS - Leibniz Instute for the Social Sciences, GermanyThe significance of cross-naonal surveys for the social sciences has increased over the past decades and withit the number of cross-naonal data sets that researchers have access to. Cross-naonal surveys are typicallylarge enterprises that demand dedicated efforts to coordinate the process of data collecon in the parcipangcountries. While cross-naonal surveys have addressed many important methodological problems, such astranslaon and the cultural applicability of concepts, the management of the data collecon process has yethad lile place in cross-naonal survey methodology.2.4.3 Checking and Balancing. Quality Assurance in the 5th European Working Condions SurveyG. van Houten 1 , M. Lyly-Yrjanainen 1 , A. Parent Thirion 2 , G. Vermeylen 21 Eurofound, Ireland; 2 European Foundaon for the Improvement of Living and Working Condions, IrelandThe European Foundaon for the Improvement of Living and Working Condions (Eurofound) carries out threerecurring Europe-wide surveys: the European Working Condions Survey (EWCS), the European Quality of LifeSurvey (EQLS), and the European Company Survey (ECS). Most recently, the 5th wave of the EWCS was implemented.To ensure the producon of high quality data each stage of the survey was carefully planned, closelymonitored and documented, and specific controls were put in place. When designing the 5th wave of the survey,close aenon was paid to user sasfacon with the previous wave and on their future informaon needs.Furthermore, an assessment was made of possible improvements in the way in which the survey addressesthe topics that are central to European policy making...2.4.4 European Values Study – the 4th wave: how to ensure quality in an all European Survey-the perspecveof collector, user and archiveE. Brislinger 2 , R. Luijkx 11 Tilburg University, Netherlands; 2 GESIS - Leibniz Instute for the Social Sciences, GermanyWe will outline the efforts undertaken to maximize the quality of the 4th wave of the European Values Study,a large-scale, cross-naonal, and longitudinal survey research program started in 1981. The 2008 survey wasconducted in 47 countries/regions with a focus on a broad range of values and was highly comparable withthe earlier waves. High standards were used in developing and translang the quesonnaire, in sampling, inmonitoring the data collecon, and in data cleaning and documentaon. To guarantee easy access and properuse, the data are available on-line along with comprehensive documentaon on the study and variable level.2.4.5 Harmonising Standards for Quality Assessment in Cross Naonal SurveysM. Petrakos 1 , T. Ieromnimon 1 , P. Stavropoulos 2 , G. Petrakos 11 Agilis SA, Greece; 2 Agili, GreeceThe presentaon describes the development of harmonised standards for the assessment of Cross Naonalsurveys based on the framework developed for Official Stascs.Cross Naonal surveys are providing comparable informaon for a number of countries and form an importantbasis for scienfic research and policy making. Quality assessment is required at various levels for survey design,assessment of survey processes and data disseminaon. It is important for both data producers and moreimportantly users that the standards used are as much as possible harmonised. This will provide importantbenefits for the development of relevant aspects of survey methods as well as enable users to compare andcombine results and findings.

68 WEDNESDAY 20 JULY2.5 Survey innovaons in data collecon for longitudinal surveys of the generalpopulaonTo be held on July 20, 2011 from: 09:00 to 10:30, in room 340.Coordinated by:• Jonathan Burton - University of Essex, United Kingdom• Emanuela Sala - University of Milano Bicocca, Italy2.5.1 Innovaons in Understanding Society: The UK Household Longitudinal StudyJ. Burton 2 , E. Sala 11 University of Milano Bicocca, Italy; 2 University of Essex, United KingdomLongitudinal surveys across the world are introducing innovaons to improve response rates, increase dataquality or to collect new types of data. Understanding Society: the UK Household Longitudinal Study (UKHLS)is a large new household panel survey for the UK. In this presentaon we give a broad overview of this newsurvey, with an emphasis on innovaons that have been introduced. One major element of the new study isan Innovaon Panel. This is a longitudinal panel which allows the tesng and development of new quesons,and new ways of asking exisng quesons. This is not the sole source of innovave data collecon methods,however. On the main sample we use geographical idenfiers to allow geo-coded linkages to facilitate analysisof local area and neighbourhood effects. Respondents have been asked for their consent to link administraveinformaon to their survey responses...2.5.2 Longitudinal Study of Australian Children (LSAC) - Evoluon of data collecon methodsJ. Corey 11 Australian Bureau of Stascs, AustraliaAs the children in our study have grown and changed, so have the methods we use to collect the data. Wechange our methods of collecon for a number of reasons:• To promote engagement in the study by ourrespondents,• To keep up-to-date with technology2.5.3 Using a Daily Prospecve Life-Chart Method to Examine the Impact of Life Events on the Course ofBipolar DisordersW. van der Vaart 3 , J. van Zaane 2 , S. Draisma 1 , J. Smit 41 Vrije Universiteit Amsterdam, Netherlands; 2 Vrije universiteit, ggzingeest, Netherlands; 3 University for Humaniscs/ GGZ inGeest Instute of Psychiatry and Mental Health, Netherlands; 4 Vrije Universiteit MedicalCenter Amsterdam, NetherlandsLife events have been found to influence the course of mood disorders, like bipolar disorders. However, themechanisms through which they operate are not clear yet. Generally, most studies claim that it is not theimpact of a single life event that triggers bipolar disorders, but assume an addive effect of mulple life events.It’s unknown though, whether this also applies to short-term effects of relavely common life events. Also it’sunclear how the effects of events may change over me; an issue that is complicated by the fact that dates oflife events oen are subject to recall error.2.5.4 Challenges in Data Collecon and Data Protecon: The Naonal Educaonal Panel StudyI. Ristau 1 , S. Meixner 1 , N. Lidzba 1 , A. Müller-Kuller 1 , J. von Maurice 1 , H. Blossfeld 11 Naonal Educaonal Panel Study (NEPS), GermanyIn modern knowledge-based sociees, educaon is not only the key for economic growth and prosperity but itis also decisive for coping with the challenges of a rapidly changing and globalizing world. The German Naonal

WEDNESDAY 20 JULY 69Educaonal Panel Study (NEPS) has been set up to find out more about the acquision of educaon in Germany,to plot the consequences of educaon for individual biographies, and to describe central educaonalprocesses and trajectories across the enre lifespan. Five interlinked theorecal dimensions form the mainstructure of the NEPS: Competence Development, Learning Environments, Social Inequality and EducaonalDecisions, Migraon Background, and Return to Educaon. Methodologically, the NEPS is based on a mulcohortsequence design. Six representave starng cohorts from different age groups - from birth to adulthood- are being recruited between 2009 and 2012...2.6 Indicators of Survey Data QualityTo be held on July 20, 2011 from: 09:00 to 10:30, in room 413.Coordinated by:• Sabrina Zuber - SHARE, MEA, University of Mannheim, Germany• Stephanie Stuck - SHARE, MEA, University of Mannheim, Germany2.6.1 Use of the R-indicator for analysing representavity of surveysJ. Bethlehem 1 , B. Schouten 11 Stascs Netherlands (CBS), NetherlandsThe response rate is oen used as a simple indicator for the quality of survey response. However, high responserates do not guarantee reliable esmates and low response rates do not necessarily lead to unreliableesmates. Recently, a new indicator was developed: the R-indicator. This indicator focuses on measuring thecomposion of the survey response, i.e. the representavity of survey response. The more the composionof the response deviates from a fully representave response, the lower the value of the indicator will be.2.6.2 Research based on Sasficing Theory: a systemac review of methods and resultsC. Roberts 2 , E. Gilbert 1 , N. Allum 11 University of Essex, United Kingdom; 2 University of Lausanne, SwitzerlandIn 1987, Krosnick and Alwin published an arcle about response order effects in survey measurement in whichthey presented a cognive theory for why some respondents might exhibit such effects based on Herbert Simon’s(1957) concept of ’sasficing’. The approach was later elaborated in an arcle by Krosnick (1991) to accountfor a range of other response effects oen observed in atudinal data, aribung them to respondentsshortcung cognive processes necessary for reporng answers accurately (Tourangeau, 1984). In the twodecades since, Krosnick’s arcle has become one of the most frequently cited in the field of survey methodology,and sasficing theory has become a popular framework for invesgang the occurrence of measurementerrors associated with the response process...2.6.3 Underreporng in Interleafed Quesonnaires: Evidence from Two Web SurveysR. Medway 3 , L. Viera Jr. 1 , S. Turner 2 , S. Marsh 11 Fors Marsh Group, United States; 2 Fors Marsh Group LLC, United States; 3 Joint Program in Survey Methodology,University of Maryland, United StatesWhen survey respondents are faced with long, cognively demanding interviews, they may take measuresto reduce their cognive burden. In parcular, when quesons are presented in a predictable paern, it isrelavely easy for respondents to learn that certain behaviors will help them get to the end of the interviewmore quickly. One frequently-used paern is to ask quesons in an interleafed format, in which each quesonin a series of filter items is succeeded by follow-up quesons only if the individual responds affirmavely tothe filter. In such cases, respondents may learn that negave responses to filter quesons help them end theinterview more quickly. Such underreporng can lead to biased survey esmates...

70 WEDNESDAY 20 JULY2.6.4 Improving the quality of cross-naonal surveys: towards a broader understanding of equivalenceW. Aschauer 2 , M. Weichbold 1 , R. Bachleitner 21 Salzburg University, Austria; 2 University of Salzburg, AustriaAlso in the coming years, survey research will be essenally characterized by a growing importance of interculturaland internaonal comparave research. Besides other well known methodological problems, equivalenceas a condion for the comparability of individual countries or cultures is essenal. Accordingly, the conceptof funconal equivalence discusses the suitability of different survey modes and different sampling strategies,the operaonalizaon of theorecal approaches in different country-specific contexts, various translaonproblems as well as configural, metric and scale equivalence.2.7 Human Values IIITo be held on July 20, 2011 from: 09:00 to 10:30, in room 415.Coordinated by:• Constanze Beierlein - GESIS - Leibniz Instute for the Social Sciences, Germany• Eldad Davidov - University of Zürich, Switzerland• Peter Schmidt - University of Giessen, Germany2.7.1 Children’s values in cross-cultural perspecveA. Döring 2 , J. Cieciuch 1 , J. Harasimczuk 3 , M. Janik 21 University of Finance and Management in Warsaw, Poland; 2 University of Münster, Germany; 3 Cardinal StefanWyszynski University Warsaw, PolandSchwartz’s (1992) theory of universal human values proved to be a highly fruiul framework for conceptualizingand researching values – to date mostly adolescents’ and adults’ values. Our findings go beyond thisusual scope and demonstrate the validity of Schwartz’s theory for values in childhood. Döring, Blauensteiner,Aryus, Drögekamp, and Bilsky (2010) developed a self-report instrument which is suited to children’s cognivedevelopmentalbackground: the Picture-Based Value Survey for Children (PBVS-C). In our study, 1,178 Germanchildren who were between 6 to 12 years old completed the German version of the PBVS-C (Döring et al.,2010), and 1,557 Polish children who were between 6 to 13 years old completed the Polish version of thePBVS-C (Cieciuch, Harasimczuk,& Döring, 2010). Furthermore, the oldest among these children (i.e.,older than 10 years) also completed thePortrait Values Quesonnaire (PVQ; Schwartz et al., 2001)...2.7.2 Meanings of Schwartz Human Values: associaons with other orientaon modelsA. Ramos 1 , J. Vala 11 University of Lisboa - ICS, PortugalThis paper aims to contribute to the understanding of the meanings of the values that are comprised in theSchwartz’s model (21 items version). This will be done through the comparison between the Human ValuesModel and four other movaonal orientaons models: Terminal Values (Rokeach 1973), Materialism/Post-Materialism (Inglehart 1977, 1990, 1997), Social Dominance Orientaon (Sidanius and Prao 1999; Sidaniuset al. 2001) and Independent and Interdependent Self-Construals (Markus and Kitayama 1991; Singelis 1994).These models represent movaonal dimensions that are broadly used in the social sciences to explain atudesand behaviours. With this exercise we do not aim to proceed to any kind of validaon of the HumanValues Model, but to idenfy what people associate to the values that integrate the model. A sample of 638Portuguese university students of different areas is used...

WEDNESDAY 20 JULY 712.7.3 The Constraints and Opportunies of 10 Schwartz Value Items in World Values SurveyM. Rudnev 11 State University - Higher School of Economics, RussiaSchwartz has developed a theory enabling researchers to measure human values, and derived 10 value domainswhich have a universal structure of the value content, are unbiased by method, have strong correlaonswith human atudes and behaviors (Schwartz, 2007). Inial Schwartz value instrument included 56/57 items,then 40-item Portrait Value Quesonnaire replaced it, in European Social Survey it was shortened to 21 itemsand in World Values Survey only 10 items have le. The reliability of 10 items instrument is rather quesonable,since even 21-item PVQ caused a large crique of its ability to measure all 10 value domains, as well asits cross-country comparability (Davidov, 2009)...2.7.4 The Structure of Implicit Values: Applying the Implicit Associaon Test to Schwartz’s Model of BasicValuesM. Vecchione 1 , F. Dentale 1 , C. Barbaranelli 11 Sapienza Università di Roma, ItalyDrawing on Schwartz’s (1992) theory of human values, the Implicit Associaon Test (IAT) was applied to measureimplicit basic values. Four IAT’s were conducted (N = 113), each measuring the relave importance attributedto two opposite values expressing conflicng movaonal domains: Achievement vs. Benevolence,Power vs. Universalism, Security vs. Self-direcon, and Tradion vs. Smulaon. The study was aimed at exploringthe reliability of value-IATs, the extent to which they differ from explicit measures of values, and theirconformity to Schwartz’s circumplex model. Internal consistencies of the IAT’s proved sasfactory for all valuetypes...2.8 Assessing the Quality of Survey Data IITo be held on July 20, 2011 from: 09:00 to 10:30, in room 410.Coordinated by: Jörg Blasius - University of Bonn, Germany2.8.1 Problems with Instuonal Control in the World Value SurveysJ. Blasius 2 , R. Dorau 2 , V. Thiessen 11 Dalhousie University, Canada; 2 University of Bonn, GermanyIn a set of papers, Blasius and Thiessen described a method which they call screening technique. The underlyinglogic of this approach consists of revealing unusual paerns in the data. Our preferred screening techniquesconsist of mulple correspondence analysis, categorical principal component analysis and in a few cases principalcomponent analysis. Oner advantage of these methods is that the numeric soluons are twinned withvisual displays of the soluons; graphics facilitate the detecon of dirty data. When an unexpected paernappears, the first line of aack is to pursue whether this is the result of a methodological artefact poinng toa data quality problem.2.8.2 Response Paerns and Scale Usage Differences Across Ethnicies and Countries: Myth or Reality?R. Thomas 11 ICF Internaonal, United StatesSome researchers have indicated that there may be differences in scale usage as a result of racial or ethnic backgroundor country of residence. In the U.S., we oen encounter anecdotes from other researchers who believethat Blacks and Hispanics are more likely to use more posive response categories though there has been onlymixed research supporng this asseron. Internaonally, some researchers have indicated that Germans areless likely to use extreme scale values (i.e. be more likely to use a middling response style) while others have

72 WEDNESDAY 20 JULYreported that the Spanish are more likely to use higher scale values or employ an extreme response style. Ina series of 4 large studies using online surveys we examined response paerns across ethnic backgrounds andacross countries. Three studies were conducted with U.S. respondents. Study 4 was conducted internaonally(Belgium, France, Germany, Italy, Spain, UK)...2.8.3 Esmang floang voters: a comparison between the ecological inference and the survey methods.L. Russo 11 Scuola Superiore Sant Anna di Pisa, ItalyIn the electoral field, one of the most frequently asked quesons is about the floang voters. How manyvoters change their preference from one elecon to another? There are two main approaches to esmate thequota of the floang voters: the survey (by asking which party the interviewed voted in the present and in theformer elecons) and the ecological esmates. Both the methods have their advantages and their problems.The main difficules with the survey are the sample coverage and the memory problems. Instead, by usingthe ecological esmates the problem is the ecological fallacy. The aim of this paper is to verify if the surveyand ecological esmates of vote swing between two elecons are significantly different. For this purpose I willconsider the 2006 and 2008 Italian Parliamentary elecons...2.8.4 The Effect of Proxy-Interviews on Data QualityR. Birkelbach 1 , C. Wolf 21 Universität Mannheim; GESIS - Leibniz Instute for the Social Sciences, Germany; 2 GESIS - Leibniz Instute forthe Social Sciences, GermanySociological research oen depends on interviews given on the behalf of another person (proxy-interviews). Associal-psychological research has shown proxy-interviews have different data quality than self-reports. Thus,it is important to do research on determinants of proxy-interviews in order to be able to adjust the measurehypothesesfor systemac distoron of data quality. Using regression analysis we will assess the micro- andmacro-determinants, as well as interacons, of proxy-interviews in the German subsample of the EuropeanLabour Force Survey (EU-LFS) in the first step. As micro-determinants individual background characteriscsof the target person are used in this study. On the macro-level effects of the household composion areconsidered, because the EU-LFS is a household-survey. Each member of the household is represented in thedata – some of them through proxy-interviews...2.9 Surveying children and young people IIITo be held on July 20, 2011 from: 09:00 to 10:30, in room 321.Coordinated by:• Lisa Calderwood - Centre for Longitudinal Studies, Instute of Educaon, University of London, UnitedKingdom• Kate Smith - Centre for Longitudinal Studies, Instute of Educaon, University of London, United Kingdom2.9.1 Italian validaon of three American scales aimed at measuring children’s well-being.b. tommasi 1 , E. Ruviglioni 11 Università degli Studi di Firenze , ItalyThis arcle focuses on the validaon in Italian context of three internaonal instruments, which aim at studyingchild well-being through the assessment of the percepon they have of their lives. The literature review hasshowed that life sasfacon in childhood has only recently become the focus of work and unl now few studieshave been conducted in this area, especially in the Italian context.

WEDNESDAY 20 JULY 732.9.2 Qualitave research to inform the collecon of sensive data among 11-year olds on the fih waveof the Millennium Cohort StudyL. Calderwood 2 , K. Smith 2 , A. Thompson 1 , E. Wallace 11 Ipsos MORI, United Kingdom; 2 Centre for Longitudinal Studies, Instute of Educaon, University of London,United KingdomMany social research projects involve the collecon of sensive data. Data collecon needs to be conducted inan ethical way that ensures that the comfort, wellbeing and rights of respondents are protected. To ensure dataquality, it is also important that approaches minimise non-response to sensive items and generate accuratereporng, for example, minimising bias towards socially desirability responses. For longitudinal studies it isalso important to minimise the risk that sensive data collecon undermines long term engagement.2.9.3 Youngs and legality: an italian caseE. Macri 1 , E. Ruviglioni 11 Università degli Studi di Firenze , ItalyLegality is currently a fundamental issue in the public debate, in Europe and parcularly in Italy. This crucialissue is strictly connected to two important issues, Public trust and instuonal legimacy. In this work, wepresent the results of a survey project carried out in a High School in Florence and realized by the LaboratoryStaRSE of the University of Florence. The study was focused on the percepon of legality in several societalaspects and concerning the acons aimed at promong sustainability. The quesonnaire is composed by thefour macro areas:1. legality and democracy, 2. environment and environmental protecon, 3. legality, environmentand instuons2.9.4 Surveying young visitors to the memorial sites of the former death and concentraon camps: methodologicalissuesK. Keler 1 , K. Stec 11 Jagiellonian University, PolandThe aim of the paper is to discuss survey design and measurement issues related to studying young people’smovaons, values and meaning of the past on the basis of the experience of conducted mixed method study.The study was carried out in 2009-2010 and includes few supplementary studies to the main survey (N=2355)among young visitors (14+) to the memorial sites of the former death and concentraon camps (Auschwitz-Birkenau, Majdanek, Treblinka in Poland). The survey was conducted just before and immediately aer the visit(to obtain panel data). These addional studies were design to achieve context data to enhance the reliabilityof the interpretaon of the survey data. The measurement was based on the declaraon of young people thusit was crucial to find out if it may be biased due to social desirability and general problems with studying youngpeople...2.10 The role and benefit of structured metadata in survey research IITo be held on July 20, 2011 from: 09:00 to 10:30, in room 414.Coordinated by: Joachim Wackerow - GESIS - Leibniz Instute for the Social Sciences, Germany2.10.1 IPUMS to IHSN: Leveraging structured metadata for discovering mul-naonal census and surveydataW. Thomas 11 University of Minnesota, United StatesThe Minnesota Populaon Center (MPC) is the home of the Integrated Public Use Microdata System (IPUMS)which, along with census microdata from the United States Census Bureau going back to 1850, includes a

74 WEDNESDAY 20 JULYcollecon of census microdata from around the world in IPUMS-Internaonal. This collecon includes datafrom 1960 to date for 55 countries in 159 censuses including approximately 326 million person records. Thedata for each country contains original data (top coded or collapsed to protect confidenality) plus harmonizeddata to support analysis of a country’s data over me or across mulple countries.2.10.2 Disseminaon of survey (meta)data in the LISS data archiveM. Streeerk 1 , S. Elshout 11 CentERdata, NetherlandsIn the year 2007 CentERdata (Tilburg University, the Netherlands) started the MESS project by seng up theLISS panel. The LISS panel is an internet panel representave of the Dutch populaon, consisng of 5,000households. On a monthly basis, the panel members fill out several quesonnaires. One of the main goals ofthe MESS project was to build a data archive making survey (meta)data accessible for other researchers. Ourpresentaon will demonstrate the LISS data archive, which we have created based on DDI 3 (Data Documenta-on Iniave). DDI is a standard for describing data from social, behavioral, and economic sciences. Questasy,an online disseminaon tool that has implemented DDI 3 was developed to disseminate both data as metadatain the LISS data archive. When archiving surveys every aspect of the survey data itself (meta data) is describedin as much detail as possible i.e...2.10.3 Microdata Informaon System MISSYJ. Bohr 11 GESIS - Leibniz Instute for the Social Sciences, GermanyMISSY provides online-informaon for official microdata in a structured design. At present MISSY focuses onthe metadata of the German Microcensus, but an extension to other surveys is intended. Though official microdataare originally not designed for research, the number of data sets, which have been made accessible forscienfic purpose, has increased. To facilitate and encourage the use of the data, there is a need of knowledgetransfer from official stascs to the scienfic community. MISSY offers the relevant metadata both in a broadand differenated way. Moreover, with the availability of annual Microcensus data sets of several decadesinformaon concerning the comparability of variables over me is required.2.10.4 Metadata Driven Survey DesignJ. Iverson 11 Algenta Technologies, United StatesUsing metadata to drive the survey research process enables data science professionals to share their work efficiently.The metadata driven process consists of three principles.1. Don’t repeat yourself. Storing informaonredundantly increases documentaon costs and makes errors more likely.2. Document the enre lifecycle.Codebooks describing datasets are not enough to give a full understanding of the underlying data.3. Makethe metadata accessible. Researchers should be able to discover and reuse data and metadata using standardformats and tools. A metadata driven process increases the transparency of social science research. This enhancescredibility of results and allows new research to take advantage of exisng resources. This talk will alsoreview tools that enable a metadata-driven process.2.11 Collecng Physical Measure and Biomarker Data in Surveys IITo be held on July 20, 2011 from: 09:00 to 10:30, in room 318.Coordinated by:• Heidi Guyer - University of Michigan, United States• Mary Beth Ofstedal - University of Michigan, United States• Barbara Schaan - University of Mannheim, Germany

WEDNESDAY 20 JULY 752.11.1 Experiences with using field interviewers vs. trained health personnel to collect biomeasures insocial surveysH. Guyer 1 , M. Ofstedal 11 University of Michigan, United StatesThe inclusion of physical measures and biomarkers in social surveys is a trend that is on the rise. This is dueboth to an increasing awareness of the ulity of integrang health measures in social surveys and the feasibilityof doing so. However, studies have approached the collecon of biomeasures differently in terms of whatmeasures are collected, when they are collected and who conducts the measurements. Measures that arecommonly collected include anthropometric measurements such as height, weight, waist circumference, hipcircumference and occasionally bioimpedence, body fat percentage or skin folds; blood pressure; measures ofperformance such as gait speed, tests of balance such as med stands or chair stands, tests of lung and gripstrength; and the collecon of biological samples for further analysis of biomarkers which may include bloodspots or whole blood draws, saliva, urine or hair samples...2.11.2 The Collecon of Biomarkers in the Survey of Health, Ageing and Rerement in Europe - Findingsand PerspecvesB. Schaan 11 University of Mannheim, GermanyA large variety of informaon of respondents’ physical and mental health has been collected within the contextof the Survey of Health, Ageing and Rerement in Europe (SHARE) from its first wave in 2004 on. Despite theirundisputable value, self-reported and subjecve health indicators turned out not to be unproblemac in internaonalcomparave analyses. The collecon of biometric data contributes to remedying such problems. Thisresearch paper presents analyses with measures of isometric grip-strength – one of the biometric measuresalready available in SHARE to date. Further, the authors discuss the inclusion of other biometric measures(especially via blood samples) into the invesgaonal program of the longitudinally designed SHARE. Relevantsociological problems and quesons (e. g...2.11.3 Incorporaon of Biomarkers and Novel Health Assessments in The Irish LongituDinal Study of Ageing(TILDA)H. Cronin 1 , C. O’Regan 1 , P. Kearney 11 Trinity College Dublin, IrelandThe Irish Longitudinal study on Ageing (TILDA) is an inter-instuonal iniave led by Trinity College Dublin,which aims to produce a significant improvement in the quanty and quality of data relang to older peopleand ageing in Ireland in order to enhance ongoing and future research and innovaon. 8000 Irish adults aged50+ are recruited at baseline and will be assessed every 2 years over a 10 year meline. Respondents firstcomplete a structured interview covering health, economic and social domains. They are then invited to aenda dedicated health centre for a health assessment or have the opon of having a modified assessment carriedout in their own home.2.11.4 Collecng Physical Measures and Biomarkers in India: Evidence from Longitudinal Ageing Study inIndia (LASI) Pilotp. Perinayagam 11 Internaonal Intsitute for Populaon Sciences , IndiaThe inclusion of biomarkers in health and social surveys is parcularly important for India, where access tohealth care tends to be limited. India’s aging populaon is at risk for undiagnosed diseases. With low literacylevel in this age group a significant level of self reporng bias exists. To deal with these challenges, direct healthexaminaons and biomarkers are crical to measure health status of elderly populaon. In the Longitudinal

76 WEDNESDAY 20 JULYAgeing Study in India (LASI), the goals of biomeasures are to (a) capture health data from populaon thatotherwise would not have this type of data recorded; (b) invesgate molecular determinants of common healthoutcomes; and (c) study interacons between biomarkers and economic condions and social circumstances.2.12 IncenvesTo be held on July 20, 2011 from: 09:00 to 10:30, in room 412.Coordinated by: Kathrin Kissau - FORS - Swiss Foundaon for Research in Social Sciences, Switzerland2.12.1 Incenves and response rates - experience from the SOEP-innovaon-sample 2009J. Schupp 1 , M. Kroh 21 SOEP - German Instute for Economic Research (DIW Berlin), Germany; 2 German Instute for Economic Research(DIW) and Humboldt-Universität zu Berlin, GermanyThe impact of different forms and values of incenves on response rates, unit non response bias and variousmeasures of data quality like e.g. item non response, has been at the heart of survey research for a long me.In the context of longitudinal surveys, the impact of respondent incenves on cooperaon rates is of parcularinterest, as compared to cross seconal ”(”one-wave”) surveys, the choice of incenves for respondents at anygiven wave t not only affects cooperaon rates in that wave but also cooperaon rates of subsequent wave(s)t+n. Bearing that in mind, the trend towards declining response rates in general populaon surveys defines acore challenge for new panel surveys: the ”non response errors” of an inial panel wave will have an ’imprint’as a kind of ”panel legacy effect” in sample structure of all subsequent waves...2.12.2 The impact of monetary incenves on compleon and data quality in online surveysF. van Veen 1 , S. Saler 1 , A. Göritz 21 Bielefeld University, Germany; 2 University of Würzburg, GermanyResearch queson: How to improve data quality and compleon rates in online surveys? Design: We conductedan online experiment in which 1,750 students were randomly assigned to one out of 4 treatments anda control group. Group 1 received a postal prenoficaon of the survey along with a prepaid voucher. Group2 received a postal prenoficaon and a postpaid voucher. Group 3 received a postal prenoficaon and aprepaid 5 EUR bank note. Group 4 solely received a postal prenoficaon. Group 5 was the control groupand was invited via e-mail. Dependent measures were compleon rate, item-nonresponse, straightlining andwillingness to self-report sensive informaon. Results: Logit models show Group 3 (prenoficaon& bank note) to significantly outperform all other groups with respect to compleon. Treatment had no effecton straightlining. The overall amount of straightlining, however, was low...2.12.3 Award distribuon within a winning team: a study of survey dataM. Zheng 11 University of macau, ChinaIn the distribuon of monetary award within a winning team, players in the team have different judgmentabout the fairness of the distribuon. Using the survey data of the player’s ranking of possible award distribuons,we explore factors influencing players’ judgment over fairness of the distribuon. The exploded logitmodel is used to study the ranked data. We find that players with high contribuon tend to prefer merit baseddistribuon; males and individuals with work experience also tend to prefer merit based distribuon than femaleand students, respecvely; the inial distribuon of the contribuon also influence the preference onaward distribuon.

WEDNESDAY 20 JULY 772.13 Surveys in policy makingTo be held on July 20, 2011 from: 11:00 to 12:30, in room 340.Coordinated by: Joan Font - Consejo Superior de Invesgaciones Cienficas (CSIC), Spain2.13.1 Community relaons in Northern Ireland: from survey to policyP. Devine 1 , G. Robinson 21 Queen’s University Belfast, United Kingdom; 2 University of Ulster, United KingdomThe paper will explore the role of survey research in policy making in Northern Ireland, with parcular referenceto community relaons (beer known internaonally as good relaons). Within a region such as NorthernIreland, which is emerging from 40 years of conflict, community relaons is a key policy area. In addion,with devoluon being a fairly recent event, community relaons policy making connues to be contested.The most recent policy consultaon on ’Cohesion, Sharing and Integraon’ is expected to report within thenext few months and has drawn heavily on atudinal survey data. Specifically, for more than 20 years publicatudes to community relaons has been recorded and monitored using two key surveys: Northern IrelandSocial Atudes Survey (1989-1996) and Northern Ireland Life and Times Survey (1998- present)...2.13.2 Survey and policies in a mullevel context: preliminary evidence from the Spanish caseJ. Font 2 , P. Alarcon 11 IESA (CSIC), Spain; 2 Consejo Superior de Invesgaciones Cienficas (CSIC), SpainFor which policies is survey informaon collected by public administraons? Does the opon for survey usedepend on the centrality of the policy in the public agenda, on the party ideology or on other factors? Whichkind of informaon and for which kinds of populaons is selected? Are the contents of surveys commissionedby public administraons different depending on the diverse territorial levels (local, regional, naonal)?2.13.3 Labour Market Data – New Challenges and User Needs in the Recent Economic CrisisK. Duspivova 1 , M. Zelený 11 University of Economics, Prague, Czech RepublicDuring the current economic crisis, the labour market has changed as a result of the negave spill-over effectof the banking sector crisis on other sectors. Exisng labour market problems became more serious andnew problems emerged. New user needs emerged among policy-makers, with new and urgent demand foraddional or new survey results because it became important to quanfy new labour market phenomenon inorder to be able to carry out labour market policy effecvely under these new condions.2.13.4 The Ulizaon of Public Opinion Research in Government: Policy Insights from Tobacco Control,Biotechnology and the Health Care Debate in CanadaL. Birch 1 , F. Pétry 1 , C. Allison Rothmayr 21 Université Laval, Canada; 2 Université de Montréal, CanadaThe Canadian federal government has invested extensively in public opinion research (POR) since the 1990s.The progressive development of policies governing the producon and use of POR in federal policy and managementwas accompanied by impressive budgets to support custom opinion research iniaves. While theDepartment of Public Works and Government Services oversees opinion research in the federal government,specialized POR units within departments coordinate the commissioning of surveys and focus groups on awide variety of subject maers. The instuonalizaon of POR at the scale of an enre government agencyresponsible for many very different programs allows us to describe POR, how it is used, by whom, and for whatpurpose, based on a large number of cases. We compare the producon and ulizaon of POR in three policyareas: 1. tobacco control; 2. biotechnologies and 3. the future of public health care in Canada...

78 WEDNESDAY 20 JULY2.14 Pung Context into Cross-naonal Research ITo be held on July 20, 2011 from: 11:00 to 12:30, in room 319.Coordinated by:• Ineke Stoop - The Netherlands Instute for Social Research/SCP, Netherlands• Eric Harrison - City University London, United Kingdom2.14.1 When is an event an event? The contribuon of media reporng to the survey contextI. Stoop 1 , E. Harrison 21 The Netherlands Instute for Social Research/SCP, Netherlands; 2 City University London, United KingdomUsers of cross-naonal surveys are aware that atudinal data may be affected by the socio-polical contextprevailing in parcular countries at the me. Some elements of naonal context, such as the rate of unemploymentor of economic growth are easier to measure than others; the data is readily available and straighorwardto merge on to the main dataset. Informaon about major ’events’ taking place during the fieldwork period ismore problemac to operaonalise.2.14.2 What is in the news? A comparison between the tradional and Polical Claims-making Approachof collecon informaon on events in the ESS.H. Fernee 11 The Netherlands Instute for Social Research/SCP, NetherlandsIn the first four rounds of the European Social Survey (ESS) data was collected about events reported in themedia for each parcipang country which resulted in an event database. Again in round five this kind ofevent data will be collected as well as data based on a different method namely the Polical Claims-makingApproach (PCA). Both methods have a different view how to code a media event. In the tradional methodevents where coded as an event itself and in the PCA method the ’claims’ about an event made by a relevantactors are coded.2.14.3 Event Data and Claims Analysis: improving European Social Survey’s media event approachA. Torres 1 , R. Brites 1 , D. Carvalho 21 Instuto Superior do Ciências do Trabalho e da Empresa - ISCTE, Portugal; 2 CIES-IUL, PortugalContextualizing longitudinal and cross-naonal surveys is crucial in order to increase its analycal capacity, notonly resorng to demographic, economic and polical stascs but namely through the collecon of mediaevents. These can be used to ”measure” the social, polical and economic climate before, aer and during thefieldwork period and are therefore an addional value for explaining the differences and changes in opinionsand atudes between and within countries.2.14.4 Reliability in the coding of polical claimsM. Cuesta Azofra 3 , M. Hierro 2 , M. Torcal 11 Universitat Pompeu Fabra, Spain; 2 UPF, Spain; 3 Centro de Invesgaciones Sociológicas (CIS), SpainThe study of the influence that the polical context yields upon respondents’ answers in atude surveys hasbeen a recurring topic of the public opinion literature. Despite this, previous research has devoted lile attenonto the measurement of the polical context on itself. Recent aempts, such as the European SocialSurvey’ events database or the polical claims database, intends to fill this gap. In this regard, coding reliabilityemerges as an issue that must be taken into account very seriously in the process of data creaon.

WEDNESDAY 20 JULY 792.15 Socio-economic variables in cross-naonal surveys: Quality enhancement inmeasurement, documentaon and data disseminaon ITo be held on July 20, 2011 from: 11:00 to 12:30, in room 318.Coordinated by:• Jürgen H.P. Hoffmeyer-Zlotnik - GESIS - Leibniz Instute for the Social Sciences, Germany• Uwe Warner - Centre d’Etudes de Populaons, de Pauvreté et de Poliques Socio-Economiques (CEPS /INSTEAD), Luxembourg• Hilde Orten - Norwegian Social Science Data Services (NSD), Norway2.15.1 How to Measure Respondent’s Labour Status and Labour Market Situaon in Cross-Naonal ComparaveSurveysJ. Hoffmeyer-Zlotnik 2 , U. Warner 11 Centre d’Etudes de Populaons, de Pauvreté et de Poliques Socio-Economiques (CEPS / INSTEAD), Luxembourg;2 GESIS - Leibniz Instute for the Social Sciences, GermanyThe social background variable ”occupaon” is based on the implicit assumpon that the posion of a personin the social structure of a modern society is primarily determined by the employment pursued. Educaonand income are closely ed to the occupaon variable. However, the reason given for the importance ofprofessional acvity and posion is that they are considered important determinants of lifestyle, atudes,and individual and group behavioursIn order to assess occupaonal presge and socioeconomic status, a range of informaon must be gathered.It is not enough to determine whether a person is in employment or not, even when employment is measuredin terms of an acvity that characterises that person.2.15.2 Measuring Employment in the Labour Force Survey – the Case of Marginal EmploymentT. Koerner 11 Federal Stascal Office Germany, GermanyFor the measurement of the employment status in cross-naonal surveys, the Labour Force Concept of theInternaonal Labour Organizaon is the key conceptual basis. It defines employment in an extensive way, i.e.any producve acvity of at least one hour per week has to be covered. The experiences with the LabourForce Survey (LFS) in the European Union have shown that measuring employment according to this conceptis challenging in several respects. One crical aspect is marginal employment. Marginal jobs are difficult tocapture in survey interviews, e.g. because respondents are guided by everyday life’s concepts of employmentand therefore tend to focus on their main social status. For instance side-jobs of pupils and pensioners areeasily overlooked in the interview...2.15.3 The life-phase as a socio-economic variable in cross-naonal surveys: Should we use a nominal oran ordinal scale?K. Komp 11 Prence Instute, University of Lethbridge, CanadaLife-phases are longer-lasng situaons that are characterized by constant acvity paerns. Those life-phasesare commonly used in life-course studies to split the me between birth and death into disnct segments. Theiradvantage is that they allow researchers to gain a quick and concise impression of the situaon an individualis in. It therefore seems worthwhile to further develop variables that can capture life-phases.

80 WEDNESDAY 20 JULY2.15.4 Measuring Family Socioeconomic Status: A Methodological Proposal for the Progress in Interna-onal Reading Literacy StudyD. Caro 11 IEA-DPC, GermanyThe paper proposes an indicator of family socio-economic status (SES) for the Progress in Internaonal ReadingLiteracy Study (PIRLS) 2006. Obtaining a valid and reliable SES indicator is crical to the study of educaonalinequalies within- and between-countries, as they are related to the SES. The SES indicator can help us understandhow inequalies related to SES are configured and can be reduced. The proposed indicator is a compositeof parental educaon, parental occupaonal status, home possessions, and family financial status variables.The theorecal model underlying the SES derived indicator is discussed. The reliability and validity of SES is assessedwith exploratory and confirmatory factor analysis. The cross-cultural comparability of SES is evaluatedwith measurement invariance tests across countries. The applied methods and stascal analyses are readilygeneralizable to other cross-naonal surveys...2.16 Pioneering Survey Translaon: Froner research in quesonnaire translaonand mullingual measurement instruments ITo be held on July 20, 2011 from: 11:00 to 12:30, in room 303.Coordinated by:• Dorothée Behr - GESIS - Leibniz Instute for the Social Sciences, Germany• Brita Dorer - GESIS - Leibniz Instute for the Social Sciences, Germany2.16.1 Avoiding deviaons due to quesonnaire translaons across countries through SQPW. Saris 1 , D. Zavala Rojas 21 Universitat Pompeu Fabra, Spain; 2 RECSM Universitat Pompeu Fabra, SpainDesigning a quesonnaire involves many decisions that influence the quality of a survey. Some have to do withthe meaning of the quesons others have to do with the form. Translaon processes in cross-seconal studiescan affect the comparability of the survey data across countries and over me when the translaon changesthe form of the request for an answer. A problem in comparave research is that it is difficult to check thecomparability of the quesons because of the different languages.2.16.2 Translaon Verificaon in the European Social Survey – help or hindrance?S. Widdop 1 , B. Dorer 2 , R. Fitzgerald 11 City University London, United Kingdom; 2 GESIS - Leibniz Instute for the Social Sciences, GermanyThroughout all stages of the cross-naonal survey life cycle there are threats to measurement equivalence.Errors can occur during the draing, pre-tesng or translaon stages of quesonnaire design as well as duringadministraon itself. In terms of quesonnaire translaon, there is increasing consensus that a commiee approachis opmal. The European Social Survey (ESS) adopted a commiee approach when it was establishedin 2001(1) and has connued to use this approach in all five rounds to date. Despite using such a thorough approachwe know that translaon errors sll occur and that queson quality differs significantly across countries(2).2.16.3 Languages and dialects of naonal surveys in the ESSB. Kleiner 2 , S. Buerli 1 , I. Renschler 21 FORS, Switzerland; 2 FORS - Swiss Foundaon for Research in Social Sciences, Switzerland

WEDNESDAY 20 JULY 81There is a growing awareness on the part of researchers of the importance of language in the chain of data producon,including the need for standardised orally-administered quesonnaires that are fully comprehensiblefor linguisc minories...2.17 Specific sub-groups of non-respondents: who are they and how can we enhancetheir parcipaon ITo be held on July 20, 2011 from: 11:00 to 12:30, in room 413.Coordinated by:• Alexandre Pollien - FORS - Swiss Foundaon for Research in Social Sciences, Switzerland• Michèle Ernst Stähli - FORS - Swiss Foundaon for Research in Social Sciences, Switzerland2.17.1 Nonresponse Bias in the Survey of Youth Percepon of Science and Technology in BogotaE. Bueno Castellanos 11 Colombian Observatory of Science and Technology, ColombiaThe Colombian Observatory of Science and Technology -OCyT- developed, in 2009, a survey about the percep-on of Science and Technology in students of the last two years of high school in Bogotá, Colombia. The surveysampling design was strafied according to the nature of school (official or private). During the data colleconstage, two main sources of nonresponse were detected. The first one, as a consequence of the importantdifference in the response probability according to the nature of school: the survey was implemented in 15out of the 16 official schools included in the original sample (94%), while only 13 out of 31 private schools(42%) allowed to collect informaon. The second source corresponds to students who belong to schools inwhich access was allowed, but did not assist during the days when survey was applied. Esmates, inially,were obtained modifying the original sample sizes by those observed...2.17.2 Do non-parcipants answer similarly as survey-parcipants? - An example of AES-quality-surveyK. Pohjanpää 11 Stascs Finland, FinlandThe quality of Finnish EU-AES 2006 (Adult Educaon Survey) was analysed by making a new contact aer oneyear with those not answered. The sample of AES-quality-survey 2007 was 1 072. The aim of study was to testif the non-parcipants answer survey quesons similarly as parcipants a year earlier. There were tree alternavesto parcipate in the survey at 2007. Namely, the new data was collected by making a personal interview(like AES), or a shorter interview by telephone. If the respondent refused both of the above-menonedchoices he was asked some quesons of survey-parcipang. The response rate of AES-quality-survey was59% (AES 65%). Most of sample persons chose a telephone interview (n=403, 38%), whereas more then everytenth took part to the longer interview (n=138, 13%). 70 persons (7%) answered only the quesons ofsurvey-parcipang...2.17.3 Factors affecng parcipaon in The Irish Longitudinal Study of AgeingG. Savva 1 , I. Clifford 1 , M. Hanly 1 , B. Whelan 11 Trinity College Dublin, IrelandMany factors affect willingness to parcipate in surveys. Here we describe the effect of socio-economic factorson parcipaon in The Irish Longitudinal Study of Ageing (TILDA), the results of a pilot experiment to determinethe effect of varying financial incenves on parcipaon, and the use of an increased incenve amongsubgroups of the populaon in which inial parcipaon was lowest.

82 WEDNESDAY 20 JULY2.17.4 Nonresponse in a Mul-Actor Survey: Evidence from the German Family PanelJ. Brüderl 1 , L. Casglioni 1 , U. Krieger 2 , V. Ludwig 1 , K. Pforr 1 , J. Schröder 31 University of Mannheim, Germany; 2 SHARE MEA University of Mannheim, Germany; 3 Mannheim Centre forEuropean Social Research MZES, University of Mannheim, GermanyMost large longitudinal surveys are designed either as individual surveys or as household surveys. For a studyof families and inmate relaonships though, a design that allows the analysis of dyads of people irrespec-ve of their living arrangements is to be preferred. With such a design specific research quesons can beaddressed, for example living-apart-together partnership arrangements or forms of intergeneraonal supportin both direcons between parents and their grown-up children outside the household.2.18 Measurement validity and reliability in cross-cultural comparison ITo be held on July 20, 2011 from: 11:00 to 12:30, in room 321.Coordinated by:• María-Dolores Hidalgo - University of Murcia, Spain• Jan Dietrich Reinhardt - University of Lucerne, Switzerland2.18.1 Analysing DIF in Polytomous Items by Mantel-Haenzsel and Ordinal logisc regression: An applica-on with PISA atudinal quesonnairesI. Benítez 1 , J. Padilla 1 , M. Hidalgo 3 , S. Sireci 21 University of Granada, Spain; 2 University of Massachuses, United States; 3 University of Murcia, SpainDifferenal Item Funconing (DIF) has received increased aenon by professionals and researchers interestedin internaonal and cross-cultural assessments over the last few decades. DIF analyses can providevalidity evidence of the equivalence level reached by different linguiscs or cultural versions of scale items.There are a wide variety of stascs for detecng DIF in both dichotomous and polytomous items. Fewerstudies have researched into ways of applying DIF stascs to detect DIF in atudinal polytomous items forcross-linguisc or cultural comparisons.The aim of this study was to illustrate how to detect polytomous DIF byMantel-Haenzsel and Ordinal Logisc Regression (OLR) procedures. DIF across English and Spanish versions ofthe seven scales included in the Student Quesonnaire of the Program for Internaonal Student Assessment(PISA, OECD, 2006), was analyzed...2.18.2 The detecon of Differenal Item Funconing in short test: A comparison of LR analysis to IRTLRDIFtechnique.M. Hidalgo 3 , M. López 3 , J. Gómez Benito 1 , J. Padilla 21 University of Barcelona, Spain; 2 University of Granada, Spain; 3 University of Murcia, SpainThere are a growing number of cross-cultural and internaonal surveys in the last decade. The comparisonof psychological and sociological variables through different cultural and linguisc groups is the aim of an increasingnumber of studies. Survey researchers usually compare measures provided by different cultural orlinguisc versions of quesonnaires. To make valid comparison involve first to have developed a carefullytranslaon process, and secondly, to test the level of equivalence reached. The analysis of Differenal ItemFunconing (DIF) is required to test whether the level of metric equivalence has been reached. Test length canaffect the right flagging items with DIF. This paper compares by a simulaon study the effecveness of LogiscRegression analysis and IRTLRDIF for detecng DIF in short tests...2.18.3 How much DIF make total-group score comparisons invalid?J. Padilla 2 , M. Hidalgo 3 , J. Gómez Benito 1 , I. Benítez 21 University of Barcelona, Spain; 2 University of Granada, Spain; 3 University of Murcia, Spain

WEDNESDAY 20 JULY 83As cross-cultural/naonal surveys are growing performed, survey researchers pay more aenon to sources ofmeasurement errors that can undermine validity, among them, the translaon errors survey quesons or thelack of construct overlap across the linguisc or cultural groups involved. Differenal Item Funconing (DIF) canprovide evidence of such sources of measurement errors. Given that survey praconers can seldom removesurvey quesons or scale items flagged DIF, the research queson is: How much DIF can survey praconeradmit without undermining total-group score comparisons? The aim of this work is to figure out how muchDIF can invalid total-group score comparisons...2.18.4 Measurement Equivalence Across Subnaonal Groups: An Analysis of the Concepon of Naonhoodin SwitzerlandO. Sarrasin 1 , E. Green 1 , A. Berchtold 21 Misc - University of Lausanne, Switzerland; 2 University of Lausanne, SwitzerlandWhen comparing survey responses of different groups, measurement equivalence ensures that revealed differences(i.e., in means or correlaons between scores) are substanal and not biased by methodological issues.While measurement equivalence tesng is frequently carried out in cross-naonal studies, it is sll rare whencomparing disnct naonal subgroups within countries. Based on data from three large surveys, the presentstudy however illustrates the importance of within-country measurement equivalence tesng by examiningwhether the measurement of the concepon of naonhood (i.e., drawing boundaries between those who arepart of the naon and those who are not) is equivalent across the two largest linguisc regions of Switzerland.Based on recent literature and referenda results, Swiss German respondents are expected to adhere toa stronger ethnic (e.g., being Chrisan) as well as civic concepon of naonhood (e.g...2.19 Open-ended survey quesons ITo be held on July 20, 2011 from: 11:00 to 12:30, in room 315.Coordinated by:• Nick Allum - University of Essex, United Kingdom• Graham Hughes - University of Surrey, United Kingdom2.19.1 Using qualitave soware to analyse open-ended survey quesons: a guide for quantave analystsG. Hughes 11 University of Surrey, United KingdomThis paper illustrates the ulity of four Computer Assisted Qualitave Data AnalysiS (CAQDAS) packages inanalyzing open-ended survey quesons (OEQs), ATLAS., MAXqda, NVivo and QDA Miner, each of which hasstrengths and weaknesses. It is argued that analyc benefits gained from systemac analysis of OEQs areworth the me required to learn a CAQDAS package. Indeed, it should usually not be necessary to undertakeextensive training as recent modificaons have made some packages parcularly accessible for handling surveydata.2.19.2 A Mixed-Methods Approach to the Analysis of Open-Ended CommentsC. Robins 1 , K. Lee 1 , S. Perry 1 , S. Berkowitz 1 , W. Hintze 11 Westat, United StatesSurvey analysts are oen challenged by open-ended survey comments, which interject textual data into anotherwise quantave analysis task. In this presentaon, we describe our team’s mixed-methods approachto sampling and analyzing lengthy open-ended comments for a large-scale military survey. We hope to showparcipants that qualitave survey data, even in large quanes, can be analyzed rigorously and painlessly. In

84 WEDNESDAY 20 JULYthe summer of 2010, the U.S. Department of Defense engaged Service members about the perceived impactson the military if Don’t Ask, Don’t Tell (the law banning openly gay individuals from military service) wererepealed. Westat supported this effort with numerous data collecon acvies, including an online surveywith a sample of 400,000 Service members...2.19.3 Combining qualitave and quantave survey data to explore public percepons of medical researchN. Allum 2 , P. Stoneman 1 , P. Sturgis 11 University of Southampton, United Kingdom; 2 University of Essex, United KingdomThe primary method by which social sciensts describe public understandings of science is to display marginalsin quantave variables and to ulise mulvariate techniques to explain the variaon observed. In this paper,we propose a different approach which could complement quantave based descripons. By delving intothe images and key concepts people make use of when thinking about science and scienfic issues, we arguethat greater insights can be offered on the cognive and psychological processes at work. This approach ismade possible by using public understanding of science data from the 2010 Wellcome Monitor dataset whichin addion to standard quantave variables, contains open ended quesons. By focusing on the verbamresponses generated from such quesons we use the ALCESTE soware to analyse any underlying paerns inthese responses...2.20 Analysis of Immigraon in Europe ITo be held on July 20, 2011 from: 11:00 to 12:30, in room 315.1.Coordinated by: Mariya Aleksynska - CEPII, France2.20.1 Ethnic Identy and Labor-Market Outcomes of Immigrants in EuropeA. Bisin 2 , E. Patacchini 0 , T. Verdier 3 , Y. Zenou 11 University of Stockholm, Sweden; 2 NYU, United States; 3 PSE, FranceThe aim of this paper is to study the relaonship between ethnic identy and labor-market outcomes of non-EU immigrants in Europe. Using the European Social Survey, we find that there is indeed a penalty to be paidfor immigrants with a strong identy. Being a first generaon immigrant leads to a penalty of about 17 percentwhile second-generaon immigrants have a probability of being employed that is not stascally different fromthat of naves. However, when they have a strong identy, second-generaon immigrants have a lower chanceof finding a job than naves. In terms of ethnic identy, we find that speaking a language at home differentthan that of the majority is always harmful in terms of employment. A strong aachment to religion has alsoa negave impact on employment while a strong aachment to tradions and customs does not seem to playa significant role...2.20.2 Locaon choice of immigrants in Belgium 1990-2007I. Ruyssen 1 , G. Rayp 1 , H. Jayet 2 , N. Ukrayinchuk 31 Ghent University, Belgium; 2 University Lille 1, France; 3 Université Lille 2, FranceThis paper analyses the locaon choice of immigrants living in Belgium between 1990-2007 and aims at separangthe so called ”network effect” from other locality-specific characteriscs. The Belgian populaon registerconstutes a rich database of migrant inflows and stocks broken down by naonality and age cohort, whichallows us to disnguish the immigrants of working age. Using these data, we empirically explain the numberof immigrants arriving in each of the 43 governmental districts as well as the 588 municipalies. The networksize is idenfied by the number of previous arrivals in the same locaon, whereas other local, geographicallyspecificcharacteriscs – such as the local labor-market condions or the presence of higher-quality amenies

WEDNESDAY 20 JULY 85– are assumed to be me invariant covariates that generate aracveness effects that can be measured usinglocaon specific fixed effects...2.20.3 Measurement of atudes toward integraon of immigrants using the EVS data from 1999 and 2008M. Valentova 11 Centre d’Etudes de Populaons, de Pauvreté et de Poliques Socio-Economiques (CEPS / INSTEAD), LuxembourgLuxembourg is one of the European countries with the largest proporon of immigrants in the populaon andwith a relavely long immigraon history. According to the latest official stascs, in 2009 the immigrantsrepresented 44 % of the Luxembourg populaon. In this context, the issues of integraon of immigrants andsocial cohesion are high on the polical and research agenda. This paper aims to contribute to these discussionsby construct a reliable and valid measurement of atudes toward integraon of immigrants and by showinghow these atudes differ among different groups of Luxembourg residents and how they have evolved duringpast ten years.2.20.4 Occupaon-Educaon Mismatch of Immigrant Workers in Europe: the Role of Home and Host CountryCharacteriscsM. Aleksynska 1 , A. Tritah 21 CEPII, France; 2 GAINS-TEPP, FranceThis paper documents the extent of the occupao n-educaon mismatch of immigrant workers in the EU inthe past decade, and analyzes its determinants. We consider three broad sets of factors: individual immigrantcharacteriscs; country of residence effects; and migraon decade-specific country of origin characteriscs.First, we find that immigrants have a higher probability of being both over- and under-qualified as opposedto the nave-born, with only marginal evidence of assimilaon. Second, we examine two types of countryof residence effects: general labor market and economic condions, such as labor market rigidies and tradeunion coverage, that may be relevant for the occupaon-qualificaon mismatch of both immigrants and naveborn;as well as migrant-specific effects, such as policies for recognion of diplomas, labor market integraonand andiscriminaon...2.21 Panel data analysisTo be held on July 20, 2011 from: 11:00 to 12:30, in room 414.Coordinated by: Jean-Marie Le Goff - University of Lausanne, Switzerland2.21.1 A model of Polya’s urn applied to panel dataJ. Le Goff 11 University of Lausanne, SwitzerlandAn invesgaon on the Swiss household panel data show that answers given to some quesons about values,especially values on family, have stable frequency distribuon at the level of the sample years aer years.However, at the individual level, it is not rare that an interviewed person does not give the same answer eachyear. In several cases, the paern of answers given by a person even look random. In this communicaon, wehowever argue that the generaon process of answers given years aer years can be assimilated as a Polya’surn. A Polya’s urn is a probabilisc urn which contains balls of two colours. Balls are drawn not only withreplacement of the ball but also by adding a ball of the same colour. The probability to draw a ball of a specificcolour in a given me of the process then depends of the past of the process...

86 WEDNESDAY 20 JULY2.21.2 An Analysis of Panel Arion and Panel Change on the 2006-08-10 General Social Survey PanelT. Smith 11 University of Chicago, United StatesThe General Social Survey (GSS) now uses a rotang panel design. In each even number year, three componentsare fielded: 1) a new cross-secon with a target sample size of 2,000, 2) the reinterview of respodents from theprevious GSS cross-secon collected two years before, and 3) the second and final reinterview with next mostrecent GSS from four years earlier. Thus, each GSS has a new panel, an intermediate panel, and an endingpanel. This paper looks at the issues of panel bias from differenal arion, net vs. gross change, and theseparaon of true change from unreliability using three-wave panel techniques.2.21.3 The relaonships between Individualism, Naonalism, Ethnocentrism&Authoritarianism in Flanders,analyzed in connuous me by means of the EDM SEMT. Toharudin 1 , H. Oud 2 , J. Billiet 3 , H. Folmer 41 University of Padjadjaran, Indonesia; 2 Radboud Universiteit Nijmegen, Netherlands; 3 Katholieke UniversiteitLeuven, Belgium; 4 University of Groningen, NetherlandsSome authors present evidence that naonalism (N) is not only related to individualism (I) and ethnocentrism(E) but also to authoritarianism (A). Therefore, we expand the naonalism-individualism-ethnocentrism modelby Toharudin et al (2008) by including authoritarianism (A) as a fourth latent variable. A was measured by 7items in two waves and 4 items in one wave giving a total of 18 observed variables measuring A which wereadded to the 48 variables measuring I, N, and E. We prove that the model is idenfied. By means of Mx the EDMsoluon for a model with dri coefficients, fixed growth intercepts, diffusion coefficients and inial parameterswas esmated. We find rather strong reciprocal effects between A and E and also a relavely strong effect ofA on I but no reverse effect from I to A. Whereas relavely small but significant effects from I and E on N arefound, no effect is found from A on N...2.21.4 Daily reports of alcohol use and disease severity: data quality, concepts and analysisS. Draisma 1 , J. van Zaane 2 , J. Smit 31 Vrije Universiteit Amsterdam, Netherlands; 2 Vrije universiteit, ggzingeest, Netherlands; 3 Vrije UniversiteitMedical Center Amsterdam, NetherlandsThe Self Rang-Prospecve Life Chart Method (LCM) is an instrument with which paents with bipolar depressionrecord behavior and mood on a daily base. One goal of the study was to establish the effect of dailyalcohol use on the course of bipolar disorder. Another goal is to establish data quality and its relaon to theuse of concepts and results of longitudinal analyses. During approximately a year, 137 paents with DSM-IVdiagnosed bipolar disorder rated their mood and number of alcohol units consumed. The paper focuses on:2.21.5 A Latent Class Approach for Esmang Labour Market Mobility in the Presence of Mulple Indicatorsand Retrospecve InterrogaonF. Bassi 2 , M. Croon 1 , A. Piarello 21 Tilburg University, Netherlands; 2 University of Padua, ItalyWith panel data analysts can esmate labour force gross flows i.e., transions in me between different states.Measurement errors in the observed state can induce bias in the esmaon of transions, leading to erroneousconclusions about labour market dynamics.A large body of literature on gross flows esmaon is based on the assumpon that errors are uncorrelatedover me. This assumpon is not realisc in many contexts, because of survey design and data colleconstrategies, especially when data are collected by retrospecve interrogaon.

WEDNESDAY 20 JULY 872.22 Human Values IVTo be held on July 20, 2011 from: 11:00 to 12:30, in room 415.Coordinated by:• Constanze Beierlein - GESIS - Leibniz Instute for the Social Sciences, Germany• Eldad Davidov - University of Zürich, Switzerland• Peter Schmidt - University of Giessen, Germany2.22.1 The Impact of Basic Human Values on Support for Instuonalized Principles in European WelfareState InstuonsJ. Kulin 11 Department of Sociology, Umeå University, SwedenTaxaon and distribuon constute main elements of welfare state redistribuve programs such as pensions,sickness insurance and unemployment benefits. As such, instuonalized principles of taxaon and distribu-on reflect core aspects of the parcular redistribuve strategies employed by welfare states in the pursuit ofequality and jusce. While there is an extensive literature on this topic, not as much is known about (1) thepublic’s preferences regarding redistribuve strategies and, in parcular, (2) their underlying moves in termsof basic human values. In other words, what redistribuve strategies do people prefer and why? In this paper,these quesons will be addressed from a cross-naonal comparave perspecve...2.22.2 Individual Values and Value Maps: Are Inglehart’s and Schwartz’s Value Dimensions Congruent orUnique?H. Dobewall 2 , M. Strack 11 University of Goengen, Germany; 2 University of Tartu, EstoniaThe aim of the current study is to examine the relaonship between two sets of value dimensions—those proposedby Ronald Inglehart and Shalom Schwartz—both at the individual and country levels of analysis. Weuse data from the World Values Survey (2005-2008) which includes direct measures for both theories, completelyassessed in 47 representave naonal samples. At the level of individuals (N = 46444), posive correlaonswere found for Schwartz’s Openness dimension with both of Inglehart’s dimensions—Self-Expressionand Secular-Raonal values. The highest correlaon with this Schwartz dimension was obtained, r = .24, aerrotang the Self-Expression score about 39 degrees clockwise. At the naonal level, the ranking of countriesbased on Schwartz’s Autonomy and the Self-Expression dimensions reached a maximum of similarity, r = .82,aer rotang Inglehart’s factor scores 27 degrees clockwise...2.22.3 Assessing refinements of the theory of basic values with a new instrument in ten countriesJ. Cieciuch 1 , M. Vecchione 2 , S. Schwartz 31 University of Finance and Management in Warsaw, Poland; 2 Sapienza Università di Roma, Italy; 3 HebrewUniversity of Jerusalem, IsraelSchwartz’s (1992) theory of human values idenfies ten broad and movaonally disnct types of values commonto people across cultures and sociees. The theory specifies the structure of dynamic relaonships amongthe values, locang them within an integrated circular structure (akin to the circular connuum of colours).The circular structure of values represents a connuum of related movaons with adjacent values sharingmovaonal emphases and distant values being movaonally incompable. The theory holds that it is possibleto paron the connuum of values into broader or narrower sets of disnct values, depending on howfinely one wishes to discriminate. In the current research project, Schwartz and colleagues propose refinementsof the theory and measurement. They paron the connuum into 19 more narrowly defined, discretevalues...

88 WEDNESDAY 20 JULY2.22.4 Quesoning the universality of the structural organizaon of human values in the ESS data: AComment on Bilsky, Janik, and Schwartz (2010).M. Strack 1 , H. Dobewall 21 University of Goengen, Germany; 2 University of Tartu, EstoniaBilsky, Janik, and Schwartz (2010) concluded (1) a ”support [of] the circular structure of basic values …” (abstract),and (2) that ”the circular model fits somewhat less well in less developed sociees” (p. 16). We commenton the conclusions drawn from the data by focusing on the migang quanfier ”somewhat”.2.23 Assessing the Quality of Survey Data IIITo be held on July 20, 2011 from: 11:00 to 12:30, in room 410.Coordinated by: Jörg Blasius - University of Bonn, Germany2.23.1 Vive la différence: What are the Implicaons of Response Non-differenaon in Web-based Studies?R. Thomas 11 ICF Internaonal, United StatesResponse non-differenaon (also known as ’straight-lining’) occurs when respondents evaluate a series ofelements and give the same answer to each element. Non-differenaon has been proposed to result fromsurvey sasficing (Krosnick, 1999) which occurs when respondents select responses that minimally meet thedemands of the task rather than select responses that are opmal. We examined 3 large datasets from U.S. respondents– each had a series of rangs to complete on a single screen within web-based surveys. All quesonswere presented in a grid structure (elements in the rows, responses in the columns) where non-differenaonappears to be most likely to occur. First, across studies, non-differenators ranged from lows of 1% to 7.5%.With regard to means and top 2 box scores, in two studies means did not differ significantly with or withoutnon-differenators, though there was a very small effect for top 2 box scores...2.23.2 Deviaons from Standardized Interviews: An Analysis of Survey Pracces in the German MicrocensusG. Güllner 2 , J. Blasius 2 , S. Köhne-Finster 11 Federal Stascal Office Germany, Germany; 2 University of Bonn, GermanyThe objecve of standardized interviewing is to create uniform condions for all respondents. Only if thecontext of data collecon is standardized and only in case that the smuli are the same across respondents,high data validity can be achieved and comparable data can be collected. The interviewer is thus required toask the survey quesons exactly as they were phrased by the researcher. Among others, they have to read outboth all quesons and all response categories to the respondents without changing formulaons. In pracce,many interviewers seem to violate interviewing rules in pracce, potenally causing data distorons which canhardly be corrected ex post.2.23.3 The research on the mobility of tourists in Sicily: detecon and esmaon of non-sampling errorsA. Oliveri 1 , A. Parroco 11 Università degli Studi di Palermo, ItalyIt is well known that in all research fields non-sampling errors affect the interpretaon of results in oen dramacterms, especially when large samples are drawn. This is also the case of the research on the mobilityof tourists in Sicily, where in years 2009-2010 almost 4000 valid interviews were obtained in the context of a

WEDNESDAY 20 JULY 89larger naonal research project. In the paper the strategies adopted by the research team in order to minimizeand esmate the impact of non-sampling errors on results are discussed, with reference to the modelspecificaon, the frame idenficaon, the measurement process and the non responses treatment.2.23.4 Behavior coding and response latency measurement as methods to detect methodologically-inducedvariaon in surveysJ. Mayerl 1 , P. Sellke 11 University of Stugart, GermanyBehavior coding and measurement of response latencies are two promising unobtrusive methods which canbe used to reach a beer understanding of respondents’ behavior and thus to enhance the data quality ofcomputer-assisted surveys.2.24 Measuring economic variablesTo be held on July 20, 2011 from: 11:00 to 12:30, in room 412.Coordinated by: Francesco Laganà - University of Lausanne, Switzerland2.24.1 Labor market dynamics and panel arionA. Mosthaf 11 Instute for Employment Research (IAB), GermanyThis study invesgates the bias produced by panel arion in models esmang transions between unemploymentand employment. The literature suggests that this bias can be decomposed into a selecon effectand a causal effect of the labor market transions on the probability of panel arion. Models esmang transionrates between labor market states are associated with the selecon effect when unobserved variableshave an influence on both the probability to be in one of the examined labor market states and the probabilityof panel arion. The causal effect exists if individuals changing from unemployment to employmenthave a higher probability of unit nonresponse, for instance because they are less oen at home. We comparedifferent methods for the correcon of arion: propensity weights and Heckman selecon models...2.24.2 Fools rush in where angels fear to tread: Are we ready to choose ALLBUS for analyzing incomedata?M. Terwey 11 GESIS - Leibniz Instute for the Social Sciences, GermanyThis paper offers some informaon about chances and risks in invesgaons which use rather similar cross sec-onal surveys. Income is taken as an example of very important and rather solid stascs for many sociologicalanalyses.2.24.3 Evaluaon of material and economic situaon of older generaons based on EU-SILC surveys inselected Central and Eastern European countriesJ. Perek-Bialas 11 Warsaw School of Economics, PolandThe aim of the paper is to discuss how the EU-SILC surveys (2007 and 2008) could be used in analysis of thesocio-economic situaon of older persons for some countries of Central and Eastern Europe (Poland, Lithuania,Hungary, Czech Republic, Slovakia, Latvia, Estonia). Having the possibility to use individual data from EU-SILC2007 and 2008, we performed an analysis but not only just for presentaon results related to the topic ofquality of life (material deprivaon) and subjecve well being (the queson of ’meeng ends’). We focus also

90 WEDNESDAY 20 JULYon the missing gaps in data/informaon with recommendaons what could be changed, adjusted, removedor added in doing research on this topic for EU-SILC surveys for selected countries also to facilitate futureinternaonal comparisons.2.25 Internet panels for the general populaon: Fact or ficon ITo be held on July 20, 2011 from: 14:00 to 15:30, in room 412.Coordinated by:• Edith De Leeuw - University of Utrecht, Netherlands• Annee Scherpenzeel - CentERdata, Netherlands2.25.1 Surveying the General Populaon: Coverage bias in European Web SurveysE. De Leeuw 1 , A. Mohorko 2 , J. Hox 11 University of Utrecht, Netherlands; 2 Utrecht University, NetherlandsA good coverage of the intended populaon is one of the cornerstones of survey research, and the successof Internet surveys is dependent on how well the intended populaon is covered. Countries differ in Internetpenetraon, threatening the internaonal comparability of web surveys.2.25.2 Noncoverage and Nonresponse in a Probability Based Mixed-Mode Access PanelS. Bartsch 1 , U. Engel 1 , H. Vehre 11 University of Bremen, GermanyThe data quality of online access panels can be threatened by self-selecon processes into the panel (nonresponse),coverage error and mode/ response effects. Self-selecon and noncoverage is likely to lead to biasedsample esmates, while mode effects and mode-specific response effects preclude any generalizaon of outcomesproduced by one survey mode to another. To study these effects, we built up a large access panel forthe adult populaon of Germany using probability sampling for the recruitment of people by phone (landlineand cell phones). Possible access panel modes are landline, cell phone and web. The project is part of thePriority Programme 1292 on ’Survey Methodology’ (see www.survey-methodology.de).2.25.3 Capability and Movaon: Analysis of Heterogeneity in the Quality of Survey Data using an Experimentin the HRS 2009 internet Post Crash SurveyR. Willis 1 , G. Kezdi 21 University of Michigan, United States; 2 Central European University, HungaryThe quality of a survey parcipant’s answer to a given survey queson depends on both the person’s capabilityof answering the queson and his or her movaon to do so. Survey designers can influence the quality of theanswers given either by affecng the person’s ability to answer the queson (e.g., by using clear language orby reducing the difficulty of what is being asked) or by affecng the person’s movaon to answer the queson(e.g., by emphasizing the importance of the queson or providing an explicit reward for a correct answer).2.25.4 Comparing (Non-)Response Between a Sequenal Mixed Mode Design and a Probability BasedInternet PanelK. van der Houwen 1 , G. Mars 11 Stascs Netherlands (CBS), NetherlandsTradionally, there has been a strong focus within Stascs Netherlands (CBS) on collecng objecve data. Fewsurveys have incorporated quesons about people’s feelings, atudes or opinions. However, policy makers

WEDNESDAY 20 JULY 91and the general public increasingly call for this type of subjecve informaon. Moreover, general opinion isthat this informaon should be current. Stascs Netherlands is now exploring how to best meet this need.2.26 Prevenng non-response on longitudinal surveys by improving survey prac-ceTo be held on July 20, 2011 from: 14:00 to 15:30, in room 414.Coordinated by: Lisa Calderwood - Centre for Longitudinal Studies, Instute of Educaon, University of London,United Kingdom2.26.1 Non-response in the 1970 Brish Cohort Study from birth to 34 yearsS. Ketende 1 , J. W McDonald 1 , S. Dex 11 Centre for Longitudinal Studies, Instute of Educaon, University of London, United KingdomLongitudinal data have the potenal to offer social science and policy makers many benefits. However, forthe potenal of longitudinal data to be fully achieved, data need to be collected systemacally over me fromthe same individuals. The analysis potenal of longitudinal data is put at risk if individuals drop out of thesuccessive data collecon contacts, more especially if they drop out in large numbers, and if those who fail toconnue are a biased sample of the original sample.2.26.2 Using informaon for previous waves to inform response maximisaon strategies on the fih waveof the Millennium Cohort StudyL. Calderwood 3 , A. Cleary 1 , K. Gallop 1 , J. W McDonald 3 , P. Smith 21 Ipsos MORI, United Kingdom; 2 University of Surrey, United Kingdom; 3 Centre for Longitudinal Studies, Instuteof Educaon, University of London, United KingdomAn advantage of longitudinal surveys is that it is possible to analyse the response characteriscs of the sampleusing survey data or para-data from previous waves, and so potenally make predicons about sample members’propensity to respond at future waves. This informaon can be used to help inform fieldwork pracce asappropriate intervenons can tailored and targeted on specific groups less likely to respond. These interven-ons in survey pracce may include case priorisaon, the use of tailored materials and varying the methodused for inial contact.2.26.3 An Experimental Test of the Effects of Incenves on Contact Update Response RatesK. McGonagle 2 , R. Schoeni 2 , M. Couper 11 University of Michigan, United States; 2 Instute for Social Research, University of Michigan, United StatesThe Panel Study of Income Dynamics (PSID) is a naonally representave longitudinal survey of approximately9,000 families ongoing since 1968. Since 1969 families have been sent a mailing asking them to update or verifytheir contact informaon to keep track of their whereabouts between waves. Analysis shows that updatedcontact informaon prior to data collecon is cost effecve, yielding less tracking, refusal conversion, fewercontact aempts, and lower arion. Given these advantages, two studies were designed with the goal ofimproving the response rate of the contact update mailing. The first study in 2008 showed that including afollow-up mailing for non-responders improved response rates by 7-10% but was inconclusive in regards tothe effecveness of incenve payments. Moreover, the overall importance of incenves could not be tested...2.26.4 Results from an experiment to increase the effecveness of between-sweep cohort maintenancemailings on the Millennium Cohort StudyL. Calderwood 11 Centre for Longitudinal Studies, Instute of Educaon, University of London, United Kingdom

92 WEDNESDAY 20 JULYThis paper presents results from an experiment to increase the effecveness of between-sweep cohort maintenanceprocedures on the Millennium Cohort Study (MCS). The MCS is following over 19,000 children bornin the UK in 2000/1. So far there have been four waves of the study at 9 months, 3 years, 5 years and 7 years.The fih wave will take place in 2012 when the study children will be aged 11 and in their final year of primaryschool.2.27 Challenges and responses in cross-cultural and comparave quesonnairepretesngTo be held on July 20, 2011 from: 14:00 to 15:30, in room 340.Coordinated by:• Kristen Miller - Naonal Center for Health Stascs , United States• Jose Luis Padilla - University of Granada, Spain2.27.1 How to combine cognive interview and field test methodologies to validate evidence for surveyquesons?J. Padilla 2 , K. Miller 1 , M. Loeb 1 , A. Maitland 31 Naonal Center for Health Stascs , United States; 2 University of Granada, Spain; 3 Centers for Disease Controland Prevenon, United StatesAs the use of cognive interview (CI) methodology to improve survey quesons is increasing, survey researchersand praconers are faced with the challenge of how to combine qualitave CI evidence with quantavefield test results. Beyond the debate about the methodological status –qualitave vs. quantave-, mulplebenefits can be obtained by combining evidence from both methods in a queson evaluaon project. In crossculturaland mul-naonal surveys, a mixed-method approach can provide evidence of the prevalence of theinterpretaon paerns found by CI, and the extent to which differences in understanding the queson’s meaningcan be linked to different type of response biases or to real differences in the intended construct. The aimof this paper is to illustrate how to use a mixed-method approach that combines CI and field test methods...2.27.2 Cognive interviewing in web surveys: the use of probing quesons in cross-naonal web surveysD. Behr 1 , M. Braun 1 , L. Kaczmirek 1 , W. Bandilla 11 GESIS - Leibniz Instute for the Social Sciences, GermanyEquivalence is a necessary prerequisite of any substanve analysis of cross-cultural survey data. Cognive interviewingis a suitable method when it comes to idenfying non-equivalence in survey data and, parcularly,its causes. Usual face-to-face cognive interviews have their shortcomings, though, e.g. their use only as apretesng device, limited sample sizes, great me and cost investment, or interviewer effects due to differentinterviewer behavior. In cross-naonal research, these problems are exacerbated and, thus, it does not comeas a surprise that the use of cognive interviewing in the comparave context brings with it parcular challenges.We now propose to test web surveys as a means to conduct cross-cultural cognive interviewing. Websurveys permit to counter the above menoned shortcomings: they allow e.g...2.27.3 Evaluang the Evaluaon: What Constutes Sufficient Evidence of Findings in Cognive InterviewReports?S. Willson 11 Naonal Center for Health Stascs , United StatesFinal reports of cognive interview studies are oen wrien and available for review; however, wide varia-on exists in the organizaon, informaon included, and level of detail present in cognive interview reports.

WEDNESDAY 20 JULY 93There is no standard format. This variaon impedes users’ ability to determine the trustworthiness and veracityof research findings. This is problemac not only in the assessment of a single report, but also in theassessment of different reports tesng the same survey quesons. If different reports have dissimilar findingsfor similar quesons, it is difficult to directly compare the conclusions if the reader cannot adequately evaluatethe findings. This paper addresses this issue and discusses factors that make it easier for readers to evaluatethe findings of a cognive evaluaon final report.2.27.4 Strategies for Analyzing Cognive Interviews in Queson Evaluaon StudiesK. Miller 1 , J. Padilla 21 Naonal Center for Health Stascs , United States; 2 University of Granada, SpainDrawing on the work performed by The Granada Group, a coalion of survey methodologists who have conducteda large-scale mulnaonal cognive interviewing study, this paper will describe various ways in whichcognive interview data can be analyzed for queson evaluaon studies. Addionally, it will discuss the advantagesand disadvantages of the various strategies for studies that require an assessment of comparabilityfor cross-cultural or mulnaonal surveys. As with quantave methods, there are numerous ways in whichan analyst can go about examining cognive interview data, and the method of analysis must be consideredin the context and purpose of the cognive interviewing project. Most importantly, and just as in quanta-ve methodology, how an analyst actually goes about conducng analysis of cognive interviews informs thefindings of that study...2.28 Specific sub-groups of non-respondents: who are they and how can we enhancetheir parcipaon IITo be held on July 20, 2011 from: 14:00 to 15:30, in room 413.Coordinated by:• Michèle Ernst Stähli - FORS - Swiss Foundaon for Research in Social Sciences, Switzerland• Alexandre Pollien - FORS - Swiss Foundaon for Research in Social Sciences, Switzerland2.28.1 Influence of the contact trials and me of contacts on the parcipaon of certain sub-populaongroups in a telephone health interview surveyP. Schmich 1 , E. von der Lippe 11 Robert Koch Instute, GermanyThe relaon between non-response and data quality in the social surveys is broadly discussed and invesgated.Sciensts apply numerous methods for improving response and the sample designs.In our study we define sub-groups of respondents which are difficult to contact and convince to parcipatein a health telephone survey. We use meta-data from the telephone interview survey German Health Updateconducted by Robert Koch Instute. The survey was carried out from July 2009 ll June 2010 and has about22 000 parcipants aged 18 and older. We examine the parcipaon of the ’problemac’ groups according tonumber of contacts and the me of contact.2.28.2 Lifestyle, accessibility and cooperaon of respondentsM. Sapin 1 , D. Joye 2 , A. Pollien 1 , M. Ernst Stähli 11 FORS - Swiss Foundaon for Research in Social Sciences, Switzerland; 2 University of Lausanne, SwitzerlandNon respondents’ universe is characterized by a great heterogeneity. Literature disnguishes two main causesfor non-parcipaon: non-contact and refusal. Accordingly, research reduces most of the me the nonrespondents’universe into two separate groups, differenang the ”hard to reach” respondents from thosewho refuse to cooperate, with the aim of assessing substanal features of both groups based for example on

94 WEDNESDAY 20 JULYnon response surveys or paradata. However, this dichotomy is insufficient to depict the variability and heterogeneityof non-response. There is ambiguity in classifying some non-respondents, for instance the respondentswho postpone endlessly the appointment that put off interview forever. In this paper, we study the reluctantand hard to reach respondents, based on Swiss data of the ISSP and the ESS...2.28.3 Strategies to reduce response bias measuring subjecve probabiliesF. van Veen 1 , S. Saler 1 , M. Beuer-Krüssel 21 Bielefeld University, Germany; 2 No organisaon atm, GermanyResearch Topic: One of the concerns in raonal choice modeling and research on risk is to find adequatemeasures of subjecve probabilies. These are in parcular important to explain deviant behavior. Due tocognive heuriscs of respondents measures are at risk of errors like base rate neglecng and overesmaon(Hoffrage et al. 2000; Tversky/Kahneman 1974; Yamagishi 1997). The use of preceding anchor quesons isknown to adjust for these shortcomings.2.28.4 Different biases due to different causes of nonresponse: The use of complete administraon datafor nonresponse analysisR. Schnell 2 , T. Gramlich 2 , A. Mosthaf 3 , S. Bender 11 Research Data Centre (FDZ) of the Federal Employment Agency at the Instute for Employment Research,Germany; 2 University of Duisburg-Essen, Germany; 3 Instute for Employment Research (IAB), GermanyNonresponse effects in a German large scale mixed-mode (CATI/CAPI) survey of welfare receiving households(PASS) were studied by linking the survey to complete social security data on households and individuals forboth respondents and nonrespondents. Weighted and imputed survey data were compared with administra-ve data. For the first me in Germany, the effect of different causes of nonresponse could be studied withknown administrave data for nonrespondents. About 26% of the 49,215 persons contacted responded to thesurvey. Different biases on different variables depending on the cause of nonresponse (refusal, illness, moved,noncontact, language problems) were observed.2.29 Pung Context into Cross-naonal Research IITo be held on July 20, 2011 from: 14:00 to 15:30, in room 319.Coordinated by:• Eric Harrison - City University London, United Kingdom• Ineke Stoop - The Netherlands Instute for Social Research/SCP, Netherlands2.29.1 The climate of crisis. Monitoring events and atudes in a cross-naonal perspecve.T. Stathopoulou 11 NCRS, GreeceEvent monitoring has been a valuable source of informaon during the successive rounds of European SocialSurvey. An overview of the reported events can provide the researcher with a comprehensive picture of theclimate within which the survey is fielded in each parcipang country. Through event reports one can traceand compare the shi of focus on certain events in between and across countries.2.29.2 Long-term events and coding media reports in PolandT. Zmijewska-Jedrzejczyk 11 Polish Academy of Sciences, Poland

WEDNESDAY 20 JULY 95Events may have a significant bearing on social beliefs. An excellent illustraon for Poland comes from ’Rywingate’,i.e. a corrupon scandal which evolved into a polical scandal and was the main driver of change onthe Polish polical scene and considerably undermined public trust in policians in general. While it is commonlyagreed that social and polical events do shape opinions, it is sll quite problemac to decide whichones should be selected and how they should be described to enable researchers to draw conclusions aboutthe effect of such events on ESS responses. In Poland, event reports have been prepared since the first round ofthe European Social Survey. Addionally, the ESS team in Poland - as one of few teams -have prepared reportsfrom events between the ESS rounds. Those reports cover longstanding issues expected to have a significantimpact on responses in the subsequent ESS round...2.29.3 Unemployed and Alone: Unemployment and Social Parcipaon across EuropeM. Dieckhoff 2 , V. GASH 11 The Cathie Marsh Centre for Census and Survey Research, University of Manchester, United Kingdom; 2 WissenschaszentruBerlin für Sozialforschung (WZB), Germany2.29.4 The introducon of the euro currency in Slovakia and its reflecon in the ESS R4 dataD. Fedáková 1 , M. Kentoš 1 , J. Výrost 11 Slovak Academy of Sciences, SlovakiaThe fieldwork of the ESS round 4 started on 17 November 2008 and finished on 15 February 2009. In themiddle of the fieldwork (on 1 January 2009) Slovakia joined the eurozone. According to the survey of SlovakStascal Office 83% of Slovak cizens posively perceived the introducon of the euro currency. The ESSdata analyses were performed to show the response tendencies in selected items (e.g. media watching, trustin polical instuons, sasfacon with naonal and European parliament, subjecve well–being, perceivedcontribuon of people of different age to economy) in a three different me periods (before, during, and aerthe introducon of the new currency) to demonstrate the possible impact of the event on the responses.2.30 Measurement validity and reliability in cross-cultural comparison IITo be held on July 20, 2011 from: 14:00 to 15:30, in room 321.Coordinated by:• María-Dolores Hidalgo - University of Murcia, Spain• Jan Dietrich Reinhardt - University of Lucerne, Switzerland2.30.1 Mul groups Confirmatory Factor Analysis of Knowledge, Atude and Pracce towards UniversalPrecauon (KAP-UP) related to Blood-borne InfeconE. Fitriana 4 , Y. Suparman 6 , K. Ibrahim 1 , L. Pinxten 2 , D. Sunjaya 3 , A. Meheus 51 Padjadjaran University, Faculty o Nurse, Indonesia; 2 IMPACT, Indonesia; 3 Faculty of Medicine PadjadjaranUniversity , Indonesia; 4 Padjadjaran University , Faculty of Psychology, Indonesia; 5 University of Antwerp, Belgium;6 Padjadjaran University, Faculty of Natural Scineces, IndonesiaBackground: No study has tested a factor structure of Knowledge, Atude and Pracce towards UniversalPrecauon (KAP-UP) related to blood-borne infecon among two groups of professional health care workers.Objecves: To test the differences factor structure of KAP-UP among physicians and nurses.2.30.2 Linking qualitave assessments and quantave tesng of translated/adapted items in compara-ve surveysP. Mohler 11 University of Mannheim, Germany

96 WEDNESDAY 20 JULYThe rise of cognive interviewing and focus groups in survey instrument design is accompanied by scepcalcomments of stascians. They are concerned that the insights of cognive interviewing and other relatedqualitave pretesng techniques cannot be directly linked to measurement properes such as (stascal) reliabilityand validity. The major issue here is, that qualitave work basically deals with paern-idenficaonwithout indicang the paern distribuon in a given populaon as it is standard in stascs.2.31 Assessing the Quality of Survey Data IVTo be held on July 20, 2011 from: 14:00 to 15:30, in room 410.Coordinated by: Jörg Blasius - University of Bonn, Germany2.31.1 Happy respondents, happy researchers? What self-disclosure and response-style can tell us aboutthe quality of our data.J. Rossmann 2 , J. Blumenberg 11 Mannheim Centre for European Social Research MZES, University of Mannheim, Germany; 2 GESIS - LeibnizInstute for the Social Sciences, GermanyEven though online surveys become more and more popular in social scienfic research, there is much doubtabout the quality of the generated data. This has much to do with the characteriscs and the response-styleof online-respondents which are harder to control than in other survey modes.2.31.2 Assessing the Quality of Face-to-Face Interviews Using a Mul-Dimenonal ModelM. Link 1 , T. Boals 1 , J. Lai 11 Nielsen, United StatesFace-to-face interviewing is thought to produce the highest quality survey data compared to other modes, yetrigorous assessments are rare and oen limited in scope. Interviews with non-tradional populaons, such asestablishment owners where recall may bias collecon of specific factual informaon, oen make assessmenteven more difficult. We provide a mulfactor conceptual framework for assessing the quality of face-to-faceinterviews in an ongoing panel of store owners...2.31.3 Quality control in web surveysM. Romano 1 , M. Nalli 11 Scuola Superiore Sant Anna di Pisa, ItalyThe development of computer-assisted methods for data collecon has provided survey researchers tools tocontrol the process of data collecon: monitoring the stream of survey data and also paradata could be usefulto improve survey cost efficiency and achieve more precise, less biased esmates. Furthermore, at the end ofthe survey process measures and paradata can be used to evaluate survey quality.2.31.4 Telephone interviewer training: How does it link to data quality?W. Jablonski 11 University of Lodz, PolandThe aim of this presentaon is to outline the results of the methodological study that was carried out duringthe period from October 2009 to August 2010. 12 major Polish research organizaons as well as 2 companiesin Norway and Iceland parcipated in the research (Norwegian and Icelandic modules of the project wereco-financed by technical assistance funds of the EEA Financial Mechanism and the Norwegian Financial Mechanismwithin the framework of the Scholarship and Training Fund). The research was based on three methods:

WEDNESDAY 20 JULY 97a self-compleon quesonnaire for CATI interviewers (942 cases were collected), a self-compleon quesonnairefor CATI studio managers, and In-Depth Interview with well experienced CATI interviewers (49 interviewswere conducted)...2.32 Analysis of Immigraon in Europe IITo be held on July 20, 2011 from: 14:00 to 15:30, in room 315.1.Coordinated by: Mariya Aleksynska - CEPII, France2.32.1 The Immigrant/Nave Wealth Gap in Germany, Italy and LuxembourgA. Porpiglia 2 , T. Mathä 2 , E. Sierminska 11 Centre d’Etudes de Populaons, de Pauvreté et de Poliques Socio-Economiques (CEPS / INSTEAD), Luxembourg;2 Banque centrale du Luxembourg, LuxembourgWealth is a comprehensive measure of household financial behaviour and economic wellbeing. It gives familiesthe possibility to rely on a buffer stock in mes of diminished income streams, to access to beer schools andenhanced health facilies. In addion it is a crucial resource to maintain the living standard in rerement.2.32.2 How Successful are Different Migrant Groups on European Labor Markets? A Comparison of Attainmentacross 28 European SocieesF. Pichler 11 University of Vienna, AustriaTypical labor market outcomes vary considerably between majority and migrant populaons. Drawing onscholarship from across the social sciences, we invesgate compeng micro and macro-level explanaons ofdifferenal occupaonal aainment among immigrant groups across 28 countries. To beer understand lowerlabor market outcomes of migrants, this paper empirically assesses a series of potenal individual and structuraldeterminants of outcomes across Europe, including human capital, social mobility, migraon backgroundon the one hand and immigraon policies, welfare regimes and economic condions on the other. The analysesof occupaonal aainment are run separately for first and second generaon migrants as well as childrenof mixed marriage to provide more meaningful insights into their variegated inroads into the labor markets ofdesnaon countries...2.32.3 Migrant-specific labour market segregaon and the earnings of immigrants and naves in GermanyB. Heizmann 1 , A. Busch 3 , E. Holst 21 Jena Graduate School GSBC, Germany; 2 German Instute for Economic Research (DIW), Germany; 3 GermanInst. for Economic Research (DIW) / Berlin Graduate School of Social Sciences (BGSS), GermanyThis paper examines the influence of migrant specific labor market segregaon on the wages of migrants andnon-migrants in Germany. Using data from the German Socio-Economic Panel Study and the German Microcensuswe test several sociological and social-psychological explanaons relevant to this topic. The noon ofQuality Sorng states that the differences in wages that are associated with the migrant share within occupa-ons are due to differences in human capital and qualificaon requirements in those jobs alone. In contrast,a Cultural Devaluaon of occupaons with a high migrant share should have a further negave influence onboth migrants’ and naves’ wages...2.32.4 The Relaonship between Immigraon and Vicmizaon in England and Wales: Evidence from theBrish Crime SurveyG. Papadopoulos 1

98 WEDNESDAY 20 JULY1 University of Essex, Department of Economics, United KingdomThis study, using data from the BCS, comprehensively examines whether vicmizaon paerns are differentbetween immigrants and naves. We first find that the probability of being a vicm of a burglary or a personalthe is higher for immigrants, but this can be perfectly explained by the fact that immigrants exhibit somedemographic characteriscs associated with higher vicmizaon. Contrary to the above, we interesngly findthat immigrants are of lower risk of violent vicmizaon. As violence is an expressive type of crime, whereinteracons between vicm-offender pairs prior to the crime act maer much more than instrumental crime,the lower risk of violence faced by immigrants could be aributed to different lifestyle choices associated withlower vicmizaon risks...2.33 Human Values VTo be held on July 20, 2011 from: 14:00 to 15:30, in room 415.Coordinated by:• Constanze Beierlein - GESIS - Leibniz Instute for the Social Sciences, Germany• Eldad Davidov - University of Zürich, Switzerland• Peter Schmidt - University of Giessen, Germany2.33.1 Tesng the Relaonship between Social Status and Religiosity in Chrisan Europe - A LongitudinalPerspecveR. Klanjsek 1 , S. Flere 11 University of Maribor, SloveniaThe correlates and possible social origins of religiosity are sll disputed. These relaonships can also be indicaveof plausibility of secularizaon. Flere and Klanjšek (European Sociees, 2009, 583-602), using WVSSurvey data (wave 1999-2001; n= 48.263) found an inverse relaonship between religiosity and income andreligiosity and educaon in European countries with a Chrisan heritage (Eastern Orthodox, Protestant andRoman Catholic). The results lent limited support to the classical deprivaon thesis, since in none of the samplesanalyzed a posive associaon between high social status and high religiosity could be found. Using thelast release of the EVS data (2008), the same analyses on the same countries will be carried out in order to testwhether the relaonships are stable over me.2.33.2 Human Values – Typologies of PreferencesT. Blank 2 , A. Pöge 1 , H. Murawski 21 Bielefeld University, Germany; 2 University Bielefeld, GermanySchwartz’s conceptualizaon of ten basic human values suggests that they are interrelated systemacally bybeing more or less similar and that this should be reflected in a hierarchy of importance among them. Thispaper deals with the queson, to which extent this conceptualizaon can be found in different groups of populaonusing ESS-2008-data for Germany, France, Sweden and Turkey. In the first step, for each of the fourcountries different homogeneous subgroups are idenfied using latent class analysis. By this, each of thesegroups is characterized by group-specific profiles of importance of the ten value types analysis. Secondly,each of these profiles is compared with a sine curve as the theorecal perfect form of an importance profilecongruent with the theorecal concept...2.33.3 The Temporal Reciprocity of Values and Beliefs: A Longitudinal Study of Polish MigrantsR. Goodwin 1 , E. Polek 2 , A. Bardi 31 Brunel University, United Kingdom; 2 University College Dublin, Ireland; 3 Royal Holloway, University of London,United Kingdom

WEDNESDAY 20 JULY 99Values and beliefs are important parts of the self-concept, but lile previous work has examined the relaonshipbetween the two, parcularly in a longitudinal context. Major life transions – such as moving to a newculture – can challenge exisng values and beliefs, and so provide a parcularly useful context for the analysisof value and belief change. In this study, we quesoned 172 Polish migrants shortly aer their arrival in the UK,and at two, subsequent, nine-month intervals, focusing on the values of tradion, self-direcon and universalism,and the social axiom ’social complexity’. Results revealed six cross-lagged effects from values to socialcomplexity, and three from social complexity to values. Findings are discussed in the light of current debatesover personality change, as well as the broader impact of significant life transions on the self-concept...2.33.4 Values or Adues? A Mixed Method Approach to Measurement Quality of the ESS- and a NewlyDeveloped Version of the Schwartz Human Value ScaleR. Latcheva 1 , G. Kessler 21 University of Zürich, Switzerland; 2 University of Vienna, AustriaValues have been object of research for several decades and different academic disciplines. A number of theorecalapproaches have been developed focusing on the nature and structure of human values, their predicveand explanatory potenal for social change and not at least the universality of their content/structure acrosscultures. Since the study of human values, grounded on the Schwartz Human Value Scale, has also become anintegral part of the European Social Survey, recent debates increasingly centre on issues of cross-cultural/crossnaonalcomparability and the quality of measurement. Studies led by Eldad Davidov could establish (paral)measurement invariance across variety of countries and over me but aer grouping some of the inial values,which seemed necessary due to lack of discriminant validity...2.34 Socio-economic variables in cross-naonal surveys: Quality enhancement inmeasurement, documentaon and data disseminaon IITo be held on July 20, 2011 from: 14:00 to 15:30, in room 318.Coordinated by:• Jürgen H.P. Hoffmeyer-Zlotnik - GESIS - Leibniz Instute for the Social Sciences, Germany• Hilde Orten - Norwegian Social Science Data Services (NSD), Norway• Uwe Warner - Centre d’Etudes de Populaons, de Pauvreté et de Poliques Socio-Economiques (CEPS /INSTEAD), Luxembourg2.34.1 The Internaonal Standard Classificaon of Educaon 2011 and its applicaon in cross-naonalsurveysS. Schneider 11 Nuffield College, United KingdomA connuing challenge of comparave research is the harmonisaon of country-specific educaonal aainmentinformaon into cross-naonally comparable variables. A classificaon that is commonly used in officialand also increasingly so in academic surveys is the Internaonal Standard Classificaon of Educaon, ISCED.In contrast to its academic ’competor’ classificaon CASMIN, ISCED covers almost all countries in the worldand is centrally maintained and documented by the UNESCO Instute for Stascs.2.34.2 The educaonal measure in internaonal Face to Face surveys carried out in Switzerland: towarda precise and internaonally comparable toolM. Ernst Stähli 1 , D. Joye 2 , M. Sapin 1 , A. Pollien 11 FORS - Swiss Foundaon for Research in Social Sciences, Switzerland; 2 University of Lausanne, Switzerland

100 WEDNESDAY 20 JULYFor the latest edion of the European Social Survey (Round 5, 2010), for which the fieldwork in Switzerland hasbeen realized between October 2010 and March 2011, the tool for the measurement of the highest educaonallevel aained (for respondent, partner, father and mother) has been revised in order to obtain an up to dateversale tool, with more informave and internaonally beer comparable indicaons. The basis for thisrevision was the European Survey Version of ISCED proposed by Silke Schneider (2009).2.34.3 The Internaonal Standard Level of Educaon (ISLED): Scaling and validang the ESS educaoncategoriesH. Schröder 1 , H. Ganzeboom 11 Vrije Universiteit Amsterdam, NetherlandsWe propose a novel way of improving the measurement of educaon level in cross-naonal research by opmallyscaling the detailed country-specific educaonal classificaons available in the ESS. Applying cause-andeffect-proporonal scaling, we esmate how the respecve naonal educaon levels are located relave toeach other with regard to a number of inputs (parental educaon and occupaon levels) and a number ofoutputs (occupaonal aainment and educaon partner). We label the thus derived measure ISLED. We findthat the increase in detail achieved by opmal scaling yields a noceably beer measurement of educaonlevel relave to both ISCED representaon and the duraon measure of educaon level. Both are present inthe ESS data.2.34.4 How to Model Parental Educaon Effects on Men and Women´s Aainment? Cross-naonal Assessmentsof Different ApproachesI. Tomescu-Dubrow 1 , H. Domanski 11 Polish Academy of Sciences, PolandResearch in social straficaon shares the assumpon that social origin operates through assets embeddedin the family structure, yet scholars’ opinions of how resources get transmied intergeneraonally vary significantly.The result of this variaon in opinions is a range of measures for family background, and disnctempirical models. A simplified schema yields three main methodological approaches: (a) one parent’s characteriscsmodels; (b) models using characteriscs of both parents; and (c) models accounng for specificeffects of social origin depending on gender. In this paper we analyze how models of each type perform whenapplied to cross-naonal data from the European Social Survey (Round 3). We focus on the impact of parentaleducaon on children’s success, while controlling for parents’ social class posion...2.35 Pioneering Survey Translaon: Froner research in quesonnaire translaonand mullingual measurement instruments IITo be held on July 20, 2011 from: 14:00 to 15:30, in room 303.Coordinated by:• Dorothée Behr - GESIS - Leibniz Instute for the Social Sciences, Germany• Brita Dorer - GESIS - Leibniz Instute for the Social Sciences, Germany2.35.1 Quality Management in Translaon of Measurement Instruments: Lessons Learnt from Large-scaleSurveysB. Upsing 2 , H. ROELKE 3 , L. Wäyrynen 1 , S. Dept 11 cApStAn Linguisc Quality Control, Belgium; 2 TBA - Deutsches Instut fuer Internaonale PaedagogischeForschung, Germany; 3 TBA-DIPF, GermanyQuality Management in Translaon of Measurement Instruments: Lessons Learnt from Large-scale Surveys

WEDNESDAY 20 JULY 101Translang measurement instruments for the ”Programme for the Internaonal Assessment of Adult Competencies”(PIAAC), the ”Programme for Internaonal Student Assessment” (PISA) and other large-scale surveysevolved on experimental procedures as well as trial and error, and led to new developments in related technologyand methodology.Some of these developments are currently being used for the translaon of PISA 2012 material:2.35.2 Advance Translaon in the European Social Survey (ESS) Round 5, 2010B. Dorer 11 GESIS - Leibniz Instute for the Social Sciences, GermanyIn cross-cultural surveys, measurement errors may arise from quesonnaire draing or from translaon.2.35.3 Supporng the Translaon Process Online in a Cross-Naonal SurveyA. Amin 1 , A. Blom 2 , M. Martens 11 CentERdata, Netherlands; 2 Survex - Survey Methods Consulng, GermanyThe Survey of Health, Ageing and Rerement in Europe (SHARE) is a muldisciplinary and cross-naonal panelsurvey on health, socio-economic status and social and family networks. The panel consists of more than45,000 individuals aged 50 or over. Having fielded the baseline of this bi-annual panel study in 2004/05, SHAREis now in its fourth wave of data collecon.2.36 Open-ended survey quesons IITo be held on July 20, 2011 from: 14:00 to 15:30, in room 315.Coordinated by:• Nick Allum - University of Essex, United Kingdom• Graham Hughes - University of Surrey, United Kingdom2.36.1 Can open-ended quesons help to idenfy voter heterogeneity?J. Blumensel 11 MZES, University of Mannheim , GermanyWhile open-ended quesons had been used in seminal works such as Stokes et al. (1958), Campbell et al.(1960), or Converse (1964), they have been largely ignored in electoral science ever since the 1980s. Currently,related to the discussion of voter heterogeneity, open-ended quesons might experience a renaissance. Thenoon that voters apply different decision-making strategies when choosing among pares has become commonsense in electoral science. Some voters, for example, are assumed to vote in order to achieve a certainpolicy-outcome, others may but put a special emphasis on candidates, and sll another group may simply votehabitually. However, how to idenfy heterogeneous subgroups in the electorate is sll unseled. Bartle (2005)advocates that directly asking the voters about their consideraons might be the most promising approach...2.36.2 Three-Valued Logic as an Instrument for Rule-Based Coding of Open-Ended QuesonsG. Müller 11 Université de Fribourg, SwitzerlandIn principle, the coding of answers to an open ended survey queson can be considered as a classificaon taskwith a Boolean decision tree. Based on the presence or absence of certain facts X1, X2, ...., Xn in the answerof an interviewee, the person is given or withhold a certain abstract aribute Y, like e.g. being a liberal, or agood patriot, etc. The advantage of such tree-based inference rules is that they allow a certain rounizaonof the coding process, which in the best case may even be done by a computer-program.

102 WEDNESDAY 20 JULY2.36.3 How do Respondents Perceive a Quesonnaire? The Contribuon of Open-ended QuesonsE. Markou 1 , B. Garnier 11 Instut Naonal d’Etudes Démographiques, FranceWhen they fill out a quesonnaire, the respondents mobilize their experience on the theme of the surveybut they also take posion towards the quesonnaire. Understanding how a quesonnaire is perceived byrespondents is thus necessary in order to improve the survey methodology and to beer appreciate the collecteddata. Open-ended quesons are an appropriate way to have a beer image of the understanding, theimpact and the relevance of the quesonnaire.2.36.4 Quality of some open-ended survey quesonsD. Oberski 1 , W. van der Veld 2 , W. Saris 11 Universitat Pompeu Fabra, Spain; 2 Radboud Universiteit Nijmegen, NetherlandsOpen-ended survey quesons have the advantage that they might provide more informaon than closed ques-ons. For example, the respondent might give condions on an opinion, provide a movaon, or give moreprecise esmates of quanes than allowed by closed categories. In addion, closed categories might imposethe researcher’s frame of reference upon the respondent, thus affecng the answers (Krosnick& Schuman 1988; Krosnick& Fabrigar 1997).2.37 Design and Implementaon of Mulmode SurveysTo be held on July 20, 2011 from: 16:00 to 17:30, in room 321.Coordinated by: Colm O’Muircheartaigh - University of Chicago, United States2.37.1 Assessing mode effects: implicaons of embedding a children’s quality of life quesonnaire withinan Internet surveyK. Lloyd 11 Queen’s University Belfast, United KingdomThe aim of this paper is to examine whether there are mode effects associated with using a children’s qualityof life measure embedded within an internet survey, Kids’ Life and Times (KLT), when compared with its moretradional use as a paper quesonnaire. Parcipants were 3440 10 and 11 year old children in Northern Irelandwho completed the KIDSCREEN-27 online as part of a general atudinal survey. The quesonnaire wasanimated using cartoon characters that are familiar to this age group and the quesons appeared on screenand were read aloud by actors. Exploratory principal component analysis of the online version of the ques-onnaire was carried out to examine whether the five domains (physical wellbeing, psychological wellbeing,autonomy and parents, social support and peers and school environment) reported for the paper version ofthe KIDSCREEN-27 could be replicated...2.37.2 Mulple modes in cross-naonal surveys – a good mix?P. Marn 1 , R. Fitzgerald 11 City University London, United KingdomCross-naonal surveys that consider mixed mode data collecon face special problems. Because naonal ”surveyclimates” differ according to respondent expectaons, fieldwork experse, and mode penetraon, mostcross-naonal surveys cannot implement the same mixed mode design in all parcipang countries. Yet differencesin data collecon designs pose a threat to measurement equivalence, potenally confounding countrydifferences with errors introduced by country-specific survey pracces.

WEDNESDAY 20 JULY 1032.37.3 A survey on end-of-life in FranceS. Pennec 3 , A. Monnier 3 , R. Aubry 2 , C. Cases 1 , S. Pontone 3 , a. Stephan 31 Instut na, France; 2 Hopital de Besancon, France; 3 Instut Naonal d’Etudes Démographiques, FranceTwo main changes concerning death arise during the last decades. Firstly death affects old and very old peopledue to the decline of mortality and the ageing of the populaon. Secondly, death occurs nowadays mainlyin instuons while just aer WWII, they mainly took place at home. Besides, the legal context of medicaldecisions, rights of paents and end of life has changed in many countries (last change occurred in 2005 inFrance).2.37.4 Social Climate Survey of Tobacco Control: A mixed-mode approachR. McMillen 1 , J. Klein 11 American Academy of Pediatrics’ Richmond Center, United StatesThe annual Social Climate Survey of Tobacco Control ulizes random samples of the US populaon. From 2001-2008, annual samples were obtained using Random Digit Dialing methods. This study examines the ulity of amixed-mode approach, used in 2009-10, and examines the validity of overall and subpopulaon findings.TheRDD frame included US households with landline telephones. Knowledge Networks provided the internetpanel. Members were randomly recruited by telephone and mail surveys, and households were provided withaccess to the Internet and hardware if needed. Weights were computed in two steps. First, both frames wereweighted based upon 2009 US Census esmates. Second, adjustments to these inial weights were computedto account for the overlap in the two samples. Comparisons to the Census demonstrated that the KN panelprovided beer representaon of adults with low educaon and young adults...2.38 Macro Mechanisms and Macro Hypotheses ITo be held on July 20, 2011 from: 16:00 to 17:30, in room 413.Coordinated by:• Heiner Meulemann - University of Köln, Germany• Hans-Jürgen Hans-Jürgen Andreß - University of Cologne, Germany2.38.1 Late careers and social structures in EuropeH. Engelhardt 2 , C. Schmidt 11 University of Bamberg, Germany; 2 University of Bamberg, GermanyIn this paper we invesgate the effects of socio-demographic and socio-economic structure on labor parcipaonand transion to rerement for older males in eleven European countries. Theorecally, our analysisis guided by consideraons of intergeneraonal compeon and intergeneraonal substuon. FollowingEasterlin’s hypotheses that intrageneraonal compeon rise with cohort size, we assume a negave effect ofcohort size on labor parcipaon and a posive effect on early rerement. Taking into account that differentcohorts are substutes at least to a certain extent we assume that the probability for early rerement will bereduced by a high intergeneraonal exchange rao in favor of older workers. Thus, labor parcipaon is bothinfluenced by socio-demographic structures when entering the labor market and during the career...2.38.2 Why are religious people happy? A social norms explanaonO. Stavrova 2 , D. Fetchenhauer 1 , T. Schlösser 11 Department of economic and social psychology, University of Cologne, Germany; 2 GK SOCLIFE, University ofCologne, Germany

104 WEDNESDAY 20 JULYDrawing on social norms theories we suggest that religious beliefs and parcipaon are related to addionalpsychological benefits if they are considered as normave in a certain naonal context. In study 1, we test thishypothesis using the data of the European and World Values Surveys (2005-2009). The results of the mullevelregression analysis in 64 countries suggest that strong descripve and injuncve norms of religiosity at countrylevel increase the happiness of religious and decrease the happiness of non religious individuals. In study 2,we use a mullevel moderated mediaon analysis to examine whether the differences in social integraon ofreligious and non religious individuals in countries where religiosity is normave account for this finding. Ourresults indicate that the effect of religiosity on happiness is parally mediated by social integraon...2.38.3 Religion, Stage & Civil Society: A Mullevel Analysis of the European Values Survey (2008)R. Norrie 11 GK SOCLIFE, University of Cologne, GermanyThis paper aims to establish what it is about religiosity that maers in explaining the link between religion andcivil society, using a mullevel research design. At the micro level, it is tested whether or not belief and/orreligiously derived social capital maer for parcipaon. Simultaneously it is tested if one’s belonging to thedominant religion maers, as a revision to exisng work on Protestant exceponalism, which tends to ignorethe underlying religious context in relaon to the individual’s confession. Exisng studies have also formulatedat the macro level, hypotheses pertaining to the overall religious devoutness of countries and their religioustradions, but have failed to take into account the religious composion of the state. It is argued here thatseparaon of church and state will increase civil society parcipaon as if the state is closed to religion, thenit will manifest itself in civil society...2.38.4 Explicit and Implicit Hypotheses Linking Maro and Micro LevelsJ. Friedrichs 2 , A. Nonnenmacher 11 Leibniz University Hannover, Instute for Polical Science, Germany; 2 University of Cologne, Research Instutefor Sociology, GermanyStudies linking macro level condions to micro level outcomes, such as characteriscs of countries to individualdecisions or acons, make assumpons on the processes relang the two levels. We refer to theseprocesses as ”social mechanisms” (Hedstrom 2005, Mayntz 2005, Opp 2004) – basically a set of interrelatedhypotheses. Unfortunately, these social mechanisms are not always made explicit, and in some cases even aclear macro-micro-hypothesis is wanng. Our methodological study aims at exploring the social mechanismsfor one selected dependent variable on the micro level: life sasfacon. We will analyze publicaons relangthe naonal to the individual level (e.g., European Values Survey) by using a sample of 30 mullevel studieswith life sasfacon as dependent variable, published in internaonal journals between 2007 and 2010. Ouranalycal procedure comprises three steps...2.39 Wider applicaons of cognive interviewing methods and novel implementaonsTo be held on July 20, 2011 from: 16:00 to 17:30, in room 303.Coordinated by:• Michelle Gray - Naonal Centre for Social Research - NatCen, United Kingdom• Joanna D’Ardenne - Naonal Centre for Social Research - NatCen, United Kingdom• Margaret Blake - Naonal Centre for Social Research - NatCen, United Kingdom2.39.1 The use of cognive interviewing methods to evaluate mode effects in survey quesonsM. Gray 2 , M. Blake 2 , P. Campanelli 1 , S. Hope 21 The Survey Coach, United Kingdom; 2 Naonal Centre for Social Research - NatCen, United Kingdom

WEDNESDAY 20 JULY 105Derived from cognive psychology and adapted to the context of quesonnaire development, cognive interviewingmethods have been extensively used to pre-test survey quesons. Usually the researcher/quesonnairedesigner uses cognive interviewing methods in the quesonnaire development stage as a qualitave tool totest specific quesons by establishing whether respondents interpret the quesons consistently and in theway intended (comprehension). The method is also used to explore recall, judgment and response issues.2.39.2 Applicaon of Cognive Interviewing For Construct RedefinionJ. Remr 11 Charles University of Prague, Czech RepublicPaper presents how the technique of cognive interviewing could be ulized for finding proper indicators whenthe research instrument is transferred from one cultural environment to the other. It is always a difficult taskto transfer the theorecal model developed in one cultural context into different one. And if it is the case, thenspecial aenon has to be paid to real meaning of each indicator proposed to measure certain type of atude.Especially in cross-cultural survey there is a high demand for achieving the comparability of the findings whichis usually ascertained by unified format of the quesons. But is the percepon of such smuli same in differentcontexts?2.39.3 ”Say What?”: Tesng of Cognive Funconing Tests with Non-Nave English Speakers in the UKHLSN. Uhrig 1 , M. Gray 2 , S. McFall 1 , J. D’Ardenne 2 , M. Balarajan 21 University of Essex, United Kingdom; 2 Naonal Centre for Social Research - NatCen, United KingdomUnderstanding Society: the UK Household Longitudinal Study (UKHLS) is a large indefinite length householdpanel survey including an oversample of ethnic minories in the UK. Interviewing in the UKHLS is conductedin English and 8 other official languages, though some respondents are interviewed in English even thoughEnglish is not their first or nave language. At Wave 3, a baery of cognive ability assessments was includedin the annual interview. To address the concern that non-nave English speakers may be prejudiced in theirperformance on these tests, a series of cognive interviews were conducted to evaluate whether respondentunderstanding and likely performance could be influenced by English language ability. Among 43 respondents,approximately half came from households where English was not the dominant language...2.39.4 Rethinking Think Alouds: A move to a qualitave research methodologyJ. Branch-Mueller 11 University of Alberta, CanadaVerbal protocol analysis is a research methodology that has long been used as a way to examine the cogniveprocesses employed while solving a problem. The theorecal framework for verbal protocol analysis comesfrom Ericsson and Simon’s seminal work entled Protocol analysis: Verbal reports as data that was first publishedin 1984. Another key work in this area, Van Someren, Barnard, and Sandberg’s (1994) The think aloudmethod: A praccal guide to modeling cognive processes, introduced researchers to the Think Aloud Method.Both works were firmly set within the posivist paradigm.2.40 The Quality of Longitudinal Survey Data: Measurement Error & Arion ITo be held on July 20, 2011 from: 16:00 to 17:30, in room 415.Coordinated by:• Ulrich Krieger - SHARE MEA University of Mannheim, Germany• Peter Lugg - Utrecht University, Netherlands• Noah Uhrig - University of Essex, United Kingdom• Emanuela Sala - University of Milano Bicocca, Italy

106 WEDNESDAY 20 JULY2.40.1 Dynamic modelling of Nonresponse in Business SurveysC. Seiler 11 Ifo Instute for Economic Research, GermanyIt is well-known that nonresponse affects the results of surveys and can even cause biases due to selecvies ifit cannot be regarded as missing at random. In contrast to household surveys, response behaviour in businesssurveys has been examined rarely in the literature. This paper is one of the first which analyses a large businesssurvey on micro data level for unit nonresponse. The data base is the Ifo Business Tendency Survey, which wasestablished in 1949 and has more than 5,000 responding firms each month. The panel structure allows to usestascal modelling including me-varying effects to check for the existence of a panel fague. The resultsshow that there are huge differences in business characterics such as size or subsector and that nonresponseis more frequent in economically good mes.2.40.2 Panel arion in the Cologne High School Panel (CHiSP: 1969/70, 1985, 1996/97, 2010)K. Birkelbach 2 , A. Grauenhorst 3 , C. Reinelt 1 , A. Weber 11 Universität Duisburg-Essen, Germany; 2 University of Cologne, Research Instute for Sociology, Germany;3 Universität zu Köln, GermanyWe will examine panel arion in a long term panel of former German high-school students. The primaryfield research was carried out in 1969/70 (n=3240, age 15); in three follow-ups (1985: n=1987, age 30; 1997:n=1596, age 43; 2010: n= 1297, age 56) the respondents’ private and occupaonal life courses have beensurveyed retrospecvely together with quesons concerning biographical, polical and religious atudes.2.40.3 Non-response and Arion - Evidence and Experience from the Brish Wealth and Assets SurveyM. Aumeyr 1 , C. Blake 11 Office for Naonal Stascs, United KingdomThe Wealth and Assets Survey (WAS) is a new longitudinal household survey in Great Britain and the first andsingle source of informaon providing a detailed picture of household and individual wealth in Great Britainbased on assets and liabilies.Several methods are being used before, during and aer the field work aimingfor a high level of response rates over me. This paper discusses the advantages and disadvantages of thesemethods as well as the lessons learned from WAS. Despite the acons taken to keep response levels up, asubstanal amount of arion was experienced, which has lead to concerns about the quality of longitudinalesmates based on the remaining sample. The paper refers to a study aiming to invesgate the differencesbetween non-responding and responding households in the first follow-up wave...2.40.4 Panel Arion in a Study of the Unemployed in the Great RecessionC. Zukin 1 , C. Van Horn 1 , J. Godofsky 11 Rutgers University, United StatesIn August of 2009 the Heldrich Center for Workforce Development at Rutgers contracted with Knowledge Networksto use their on-line probability panel to conduct a naonal survey of 1200 unemployed U.S. adults whosaid they had been unemployed at some point in the preceding 12 months. Quesons centered on how longpeople had been out of work, their job searching strategies, their psychological state, their financial condionand coping mechanisms, among others.2.41 Measurement equivalence in cross-cultural survey research ITo be held on July 20, 2011 from: 16:00 to 17:30, in room 410.Coordinated by: Jouni Kuha - London School of Economics, United Kingdom

WEDNESDAY 20 JULY 1072.41.1 Measuring Authoritarianism with Different Sets of Items in a Longitudinal StudyT. Toharudin 1 , H. Oud 2 , J. Billiet 3 , H. Folmer 41 University of Padjadjaran, Indonesia; 2 Radboud Universiteit Nijmegen, Netherlands; 3 Katholieke UniversiteitLeuven, Belgium; 4 University of Groningen, NetherlandsIn the General Elecon Study in Belgium (Interuniversitair Steunpunt Polieke-Opinieonderzoek K.U. Leuven,1991, 1995, and 1999), authoritarianism was measured in each of the wave years 1991, 1995, 1999 by sevenitems. However, only two items were the same in all of the three waves. A series of congenericness tests asproposed by Jöreskog (1971, 1974) is performed to find out which of the items in each of the waves can beconsidered to measure the same latent variable and can therefore be used in a longitudinal study to assessthe development of authoritarianism over me. We find that three of the items in 1999 did not pass thecongenericness test convincingly.2.41.2 Sensivity of conclusions to incorrect assumpons about cross-naonal measurement equivalenceJ. Kuha 2 , I. Moustaki 1 , S. Stares 21 London School of Economics and Polical Science, United Kingdom; 2 London School of Economics, UnitedKingdomIn latent variable modelling of cross-naonal survey data, substanve interest usually focuses on the distribu-ons of the latent variables, for example on cross-naonal comparisons of averages or on models for associa-ons between the latent variables. In contrast, the measurement models for the observed indicator variablesare typically of lesser interest, and any complexies in these models a source more of nuisance than of excitement.2.41.3 Cross-Naonal and Cross-Ethnic Differences in Atudes. A Case of LuxemburgM. Kankaraš 1 , G. Moors 11 Tilburg University, NetherlandsUsing a case of Luxembourg a cross-cultural comparave perspecve is linked to between as well as withincountry comparisons by answering a two-folded queson. First we analyzed the level of measurement equivalence,i.e. the extent to which ethnic groups in Luxembourg and cizen of their countries of origin assignthe same meaning to atude quesons. Secondly, we examined whether ethnic-cultural groups within Luxembourgresemble cizens from their nave country more than Luxembourger’s atudes, i.e. we comparedthe relave influence of a given naonal context and cultural background of Luxembourg’s minories on theiratudes. We selected three scales from the European Value Study, wave 2008/2009 to demonstrate differenttypes of result from such analyses...2.41.4 When are item intercept differences substanal in measurement equivalence tesng?B. Meuleman 11 University of Leuven, BelgiumApplied comparave researchers are becoming increasingly aware of the issue of measurement equivalence.By now, there exists considerable agreement on the concrete operaonalizaon and implicaons of (the variouslevels of) measurement equivalence. Mulple group confirmatory factor analysis (MGCFA) has becomewidely recognized as a useful stascal tool to test for equivalence. In this framework, measurement equivalenceis assessed by constraining certain parameters – e.g. factor loadings or item intercepts - across groups.2.42 Effects of Interviewers in Sample Surveys ITo be held on July 20, 2011 from: 16:00 to 17:30, in room 319.

108 WEDNESDAY 20 JULYCoordinated by:• Gabriele Durrant - University of Southampton, United Kingdom• Ian Brunton-Smith - University of Surrey, United Kingdom2.42.1 Using paradata to explain the interviewer contribuon to survey design effectG. Turner 11 University of Southampton, United KingdomThe interviewer is a well recognized source of survey error. In this paper I focus on the interviewer contribuonto measurement error and, more precisely, on the variability of survey esmates that is introduced by the interviewer– the interviewer effect. Face-to-face interview surveys generally employ a clustered sample design,in which geographical clusters are first selected and then individuals or households are selected within clusters. This design can lead to inflaon of the variance of survey esmates, relave to a simple random sample, dueto the greater similarity between respondents in the same cluster than is evident in the populaon as a whole.This phenomenon is referred to as the design effect. Because, usually, there is only one interviewer workingin each geographical cluster it is difficult to separate the design effect due to areas from that which is causedby interviewers...2.42.2 Is success on the doorstep correlated with the magnitude of interviewer variance?I. Brunton-Smith 1 , P. Sturgis 21 University of Surrey, United Kingdom; 2 University of Southampton, United KingdomEvidence is now beginning to accumulate which shows that interviewer atudes, personality, and behaviourare predicve of success in achieving contact and cooperaon on the doorstep. A less frequently acknowledgedpossibility, however, is that these same characteriscs might also be predicve of the extent to whichinterviewers follow best pracce in the implementaon of standardized interviewing. That is to say, there maybe a correlaon between interviewer-induced nonresponse bias and measurement error. This raises the intriguingpossibility that, although recruing certain types of interviewers and training them to adopt parcularstrategies on the doorstep might serve to reduce nonresponse bias through raising response rates, total surveyerror might be unaffected (or even increased) due to an increase in interviewer variance...2.42.3 Movated underreporng in screening interviewsF. Kreuter 2 , S. Eckman 1 , R. Tourangeau 21 Instute for Employment Research, Germany; 2 University of Maryland, United StatesMany naonal surveys include screening interviews intended to idenfy members of the eligible populaonor members of rare subgroups slated for oversampling. Underreporng of members of these groups drives upsurvey costs and may introduce bias into the esmates. There is evidence that members of the target populaonsare somemes underreported in screening interviews. Interesngly members of these same groupsare oen well covered in surveys that do not parcularly screen for them. One of the best documented instancesof such a screening shorall occurred in the Naonal Longitudinal Survey of Youth, 1997 Cohort, witha coverage rao of only about 70 percent of the targeted age group. All other age groups have rates above 90percent. Such screening shoralls could reflect respondent movaon to screen out rather than refuse (andbe subject to conversion aempts)...2.42.4 Measuring Interviewer Effects across Countries and SurveysA. Blom 11 Survex - Survey Methods Consulng, Germany

WEDNESDAY 20 JULY 109Interviewer effects are found across all types of interviewer-mediated surveys crossing disciplines and countries.While studies describing interviewer effects are manifold, idenfying characteriscs explaining theseeffects has proven difficult. This paper proposes a conceptual framework of interviewer characteriscs for explaininginterviewer effects. The framework encompasses four dimensions of interviewer characteriscs: generalinterviewer atudes, interviewers’ own behavior, interviewers’ experience with measurements, and interviewers’expectaons. An interviewer quesonnaire based on this conceptual framework was implementedon fourth wave of SHARE Germany in early 2011. Exploratory analysis of the interviewer data collected showassociaons between the response and consent rates that interviewers expect to achieve and the other dimensionsof the characteriscs collected...2.43 A Vital Use of Sample Surveys and Censuses to Inform Operang Data SystemTo be held on July 20, 2011 from: 16:00 to 17:30, in room 340.Coordinated by:• Young Chun - NORC at the University of Chicago, United States• Fritz Scheuren - NORC at the University of Chicago, United States2.43.1 Combining registers into a fully register-based census - some methodological issuesI. Jansson 1 , D. Hedlin 1 , A. Holmberg 11 Stascs Sweden, SwedenStascs Sweden faces the challenge of conducng Sweden’s first fully register-based census. Several registers,for example the exisng Populaon register, the Real property register and the new Register of dwellings willbe matched to allow us to esmate distribuons of variables. There are number of methodological issuesinvolved, such as stascal matching, disclosure control and evaluaon of model assumpons.Missing data pose a parcular problem. There will be individuals with no recorded dwelling in populaonregister, as well as dwellings with no residents according to the register of dwellings. Methods for matchingindividuals and dwellings will be discussed.2.43.2 Use of Administrave Sources for Censuses and Demographic and Social Stascs – Merits andChallengesL. Thygesen 11 Stascs Denmark, DenmarkDemographic and social stascs in Denmark, including censuses, have been mostly based on administraveregisters since 1981. Surveys based on interviews or quesonnaires are important supplementary sources infields where suitable register data can not be obtained. The stascs based on surveys and registers are closelyconnected in one coherent system which is briefly described in this paper. The philosophy behind the system,as well as its merits, problems and challenges, are discussed.2.43.3 Integrang surveys and administrave data within a common framework - a systems approach tostascs produconB. Sundgren 11 Stockholm University, SwedenIn countries like Sweden, up to 99% of the data used for official stascs emanate from administrave datasources, and only 1% from tradional surveys. According to a Dutch study, it is roughly 100 mes more expensiveto collect data by a tradional census than by using administrave data.

110 WEDNESDAY 20 JULY2.43.4 Comparing Administrave Data to the 2010 U.S. CensusL. Marshall 11 U.S. Census Bureau, United StatesThe U.S. Census Bureau has acquired and evaluated administrave records data since 1999. Previous studieshave compared administrave records data to the U.S. Census – reviewing populaon counts and distribuonsof demographic characteriscs. This paper discusses plans for the U.S. Census Bureau’s ”2010 Census MatchStudy” which intends to fully analyze the coverage of administrave records data compared to the 2010 U.S.Census. Topics to be discussed include acquision plans for new administrave records data, development ofimproved record linkage procedures, and plans to use the match results to compare the individuals and theaddresses in each database. While populaon counts, housing counts, and even demographic distribuonsmay be similar between administrave records data and the 2010 U.S. Census, the specific individuals andaddresses may differ. This paper discusses plans to evaluate those differences...2.44 Internet panels for the general populaon: Fact or ficon IITo be held on July 20, 2011 from: 16:00 to 17:30, in room 412.Coordinated by:• Edith De Leeuw - University of Utrecht, Netherlands• Annee Scherpenzeel - CentERdata, Netherlands2.44.1 Why do Internet panel members become inacve and how can they be re-acvated?A. Scherpenzeel 11 CentERdata, NetherlandsThe LISS panel (Longitudinal Internet Studies for the Social sciences) is an online panel which is based on a trueprobability sample of households. Households that cannot otherwise parcipate are provided with a computerand Internet connecon.2.44.2 Can Tailor-made Quesons Movate Respondents in an Internet Panel?M. Oudejans 11 CentERdata, NetherlandsAlmost no literature exists on the effects of the topic of the quesonnaires on panel parcipaon. In this paperI will explore how the addion of ”tailor-made quesons” at the end of exisng quesonnaires will affect theevaluaon of the quesonnaire. The experiment will be embedded in the Dutch CentERdata LISS panel; a panelof 5.000 households that complete surveys every month.2.44.3 Answer behavior in a probability based internet panelC. Vis 2 , A. Wijnant 11 CentERdata Tilburg University , Netherlands; 2 CentERdata, NetherlandsIn 2007 CentERdata (Tilburg University, the Netherlands) started the MESS project by seng up the LISS panel.The LISS panel is a probability based internet panel representave of the Dutch populaon, consisng of 5,000households. On a monthly basis, the panel members fill out quesonnaires for about 30 minutes.2.44.4 Design and Implementaon of an Online Weekly Survey to Study Unintended Pregnancies: PreliminaryResultsJ. Barber 1 , Y. Kusunoki 2 , H. Gatny 2

WEDNESDAY 20 JULY 1111 University of Michigan, United States; 2 Instute for Social Research, University of Michigan, United StatesWe are conducng new research to invesgate unintended pregnancies during the transion to adulthood.Invesgaon of these issues requires detailed, ming-specific measures of atudinal and behavioral aspectsof relaonships (including sexual behavior), contracepve use, acvies that compete with childbearing (includingschool and work), and community context. Our invesgaon also requires detailed, ming-specificmeasures of beliefs, expectaons, and willingness to engage in behaviors related to unintended pregnancy,such as sexual relaonships, contracepon, pregnancy, and parenthood. Available data resources for studyingunintended pregnancy suffer from fundamental weaknesses including retrospecve reporng error.2.45 Analysis of Immigraon in Europe IIITo be held on July 20, 2011 from: 16:00 to 17:30, in room 315.1.Coordinated by: Mariya Aleksynska - CEPII, France2.45.1 Does ethnic diversity make your work beer? A longitudinal approach. (with Simonea Longhi)L. Fumagalli 11 University of Essex, United KingdomEmpirical and theorecal studies find a relaonship between ethnic diversity and indicators of performancesuch as producvity and growth. Moreover, research suggests that diversity facilitates problem solving andincreases creavity, but erodes trust and discourages the provision of public goods. However, these findingsare generally based on cross seconal data or on very imprecise measures of cultural diversity, which makes itdifficult to isolate causality.2.45.2 Educaonal inequalies across ethnic minories at the end of compulsory educaonM. Fernández-Reino 11 Universitat Pompeu Fabra, SpainIn my paper I try to disentangle the mechanisms that lie behind the differences in educaonal performanceat the end of compulsory educaon among ethnic minories in England. For this purpose, I use the first threewaves of the Longitudinal Study of Young People in England. Using binary logisc regressions, I invesgatethe variaons in achievement of 5 A*-C GCSE (including English and Maths) in the final examinaons, whichusually determines entering in the more academic track. As some studies have previously shown, there areconsiderable differences in achievement across ethnic groups in comparison to the White-Brish. While Indiansare on average outperforming autochthonous students, a higher proporon of Pakistani, Bangladeshi andBlack-Caribbean students are failing compared to White-Brish...2.45.3 Ethnic Identy and Labor Market Integraon: Evidence from DenmarkC. Gorinas 11 Århus University - The Danish Naonal Centre for Social Research, DenmarkEthnic identy – the degree of commitment of immigrants towards the country of origin and the host country– together with the economic integraon of minories are now central in the public debate. This paperexplores both identy formaon and the link between identy and employment for ethnic groups, exploingsurvey data from Denmark. Based on the two-dimensional acculturaon framework of Berry (1997), identyis modeled through both direct and composite measures successively following the methodologies of Nekbyand Rödin (2007), Bau and Zenou (2010) and Constant and Zimmermann (2008). First, I find that idenficaonwith the majority is posively associated with post-migraon characteriscs such as the pracce ofDanish at home and educaon gained in Denmark, and to some extent it is influenced by ethnic and religiousbackgrounds...

112 WEDNESDAY 20 JULY2.45.4 Sampling special populaons: Why is it important to ask immigrants about their year of arrival?C. Ródenas 1 , M. Mar 11 Universidad de Alicante, SpainThe objecve of this study is to show that obtaining demographic and social informaon regarding the immigrantpopulaon through sampling techniques requires a specifically designed survey. In order to guaranteethe reliability of the results, the sample design should not be limited to using those demographic variables thatare commonly used, but should also contemplate the variable referring to the ”year of arrival” of the immigrantsto the country. If this variable is not considered in the inial design or subsequent adjustments of thesurvey, there is a high risk that its esmate will be biased and consequently, that of all those variables that arecorrelated with the year of arrival.2.46 Socio-economic variables in cross-naonal surveys: Quality enhancement inmeasurement, documentaon and data disseminaon IIITo be held on July 20, 2011 from: 16:00 to 17:30, in room 318.Coordinated by:• Jürgen H.P. Hoffmeyer-Zlotnik - GESIS - Leibniz Instute for the Social Sciences, Germany• Hilde Orten - Norwegian Social Science Data Services (NSD), Norway• Uwe Warner - Centre d’Etudes de Populaons, de Pauvreté et de Poliques Socio-Economiques (CEPS /INSTEAD), Luxembourg2.46.1 The New Educaon Measures in the European Social Survey - Advantages in Using DDI for DocumentaonH. Orten 1 , H. Midtsæter 1 , J. Wackerow 21 Norwegian Social Science Data Services (NSD), Norway; 2 GESIS - Leibniz Instute for the Social Sciences, GermanyThe European Social Survey is a biennial mul-country survey covering more than 30 countries.For its fih round, currently in field, the survey will carry out a major revision of the measurement of respondent’s,partner’s, father’s and mother’s highest level of educaon.The ESS will change its current targetharmonised measure to a more detailed ISCED coding.There will also be an addional variable named ES-ISCED that will be mapped from the new detailed ISCEDmeasure.The process of construcng the new measure has involved conceptual consideraons, set-up of bridging specificaonsfrom source to target variables as well as quesonnaire items for each parcipang country.2.46.2 The ”new” survey queson for the income measurement in the European Social Survey and theimprovements for cross-country comparisonU. Warner 1 , J. Hoffmeyer-Zlotnik 21 Centre d’Etudes de Populaons, de Pauvreté et de Poliques Socio-Economiques (CEPS / INSTEAD), Luxembourg;2 GESIS - Leibniz Instute for the Social Sciences, GermanyComparing the total net household income across European countries, the household definion and size, therespondent’s family relaon to the main income earner, the main source of income, the income composion,the respondent’s capability to remember income element, and the naonal income distribuons have obviousand visible impacts on the income informaon given by the respondent.

WEDNESDAY 20 JULY 1132.46.3 How much confidence can we have in Europe’s poverty figures? EU-SILC and the standard error ofthe Europe 2020 poverty indicatorsT. Goedemé 11 Herman Deleeck Centre for Social Policy, University of Antwerp, BelgiumThe European Union Stascs on Income and Living Condions (EU-SILC) are one of the most important datasources for cross-naonal comparave research on income and living condions in the European Union. AsEU-SILC consists of a sample of households in all EU member states, point esmates should be accompaniedby appropriate standard errors and confidence intervals. This is especially so if indicators are constructed formeasuring progress towards pre-defined targets such as those of the Europe 2020 poverty reducon strategy.All too oen this has been neglected in European poverty research and official publicaons. In contrast, thispaper pays explicit aenon to the calculaon of standard errors and confidence intervals. Unfortunately,informaon on the sample design is incomplete in the EU-SILC User Database and adequate documentaon islacking...2.46.4 Harmonising applicaon form informaon in consumer credit: a case of Belgium, Germany and theNetherlandsG. Andreeva 11 University of Edinburgh, United KingdomIn consumer credit, risk assessment relies on the applicaon form informaon supplied by prospecve borrowers.The applicaon form includes socio-economic variables (marital status, occupaon, etc.) and is similar toa survey. By linking the applicaon characteriscs to the re-payment behaviour of previous borrowers, creditscoring esmates creditworthiness of new applicants. Tradionally applicaon forms are country- and productspecific.Yet the European economic integraon makes it possible for large internaonal lenders to assess riskof their naonal and regional porolios with a single credit scoring model, which is cheaper and operaonallymore efficient.2.47 Open-ended survey quesons IIITo be held on July 20, 2011 from: 16:00 to 17:30, in room 315.Coordinated by:• Nick Allum - University of Essex, United Kingdom• Graham Hughes - University of Surrey, United Kingdom2.47.1 Open-ended quesons in the context of temporary work researchK. Siponen 11 School of Social Sciences and Humanies, University of Tampere, FinlandTemporary work is characterized by heterogeneity which oen refers to temporary employees’ different mo-ves for accepng temporary employment. Previous research suggests that these moves may be an importantelement in explaining differences in temporary employees’ well-being. In many studies, moves arepredominantly considered as a dichotomy between voluntary and involuntary temporary work, e.g. by askingwhether the employee would prefer an open-ended contract or not. However, offering dichotomized responsecategories may easily lead to too narrow view on moves.2.47.2 A Comparison of On-site and Office Coding of Occupaons: The Case of the Chinese Family PanelStudyL. Ren 1 , L. Li 1 , Q. Xu 1 , J. Yan 1 , Z. Qiu 11 Peking University, China

114 WEDNESDAY 20 JULYThe importance of occupaons in social science research and the widely-known error-prone coding process ofoccupaons in survey research are a dilemma for social sciensts and a challenge for survey data managers.The Chinese Family Panel Study (CFPS), a Computer-Aided Personal Interviewing (CAPI) survey, employed twomanual approaches to occupaonal coding: interviewers carrying out coding during the interviewing; andexperienced coders performing coding within the survey organizaon aer data collecon. This study concernsthe coding quality of these two methods and the factors that have impacts on the quality. It first describes thecoding processes of the two methods and compares their advantages and disadvantages. Then it checks theconsistency of their results and tries to clarify the inconsistencies with the help of supplemental informaonfrom the audio records of interviewing...2.47.3 Quality of responses to an open-ended queson on a mixed-mode surveyJ. Gibson 1 , F. Vakalia 1 , S. Turner 11 Fors Marsh Group LLC, United StatesWeb surveys pose a number of advantages over paper or telephone surveys, including speed, cost, and easeof data entry (Couper, 2000; Ferrara& Nolan, 1974; Kiesler& Sproull, 1986). However, research regarding survey mode differences highlights a number of potenal challengessuch as sample representaveness and data quality (Couper, 2000; Kiesler& Sproull, 1986). Likewise, open-ended quesons (OEQs) offer both advantages and disadvantages (Poncheri,Lindberg, Thompson,& Surface, 2008). They provide an opportunity to gather rich data that might otherwise be lost if responseopons do not adequately capture the relevant aspects of the quesonnaire item; however, decisions regardingresponse coding and manipulaon are crucial to determining what researchers glean from the data. Thisstudy applies human coding to an OEQ on a mixed-mode (paper and electronic) quality of work life survey ofmilitary recruiters (N = 3,997)...2.47.4 Quanfying Open-Ended Responses: Results from an Online Adversing Tracking SurveyA. Jacobe 2 , L. Brewer 2 , F. Vakalia 2 , S. Turner 2 , S. Marsh 11 Fors Marsh Group, United States; 2 Fors Marsh Group LLC, United StatesOpen-ended quesons are known to yield useful informaon, especially when researchers need to explorecomplex issues that do not have a finite or pre-determined set of responses.This paper will highlight the ulity and reliability of using open-ended quesons in surveys measuring adver-sement recall by detailing the process in which a coding methodology was established and applied to analyzedata from an adversement tracking study.2.47.5 The Impact of Open-Ended Quesons: A Mulvariate Study of Respondent EngagementS. Gielman 11 Mktg, Inc., United StatesIn survey research those that are less engaged are less likely to parcipate in the survey process and those whodo but fail to provide quality responses drive measurement error.In this analysis the metadata from 1100 diverse studies emanang from a great variety of online sources aswell as covering an abundance of sources was examined to determine the driving factors of respondent engagement.There has been considerable discussion as to the influence of varying structural designs within a quesonnairethat might improve respondent engagement. Surely, instruments replete with grids that drag on over fortyminutes beg the quesons as to the quality of responses that we are elicing.

WEDNESDAY 20 JULY 1152.48 Quality of Life/well-beingTo be held on July 20, 2011 from: 16:00 to 17:30, in room 414.Coordinated by: Marieke Voorpostel - FORS - Swiss Foundaon for Research in Social Sciences, Switzerland2.48.1 Money, sociability and happiness: are developed countries doomed to social erosion and unhappiness?F. Sarracino 11 Centre d’Etudes de Populaons, de Pauvreté et de Poliques Socio-Economiques (CEPS / INSTEAD), LuxembourgDiscovering whether social capital endowments in modern sociees have been subjected or not to a processof gradual erosion is one of the most debated topics in recent economic literature. Inaugurated by Putnam’spioneering studies, the debate on social capital trends has been recently revived by Stevenson and Wolfers(2008) contending Easterlin’s assessment. Present work is aimed at finding evidence for the relaonship betweenchanges in social capital and subjecve well-being in eight European countries and in Japan between1980 and 2005...2.48.2 Self Assessed Subjecve Wellbeing as Conflict Indicator in Transional SocieesY. Deshmukh 11 World Associaon for Public Opinion Research, United StatesCan a self assessed subjecve indicator on wellbeing be used as Conflict Indicator in Transional Sociees byusing surveys to understand a state of mind wherein the feeling of well being and happiness is on the top of themind recall? There are several globally accepted QOL indicators like health, housing, social security, educaonand many more. But, the meaning of QOL varies when we compare a economically stable and democracallyevolved society to a conflict and disaster hit transional sociees. Several QOL indicators have been formulatedfor ”Normal” sociees but hardly any for ”Transional” ones. This study is aimed at analyzing Quality of Lifedata as Conflict Indicator in transional sociees of Sri Lanka and Indonesia aer the Tsunami disaster and itsimpact on the ongoing armed conflict in these countries...2.48.3 The use of the me as a dimension of the quality of life: an applicaons. ceccherini 11 università firenze, ItalyMany studies (the Global Project presented by OCSE; GDP and Beyond presented by the Europeean Commission;the IX Congress ISQOLS carry out in Firenze 2009)show the importance of me use as a dimension ofthe individual quality of life. This is reasserted also by the Commission on measurement the economic performanceand social progress chaired by Joseph E. Sglitz, Amartya Sen, Jean-Paul Fitoussi.2.48.4 Homemaking and women’s well-being in Europe. Effect of divorce risk, selecon and dominanggender-role atudes.M. Mikucka 11 Centre d’Etudes de Populaons, de Pauvreté et de Poliques Socio-Economiques (CEPS / INSTEAD), LuxembourgWhereas it is known that employment status strongly impacts individual well-being, the literature concerningthe effect of homemaking is so far inconclusive. The paper invesgates the effect of being a housewife onwell-being of women, using European Values Study data for 36 European countries (year 2008) and mullevelregression methodology.

116 WEDNESDAY 20 JULYResults show that the impact of homemaking on well-being is overall posive and it varies across countries.Three possible explanaons of this variaon are tested.

117Day 3Thursday 21 July3.1 Use of web surveys in government data colleconTo be held on July 21, 2011 from: 09:00 to 10:30, in room 315.1.Coordinated by:• Ma Jans - US Census Bureau, United States• Jennifer Romano Bergstrom - US Census Bureau, United States• Sue Ellen Hansen - University of Michigan, United States3.1.1 What We Know About Our Online Respondents…And How We Plan to Use That KnowledgeC. Goschall 11 Naonal Agricultural Stascs Service, United StatesThe 2007 United States Census of Agriculture was the first to offer online reporng and nearly 100,000 Censusof Agriculture reports were received online. We examined traits of web respondents and quality of their data.Web respondents were smaller, newer, more urban farms with high speed internet access. Their data werealso beer quality. Thus, several experiments were designed to target and increase web response during thetesng for the 2012 Census of Agriculture. Comparisons were made between: groups 1) mailed a cover leerand instrucons for online reporng, without a paper form; and 2) mailed a cover leer requesng them toreport online but also included a paper form. In addion, several different cover leer treatment groups weretested. Response rates (both online and overall) and plans for increasing online reporng for the 2012 Censusof Agriculture will be discussed...3.1.2 Web Surveys: Re-(Search) for an Appropriate DesignK. Blanke 11 Federal Stascal Office, GermanyAbstractGerman official stascs covers more than 380 stascs, of which 170 surveys based on quesonnaires. Onlya few are related to social stascs, the majority of surveys deals with informaon on businesses, instuons,or public administraons.3.1.3 Observaons about Respondent Behaviors and Survey Data from a Web Survey Commissioned bythe U.S. Food and Drug AdministraonJ. Lin 11 U.S. Food and Drug Administraon, United StatesThe U.S. Food and Drug Administraon (FDA) commissioned a Web survey about food labeling with two independentsamples: an online consumer panel and shopping mall visitors. The purpose of the survey was to

118 THURSDAY 21 JULYcompare and contrast respondent behaviors and survey data between the two sample sources and to help usassess potenal advantages and disadvantages of replacing online consumer panels for shopping mall visitorsin data collecons aimed at quanfying causal relaonships rather than at developing populaon esmates.The 15-minute survey was conducted in 2006 concurrently with 765 online parcipants and 600 mall parcipants,using the same mode of collecon and the same quesonnaire. Parcipants were randomly assignedto view different food labels and to answer quesons about the label and the food they viewed...3.1.4 Large Companies, Mulple Surveys: The Promise and Challenge of Web (Survey) 2.0A. Tule 11 US Census Bureau, United StatesLarge companies tend to receive mulple different surveys, since they comprise a greater proporon of economicacvity in their respecve sectors or industries. The U.S. Census Bureau is conducng a series of meetingswith respondents at large companies to beer understand their survey response processes, with the aimof developing tools to assist them and reduce the burden of parcipang in mulple surveys. One conceptualtool which has been well-received is a secure online ”Company Portal,” which would provide a nexus for centralizedsurvey management by corporate-level respondents, leng them review the surveys they will receivein a given year for planning purposes, update contact and mailing informaon, and monitor survey compleonstatus to ensure mely responses from informants distributed throughout their companies...3.2 Exploring innovave methods for nonresponse bias assessment and adjustmentusing paradataTo be held on July 21, 2011 from: 09:00 to 10:30, in room 315.Coordinated by:• Hideko Matsuo - Katholieke Universiteit Leuven, Belgium• Jaak Billiet - Katholieke Universiteit Leuven, Belgium3.2.1 Using Neighborhood Characteriscs to Predict Nonresponse: Results from a Naonal Mail Survey ofYoung AdultsL. Viera Jr. 1 , S. Turner 2 , S. Marsh 1 , R. Medway 31 Fors Marsh Group, United States; 2 Fors Marsh Group LLC, United States; 3 Joint Program in Survey Methodology,University of Maryland, United StatesResearchers in the industry are quickly coming to the realizaon that nonresponse bias in esmates is onlyindirectly related to nonresponse rates. The general statement that ”the lower your response rate, the morevulnerable your study is to nonresponse bias” has been proven a gross oversimplificaon. Because even veryhigh response rates have the potenal for very high bias on specific esmates within a survey, recent calls formore nonresponse bias studies have been made (Groves, 2006). However, conducng such analyses in manylarge, naonal surveys is oen problemac when limited informaon about nonrespondents is available on thesurvey frame. Therefore, researchers have sought to find new ways of conducng nonresponse bias analysesby appending auxiliary data to their samples...3.2.2 Assessment of propensity score methods on nonresponse bias adjustmentA. Alanya 2 , H. Matsuo 1 , J. Billiet 11 Katholieke Universiteit Leuven, Belgium; 2 Gallup Europe, BelgiumUnder the project of ’Joint Research Acvies (JRA2): improving representaveness ofthe samples’, one of the goal of this project is to study non-response bias using fourdifferent approaches for detecon and adjustment of nonresponse bias used in previous

THURSDAY 21 JULY 119rounds of ESSi: (1) post-straficaon; (2) analysis of reluctant respondents; (3). usingobservable data; and (4) esmang bias and adjustment using surveys amongnonrespondents. These four approaches on detecon and adjustment of non-response biasare used independently from each other and each approach has its own advantages anddisadvantages. The last approach applied propensity score straficaon method (10classes) on the basis of the informaon of co-operave respondents and non-respondents.3.2.3 Exploring use of informaon in paradata through calibraon method to detect and adjust nonresponseH. Matsuo 1 , J. Billiet 11 Katholieke Universiteit Leuven, BelgiumUnder the JRA2 ESSi programme, four approaches on detecon and adjustment of non-response bias are studiedusing different types of informaon such as populaon stascs, reluctant respondents, observable dataand non-response survey. Taking the perspecve of previous approaches, our focus of this paper is to exploresuitable informaon/variable for nonresponse bias detecon and adjustment. Based on the conceptual frameworkof survey cooperaon from Groves and Couper (1998), this paper firstly idenfies the type of informaonrecorded in the paradata, such as dwelling and neighborhood as well as contact procedure variables in rela-on to survey cooperaon, which is suitable for non-response detecon. Different types of respondents arecompared based on these paradata variables...3.2.4 The potenal of call record data to study nonresponse bias in the french survey on sexual behaviourN. Razafindratsima 1 , E. Morand 1 , S. Legleye 21 Instut Naonal d’Etudes Démographiques, France; 2 INED, FranceComputer assisted telephone interview (CATI) sowares used in telephone surveys oen provide, along withthe survey datafiles, a record of the historic of all contact aempts (such as date and status of the aempt:refusal, appointment, invalid number, acceptance, etc.). This informaon is nevertheless rarely used, partlydue to methodological issues concerning their stascal treatment.3.3 Randomized response methodology for sensive topicsTo be held on July 21, 2011 from: 09:00 to 10:30, in room 318.Coordinated by: Pier Francesco Perri - University of Calabria, Department of Economics and Stascs, Italy3.3.1 Making Use of ”Benford’s Law” for the Randomized Response TechniqueA. Diekmann 11 ETH, SwitzerlandIn my presentaon ”Benford’s law” is applied to the ”randomized response technique” (RRT) to increase thevalidity of answers to sensive quesons. Using the Newcomb-Benford distribuon as a randomizing devicehas several advantages. It is easy to explain and follow the procedure as no physical device such as a coin ora dice is necessary and the method guarantees full anonymity. As is well known, the price for the anonymityof the RRT is a decrease in the efficiency of the esmator. However, because of the subjecve overesmaonof certain numbers (”Benford illusion”), the conflict between the variance of the esmates and the degree ofanonymity is less pronounced compared to other RRT methods. Making use of Benford’s law improves theefficiency of the esmator for any given level of anonymity.

120 THURSDAY 21 JULY3.3.2 Distribuon funcon esmaon by using randomized response proceduresL. Barabesi 1 , G. Diana 3 , P. Perri 21 University of Siena, Italy; 2 University of Calabria, Department of Economics and Stascs, Italy; 3 Universityof Padua, ItalyMuch of the literature on survey sampling focuses on the esmaon of populaon parameters of a targetvariable, say Y. A typical parameter of interest is the populaon mean (or total). Nonetheless, final usersof survey data are oen interested also in the distribuon funcon F(y). The esmaon of the populaondistribuon funcon has received considerable aenon in a non-sensive quesons framework when dataare directly observed by means of convenonal techniques.3.3.3 Mulstage randomized response strategies for all probability sampling designsA. Quatember 11 Johannes Kepler University Linz, AustriaIn the talk a generalizaon of mulstage randomized response quesoning designs for the esmaon of proporonsapplicable for all probability sampling methods will be discussed under the aspect of efficiency andprivacy protecon: This strategy starts at stage 1 of h stages by asking a survey unit the queson ”Are you amember of group UA?” with probability p11. With the remaining probability 1-p11 the unit is directed to a secondstage, where the same queson is asked with probability p21. With the remaining probability 1-p21 theelement is directed to another stage and so on. Finally, at the h-th stage of the strategy the unit is asked withprobability ph1 the queson on membership of group UA. With probability ph2 the respondent is asked thequeson ”Are you a member of the complementary group?”. With ph3 the survey unit is asked the queson”Are you a member of group UB?” with UB being a completely non-sensive group...3.3.4 Improved Randomized Response Models Using Three Decks of CardsS. Abdelfatah 2 , R. Mazloum 2 , S. Singh 11 Texas A&M University - Kingsville, United States; 2 Faculty of Economics and Polical Science, Cairo University, EgyptOdumade and Singh (2009) have extended the Warner (1965) randomized response model by the use of twodecks of cards. In a trial to obtain more efficient esmator of the populaon proporon possessing a sensivecharacterisc, a new randomized response model based on the use of three decks of cards is proposed as anextension to the Odumade and Singh (2009) model. The condion under which the proposed model is moreefficient than the Odumade and Singh (2009) model has been obtained. An empirical study has also beenperformed to examine the relave efficiency of the proposed esmator with respect to the Warner (1965),Mangat and Singh (1990), Mangat (1994) and Odumade and Singh (2009) esmators. The proposed modelcan be easily adjusted to be more efficient than the previously menoned models but the problem of low levelof cooperaon from the respondents can be accompanied with such model...3.4 Collecon and analysis of social media data for survey researchTo be held on July 21, 2011 from: 09:00 to 10:30, in room 340.Coordinated by: Craig A. Hill - RTI Internaonal, United States3.4.1 Current state of social media use for survey researchC. Hill 1 , E. Dean 11 RTI Internaonal, United StatesAs the use of social media (Facebook, Twier, etc) rises, the opportunies to use these plaorms and the dataresident on these plaorms to supplant or supplement ”tradional” survey data also increases.

THURSDAY 21 JULY 121How are researchers in the survey community using these data now, and how will be they be using them inthe near future? We will present a review of the current landscape of social media data use and offer thoughtsabout near-future uses and potenal lines of research to pursue along these lines.3.4.2 Errors within web-based surveys: a comparison between two different tools for the analysis oftourist desnaons websites qualityP. Gabriella 1 , A. Oliveri 11 Università degli Studi di Palermo, ItalyThe Internet has recently proved to be a very large place where social phenomena can be surveyed and anenvironment where powerful data collecon tool can be used.However, research reports have paid lile aenon to the errors affecng different techniques used withinweb-based surveys.This paper discusses such issue by considering both sampling and non-sampling errors.To reach this aim, the first part presents a preliminary review on the most popular web-based data collecontechniques, so as to idenfy the nature and the features of the errors affecng each technique.3.5 Informed pursuit of high response rates ITo be held on July 21, 2011 from: 09:00 to 10:30, in room 414.Coordinated by:• Ineke Stoop - The Netherlands Instute for Social Research/SCP, Netherlands• Achim Koch - GESIS - Leibniz Instute for the Social Sciences, Germany• Andy Peytchev - RTI Internaonal, United States3.5.1 Experiences in Implemenng Responsive Collecon Design for CATI SurveysF. Laflamme 1 , F. Bilocq 11 Stascs Canada, CanadaWith the recent emphasis on the analysis of collecon process data, paradata research has been focussed ona beer understanding of the data collecon processes to idenfy strategic opportunies to improve surveyoperaons. Research findings have oen indicated that the same data collecon approach does not workeffecvely throughout a data collecon cycle, stressing the need to develop a more flexible and efficient datacollecon strategy. Over the last two years, the relaonship between quality, cost, producvity and respondingpotenal of outstanding cases over the course of collecon has been invesgated. Addional tools have alsobeen developed to beer assess and monitor progress, quality and performance during collecon to allow thedevelopment and implementaon of Responsive Collecon Design (RCD) strategy for CATI surveys...3.5.2 Maximising response using a responsive design strategy on the Brish Naonal Survey of SexualAtudes and Lifestyles 3 (Natsal 3)B. Erens 1 , A. Phelps 2 , S. Nicholson 2 , J. Sinibaldi 31 London School of Hygiene and Tropical Medicine and University College London, United Kingdom; 2 NaonalCentre for Social Research - NatCen, United Kingdom; 3 Instute for Employment Research (IAB), GermanyNatsal 1 and Natsal 2, undertaken in 1990 and 2000, are the largest surveys in the world on sexual behaviour.Natsal 3 will provide updated measures of sexual behaviour by interviewing 15,000 respondents aged 16-74 in Britain, using a strafied, clustered probability sample design. Fieldwork takes place in 8 waves fromSeptember 2010 to August 2012.

122 THURSDAY 21 JULY3.5.3 Is it what you say, or how you say it? Exploring the effects of email messages for online panel surveysZ. Fazekas 1 , M. T. Wall 2 , A. Krouwel 21 Department of Methods in the Social Sciences, University of Vienna, Austria; 2 Kieskompas, NetherlandsVariaon in levels of survey parcipaon and in the quality of individual responses, as well as the problem ofarion in panel studies are ongoing core concerns for all survey researchers. Various reward and incenvestructures are employed in order to maximise response rates and response quality. One component of thisstructure is the content and tone of covering leers or emails that ask respondents to parcipate (or, in somecases to connue to parcipate) in surveys. However, there is lile evidence of a systemac approach to tesngthe effects of formulaon, tone, and content of the messages that accompany survey requests on responsepropensity and response quality in the exisng literature. We intend to fill this gap using a quasi-experimentalresearch design in which 14,000 respondents to an online survey undertaken by Kieskompas...3.5.4 Using respondent incenves in the German General Social Survey 2010: Are they an effecve meansto achieve a more balanced sample?M. Blohm 1 , A. Koch 11 GESIS - Leibniz Instute for the Social Sciences, GermanyLike many other surveys the German General Social Survey (ALLBUS) has been facing an increase in nonresponsein the past years. Between 1994 and 2008, e.g., the response rate has decreased from 54 % in 1994 to40 % in 2008. Against this background, in ALLBUS 2010 an experiment was mounted to invesgate whether ornot the provision of a condional respondent incenve (0 EUR, 10 EUR, 20 EUR) might help to stop this trend.3.6 Reporng cognive interview studies using the Cognive Interviewing ReportingFramework (CIRF): praccal examplesTo be held on July 21, 2011 from: 09:00 to 10:30, in room 303.Coordinated by:• Hennie R. Boeije - University of Utrecht, Netherlands• Gordon Willis - Naonal Instutes of Health (NIH), United States3.6.1 Using the CIRF to Report on a Dutch Pre-Test of a European Health Survey QuesonnaireV. Meertens 1 , R. Vis 11 Stascs Netherlands (CBS), NetherlandsUsing the CIRF to Report on a Dutch Pre-Test of a European Health Survey QuesonnaireVivian Meertens and Rachel Vis-VisschersDivision of Methodology and Quality, Stascs NetherlandsAbstract for ESRA 2011Session: Reporng cognive interview studies using the Cognive Interviewing Reporng Framework (CIRF):praccal examples.3.6.2 Using the CIRF to report on a mixed-method validaon study of a psychological scaleJ. Padilla 1 , I. Benítez 1 , M. Casllo 11 University of Granada, SpainThere is a growing interest in performing cognive interviewing to get validity evidence of psychological scales.Psychologists have tradionally conducted validaon studies using psychometrics. Therefore, they may havedifficulty reporng cognive interviewing findings and combining them with quantave results. The aim of

THURSDAY 21 JULY 123the paper is to present the lessons learned using the CIRF to report a mixed-method validaon study of a”family funcon” scale. The scale consists of 5 items in a rang scale format with 3 response opons. 21cognive interviews were performed during the cognive pre-test of the ”family funcon” scale. Moreover,28371 people responded to the Spanish version of the APGAR scale included in a naonal health survey. Theiranswers were used to analyse basic psychometrics which were computed comparing groups defined by the”type of home” variable...3.6.3 The Cognive Interviewing Reporng Framework (CIRF): Incorporang the Principles of QualitaveResearchH. Boeije 1 , G. Willis 21 University of Utrecht, Netherlands; 2 Naonal Instutes of Health (NIH), United StatesThe crux of this presentaon is that cognive interviewing as a way of pretesng quesonnaires is in essencea qualitave method. Although different techniques can be used in cognive interviews, such as think aloudand verbal probing, many of these are equivalent to those frequently used in the qualitave research field generally.Of course, cognive interviewing involves some unique elements. We begin with a comparison of theassumpons, purposes and current pracces in qualitave research with the pracce of cognive interviewing,to point out what they have in common and in what regards they differ.3.6.4 How Should Cognive Interviewing Reports be Wrien and Made Use of?G. Willis 2 , H. Boeije 11 University of Utrecht, Netherlands; 2 Naonal Instutes of Health (NIH), United StatesThe Cognive Interviewing Reporng Framework (CIRF) proposes to (a) increase coverage of the importantelements contained in Cognive interview reports; and (b) enhance the uniformity of presentaon in thesereports. These developments are expected to benefit the field of survey pretesng, by: (a) providing a frameworkthat leads researchers to consider how the pretesng work should be conducted in the first place, andto perhaps improve their pracces; (b) forcing report-writers to think through all the elements that shouldbe considered when wring reports; and (c) leading the survey pretesng field toward the producon of reportsthat are comprehensive enough, and similar enough, to be made use of beyond the original authors ororganizaon...3.7 Occupaon and Educaon variables for Comparave ResearchersTo be held on July 21, 2011 from: 09:00 to 10:30, in room 321.Coordinated by: Eric Harrison - City University London, United Kingdom3.7.1 The use of cross-naonal survey data to operaonalise social classE. Harrison 11 City University London, United KingdomThe accurate measurement of occupaon is valuable not only in its own right, but also because it is one of thebasic pieces of informaon needed to derive many widely-used class schemas. In the context of cross-naonalresearch this also requires a high degree of harmonisaon across countries. Despite considerable efforts inthis direcon, users accept that a combinaon of instuonal differences and human error lead to less thanopmal validity. Thus however conceptually rigorous the class measure, it is likely to be operaonalised usingsurvey data that is frequently imperfect and always incomplete.3.7.2 Drop-out rates during compleon of an occupaon search tree in web-surveysK. Tijdens 1

124 THURSDAY 21 JULY1 Erasmus University, NetherlandsWeb-surveys are disadvantageous for self-idenficaon of occupaon because unidenfiable and too aggregateresponses can’t be corrected during survey compleon. Therefore, stascal agencies judge the measurementof the occupaon variable in web-surveys too risky. Two soluons are discussed: online text stringmatching and a search tree, both using an occupaonal database. The laer is commonly used by onlinejobsites, the former not well developed today.3.7.3 The standardizaon of background variables related to occupaon in the ISSPJ. Edlund 1 , I. Bechert 21 Department of Sociology, University of Umea, Sweden; 2 GESIS - Leibniz Instute for the Social Sciences, GermanyThe Internaonal Social Survey Program (ISSP) is one of the most used data bases among scholars involved incross-naonal research. The program originally started out as a collaborave item exchange project, focusingmostly on the measurement equivalence of atudinal items. Less effort was put on creang standardizedmeasurements of key sociological background variables, such as those related to the labor market, e.g., occupaon,sector, and labor market status. With the growth of the ISSP – at present more than 40 countriesare involved – the less than opmal standardizaon of background variables created increasing problems forscholars. Some years ago, the ISSP set up a task force, the Demographic Methods Group (DMG) addressingthese problems. In 2009, the ISSP implemented the work by the DMG and started changing the measurementof background variables in order to improve standardizaon...3.7.4 Occupaon as a key survey variable: concepts, measurement, coding methods and accuracyP. Elias 11 Insiitute for Employment Research, University of Warwick, United KingdomThis presentaon reviews progress that has been made over the last 20 years towards improving the quality ofinformaon on occupaons as collected in censuses and surveys. The author draws on his experience in developingthree UK naonal occupaonal classificaons (SOC90, SOC2000 and SOC2010), together with the workhe has performed for the Internaonal Labour Office and Eurostat to help develop a harmonised classiifciaonof occupataons for use within the European Research Area. The presentaon will cover conceptual issues,procedures for the collecon of occupaonal informaon, coding problems, the use of soware for coding. Itconcludes with an assessment of the current quality of occupaonal informaon in a variety of survey sources.3.8 Internet panels for the general populaon: Fact or ficon IIITo be held on July 21, 2011 from: 09:00 to 10:30, in room 412.Coordinated by:• Edith De Leeuw - University of Utrecht, Netherlands• Annee Scherpenzeel - CentERdata, Netherlands3.8.1 Are reluctant panelists worth the effort? Analysis of measurement error in a probability-based Internetpanel.C. Roberts 3 , P. Sturgis 1 , N. Allum 21 University of Southampton, United Kingdom; 2 University of Essex, United Kingdom; 3 University of Lausanne,SwitzerlandInternet panel surveys based on probability samples of the populaon present new and excing opportuni-es for gathering social science data. Yet despite their considerable appeal, parcularly with respect to measurementquality (for example, the potenal to reduce bias on socially sensive quesons, the avoidance of

THURSDAY 21 JULY 125interviewer effects, and the possibility to exploit all the benefits of computer-assisted quesonnaire administraon),threats to data quality persist in the form of nonresponse and arion, and the possibility that overme, the responses of even the most loyal panelists may exhibit effects associated with learning or decliningmovaon...3.8.2 Quality of a probability-based online panel for the social sciences in Germany: first results.B. Struminskaya 2 , I. Schaurer 1 , W. Bandilla 2 , L. Kaczmirek 21 GESIS, Germany; 2 GESIS - Leibniz Instute for the Social Sciences, GermanyIt has been acknowledged, that only probability-based online surveys can claim their results to be generalizableto the target populaon. In Germany a probability-based online access panel is organized by GESIS.Respondents are recruited by telephone (landline and mobile).Such a recruitment strategy poses a problem in terms of non-response as there are several decisions to bemade by potenal panelists. They have to take part in the telephone interview, agree to become a member ofthe panel and fill in the online quesonnaires. This mulstep selecon is a potenal threat to the quality andgeneralizability of data collected online.3.8.3 A Comparison Web Study to Examine Panel Reuse Effects in an Adversing Tracking StudyF. Vakalia 2 , S. Turner 2 , L. Brewer 2 , A. Jacobe 2 , S. Marsh 11 Fors Marsh Group, United States; 2 Fors Marsh Group LLC, United StatesAlthough probability-based online panels provide many advantages such as quality sample, higher responserates and potenal for longitudinal design surveys, this paper discusses the limitaons and drawbacks to theconnued use of a probability based web panel. One limitaon that hinders sample quality is when panelrespondents become condioned to taking surveys, and may start to demonstrate bad survey taking behaviors,such as non-response and sasficing.3.8.4 Applying maximum entropy weighng to on line panel data colleconA. Bianchi 1 , S. Biffignandi 1 , E. Hartmann 2 , J. Sekhon 21 University of Bergamo, Italy; 2 University of Berkley, United StatesA generalizaon of maximum entropy weighng (Maxent) is studied to allow for soluons when the data arenot full rank and, more importantly, to find the best soluon possible when the populaon moments cannotbe perfectly matched or when a perfect match would involve a large degree of extrapolaon. Maxent generalizaonis used to reweight a sample to match observed populaon moments. It is noted that by freeing theunit weights to vary smoothly across units greatly increases the space of potenally good matching soluonsand maximazes the use of available informaon. Propensity score weighng can be combined with MaxEntWhen the weights are highly variables, alternave methods might be used, such as PSM or a doubly robustmethod—i.e., first weight and then esmate a model on the data.3.9 The Quality of Longitudinal Survey Data: Measurement Error & Arion IITo be held on July 21, 2011 from: 09:00 to 10:30, in room 415.Coordinated by:• Ulrich Krieger - SHARE MEA University of Mannheim, Germany• Peter Lugg - Utrecht University, Netherlands• Emanuela Sala - University of Milano Bicocca, Italy• Noah Uhrig - University of Essex, United Kingdom

126 THURSDAY 21 JULY3.9.1 Panel Arion Arising from a Switch of Fieldwork Agencies in the IAB Household Panel Study PASSG. Mueller 11 Instute for Employment Research, GermanyOngoing longitudinal surveys seldom switch fieldwork agencies between waves. The possible effects of sucha change are difficult to esmate in advance and have, unl now, rarely been explored. In addion, dataprivacy statements hamper the switch between fieldwork agencies in pracce. Against this backdrop, thisarcle documents the change of fieldwork agency in the IAB Household Panel Study ”Labour Market and SocialSecurity” (PASS). As in most other academic surveys within the Federal Republic of Germany, the householdsto be surveyed in PASS were ensured in advance in wring (”Datenschutzbla”), among other things, that theirname and address data would not be passed on to third pares...3.9.2 Survey Burden, arion and alteri parcipaon in the pairfam panelU. Krieger 11 SHARE MEA University of Mannheim, GermanyThe German family panel study is an annually conducted panel survey of individuals. In addion to the mainrespondents and their partners, since wave two their parents and their children are also approached for aninterview. The aim of this design is to draw an accurate picture of the family situaon of the survey respondent.The main respondent has a gatekeeper funcon in providing access to the alteri respondents connectedto them by facilitang or declining informaon and or consent to interviews on their partners, parents andchildren.3.9.3 Re-engaging with survey non-respondents: Evidence from three household panelsN. Watson 1 , M. Wooden 11 University of Melbourne, AustraliaMost analyses of non-response in longitudinal surveys are concerned with the decisions respondents make toconnue parcipaon. Yet a poron of the fieldwork effort is spent in re-engaging non-respondents with thesurvey, resulng in non-monotonic arion. This suggests that analyses of the decision to connue parcipaonneeds to be complemented with analyses of the decision taken to recommence parcipaon. Are thedesign features and fieldwork pracces that we adopt to encourage connued parcipaon also appropriateto encourage non-respondents to be re-interviewed at a later wave?3.9.4 ”I sll don’t know”: Non-substanve responses in longitudinal dataR. Young 1 , D. Johnson 11 The Pennsylvania State University, United StatesThe presence of item non-response in longitudinal data is typically regarded as a nuisance that may bias stas-cal inference. Researchers have rarely considered the possibility that some types of non-substanve responsein fact may offer informave missing data in a longitudinal framework. In this paper we use two waves of theNaonal Survey of Midlife in the United States (n = 7,108) to explore whether ”Don’t know” (DK) responsepropensity is consistent among individuals over me. The stability and predicve power of DK responses fromone survey wave to the next may have important implicaons for survey design features, methods for reducingunit and item non-response, and predicng future panel arion. Uncorrelated DK responses over me mayindicate contextual survey factors rather than reflecng a stable personal trait...3.10 Macro Mechanisms and Macro Hypotheses IITo be held on July 21, 2011 from: 09:00 to 10:30, in room 413.Coordinated by:

THURSDAY 21 JULY 127• Hans-Jürgen Hans-Jürgen Andreß - University of Cologne, Germany• Heiner Meulemann - University of Köln, Germany3.10.1 Status Aainment Paerns in European SocieesP. Tufis 11 The Research Instute for Quality of Life, Romanian Academy, RomaniaThe paper proposes a systemac comparison of status aainment models across a wide array of Europeansociees with different types of socio-polical systems. The objecve is to esmate the degree of homogeneityin status aainment paerns within regions of countries classified according to several criteria: geographical,cultural, and linguisc, labor markets and polical systems, and the welfare state. Four broad regions areemployed: Connental European, social-democrac, Mediterranean, and Central and Eastern European (CEE)countries. The research also aempts to give an answer to the queson regarding the influence of sociopolicalarrangements on status aainment paerns and discusses differences in status aainment modelsbetween former communist countries from CEE and Western ’seled’ capitalist countries...3.10.2 Employment Protecon Legislaon: Contrasng Effects on Individual Unemployment RisksJ. Benne 11 GK SOCLIFE, University of Cologne, GermanyIndividual unemployment risks are primarily determined by the level of educaon a person has acquired. Thehighly-skilled are remarkably less oen unemployed in comparison to the low-skilled.However, the extent of such an imbalance is affected by instuonal factors. Hereby, the paper builds on thesll ongoing discussion of labour market flexibility. The focus is on the strictness of employment proteconlegislaon (EPL), which has parcularly been cricized as employment barrier for disadvantaged groups in thelabour market.3.10.3 The Experience of Social Mobility and the Formaon of Atudes Toward Income RedistribuonA. Schmidt 11 GK SOCLIFE, University of Cologne, GermanyThe study aims to explain the genesis of atudes toward redistribuon by applying the theory of causal attribuonto the phenomenon of subjecve social mobility. The paper asks two quesons: (1.) Are atudestoward income redistribuon affected by the subjecve experience of social mobility, and, (2.) how are theseeffects moderated by cultural contexts? These quesons refer to a potenal long-term feedback process betweena welfare state’s success in providing equal opportunies and individual atudes toward welfare stateacons. The hypotheses are tested with a mullevel design based on 21 countries and three me-points usinginternaonal survey data from the ISSP (Internaonal Social Survey Programme). The results suggest that theexperience of downward mobility leads to an increase in the support for redistribuon, while the experienceof upward mobility leads to a decrease in support for redistribuon...3.11 Measurement equivalence in cross-cultural survey research IITo be held on July 21, 2011 from: 09:00 to 10:30, in room 410.Coordinated by: Jouni Kuha - London School of Economics, United Kingdom3.11.1 Using indicators of stascal fit to explore cross-naonal paerns in latent class modelsS. Stares 1 , N. Mejlgaard 21 London School of Economics, United Kingdom; 2 University of Aarhus, Denmark

128 THURSDAY 21 JULYLatent variable models of cross-naonal survey data oen fit beer in some countries and worse in others,suggesng that the interpretaon of the measurement model is more substanvely plausible in some thanin others. We use marginal residual fit stascs to invesgate whether a 32-country latent class analysis canbe improved by idenfying those groups of countries that are most similar and most different in terms ofinterpretaon, breaking a single joint model for the total sample into a few more sasfactory models for smallergroups of countries. We use it to inform our substanve understanding of public parcipaon in science andtechnology across Europe.3.11.2 Examining European atudes towards the governance of biotechnologies: Who trusts whom?N. Allum 3 , P. Stoneman 1 , G. Revuelta 4 , P. Rudolf 21 University of Southampton, United Kingdom; 2 London School of Economics and Polical Science, United Kingdom;3 University of Essex, United Kingdom; 4 Universitat Pompeu Fabra, SpainIt is a well-established finding that cizens who approve of new or controversial technologies tend to trustthose seen as responsible for their development. This trust-approval nexus encompasses sciensts and theirinstuons, governments, regulators and the industries involved in bringing products to the market. In thispaper we examine the social-structural factors and value-orientaons that predispose Europeans to trust ordistrust these actors with respect to two controversial technologies – animal cloning and synthec biology.The former is relavely familiar to many people in 2010 but the laer has barely registered with any but themost informed strata of the European public. We capitalise on this contrast to examine the variables that act toproduce default posions of trust and distrust amongst Europeans for unfamiliar and familiar technologies...3.11.3 Worry about Crime among European Cizens: A Latent Class Analysis of Cross-Naonal DataJ. Jackson 1 , J. Kuha 11 London School of Economics, United KingdomFear of crime troubles cizens and policy-makers across the industrialised and industrialising world, yet comparaveanalyses are surprisingly rare. Drawing upon data from the European Social Survey, we assess thedimensionality of four categorical variables that together address the incidence and impact of worry aboutcrime. A model-assisted method based on latent class modelling produces a categorical scale with six levelsthat is comparable across countries. We esmate naonal levels of worry about crime in 23 European countriesand we compare the findings with those produced by the standard single indicator (percepons of thesafety of streets aer dark). We finish with a call for comparave work on fear of crime that goes beyonddescripon to address the complex psychological (micro) and sociological (macro) mechanisms underpinningthis far-reaching social and polical phenomenon...3.12 Effects of Interviewers in Sample Surveys IITo be held on July 21, 2011 from: 09:00 to 10:30, in room 319.Coordinated by:• Gabriele Durrant - University of Southampton, United Kingdom• Ian Brunton-Smith - University of Surrey, United Kingdom3.12.1 Interviewer Effects on Wave Nonresponse in Longitudinal SurveysG. Durrant 1 , R. Vassallo 1 , P. Smith 11 University of Southampton, United KingdomIt is widely recognized that interviewers play an important role in achieving contact and cooperaon from samplemembers in face-to-face surveys. So far, however, much research on interviewer effects on nonresponsehas been carried out for cross-seconal surveys. For longitudinal surveys, although some processes may be

THURSDAY 21 JULY 129similar to those in cross-seconal surveys, a number of complicang factors exist, such as that response needsto be achieved across several waves and that interviewers may change over me.3.12.2 Personality Traits and Interviewer Effects in Face-To-Face SurveysM. Kroh 11 German Instute for Economic Research (DIW) and Humboldt-Universität zu Berlin, GermanyThe influence of interviewer characteriscs on survey responses in face-to-face interviews is typically hard tostudy. On the one hand, the regional assignment of interviewers in cross-seconal surveys makes it difficultto disentangle regional from interviewer variaon in survey responses. On the other hand, the experimentalassignment of interviewers is oen limited to single characteriscs (e.g., gender, race, etc.). The present studyidenfies interviewer effects by longitudinal changes in interviewers in the German Socio-Economic Panel. Aninterviewer-survey that mimics the personal quesonnaire allows us to study the adjustment of respondents’answers to the opinions of changing interviewers in a variety of reported atudes and behaviours...3.12.3 Knocking on Respondents’ Doors - Unit Non-Response and Non-Contact in a Large Wealth SurveyJ. Le Blanc 2 , I. Noack 1 , T. Schmidt 21 Saarland University, Germany; 2 Deutsche Bundesbank, GermanyThis paper analyses the impact of interviewers and straficaon on unit non-response and non-contact in faceto-facesurveys. While most studies invesgate differences across geographical areas, we focus on differencesacross wealth strata within Germany. We also contribute to the exisng literature by studying straficaonand interviewer effects as well as interacons of both.The data for our analysis comes from the new German survey on Private Households and their Finances (PHF),which oversamples ”wealthy” areas in Germany. Making use of the special sampling design of the PHF, we analysedifferences between wealthy and other strata as contact and cooperaon behaviour across these groupsmay differ.3.12.4 Interviewers and data accuracy in the survey on household income and wealth (SHIW)A. Neri 1 , G. Ilardi 11 Banca d’Italia, ItalyThe paper invesgates the interviewer contribuon to data accuracy in the survey on household income andwealth conducted by the Bank of Italy (SHIW). In parcular, I focus on the interacon of the two major surveyerror components in the SHIW: unit nonresponse and measurement error. The main research queson iswhether interviewers who are good at recruing respondents also good at collecng data of good quality. Thisinformaon has implicaons for interviewer recruitment and training and it is of paramount importance forimproving data quality.3.12.5 Measurement of vaccinaon coverage based on vaccinaon cards in a health survey of childrenand adolescents – interviewer effects and alternavesC. Poethko-Mueller 1 , M. Schlaud 11 Robert Koch Instute, GermanyAimGiven the absence of a ’gold standard’ for measuring vaccinaon status in health surveys, on-site assessmentof vaccinaon cards by study physicians and algorithm-based analyses of complete vaccinaon data were compared.MethodThe naonally representave German Health Interview and Examinaon Survey for Children and Adolescents(KiGGS) of the Robert Koch Instute surveyed 17,641 children. Parcipants were invited to study centres,

130 THURSDAY 21 JULYwhere they filled in quesonnaires and had a computer assisted personal interview by a physician. Vaccina-on cards were presented and photocopied for later data entry. The frequency of missing values and interviewervariability were esmated for on-site assessment. Rates of false posive and false negave rangs werecalculated.Results3.13 Causal designs in non-experimental researchTo be held on July 21, 2011 from: 11:00 to 12:30, in room 321.Coordinated by:• Samuel Bendahan - University of Lausanne, Switzerland• Marius Brulhart - University of Lausanne, Switzerland• Lorenz Goee - University of Lausanne, Switzerland• Rafael Lalive - University of Lausanne, Switzerland• Philippe Jacquart - University of Lausanne, Switzerland3.13.1 Can We Do Well by Doing Good? A Reassessment of the Health Benefits of Volunteering UsingPropensity Score ModelsL. TAO 1 , L. JIN 11 The Chinese University of Hong Kong, ChinaIt has long been argued that altruisc behavior, such as volunteering, may benefit health through a variety ofpsychological and social pathways. A number of longitudinal studies have linked baseline volunteering acvi-es with posive health outcomes during the follow-up. However, individuals who volunteer can be drascallydifferent from those who do not, in ways that are highly consequenal to health outcomes. The problem ofendogeneity can be serious even with longitudinal study design and an extensive array of control variables.The main goal of this study is to reassess the presumed health benefits of volunteering using propensity scoremodels. We use the data from the Health and Rerement Study (HRS) in the USA, which is a longitudinal surveyspanning 15 years from 1993 to 2008. We first replicated the results from a published study by Luoh andHerzog on volunteering and health using the data from 1997 to 2000...3.13.2 Esmang Causal Effects Using Observaonal Design: Adjustment for Selecon Bias in ComplexLongitudinal Data on Early Maternal Employment EffectsY. Im 11 University of Chicago, United StatesThis paper arculates causal inference concepts and methods in the context of a non-experimental study ofme-varying treatments. In the presence of me-varying confounders, defined as covariates that are outcomesof prior treatments but also predictors of later treatment assignments, standard methods of adjustment canlead to bias—due to special problems of endogeneity. To cope with this problem, the current study appliesinverse-probability-of-treatment weighng (IPTW) to complex longitudinal data. Unlike standard stascalmethods, IPTW can appropriately adjust for confounding and selecon bias due to measured me-varyingcovariates affected by prior exposure. IPTW is a direct funcon of the propensity score, as the weight is completelydetermined by the fied propensity score model and is esmated parametrically...3.13.3 Propensity Score Methods for Causal Inference: The Relave Importance of Covariate Selecon,Reliable Measurement, and Choice of Analyc TechniqueP. Steiner 11 University of Wisconsin - Madison, United States

THURSDAY 21 JULY 131The popularity of propensity score methods for esmang causal treatment effects from observaonal studiesusing survey data strongly increased during the past decade. However, the success of these methods inremoving selecon bias mainly rests on strong assumpons and the competent implementaon of a specificpropensity score technique. Using a within-study comparison by Shadish, Clark& Steiner (2008) that compares treatment effect esmates from a non-equivalent control group design withself-selecon and a corresponding randomized experiment, the relave importance of four factors in removingselecon bias is assessed (Steiner et al...3.14 Mulple imputaon of complex data designs ITo be held on July 21, 2011 from: 11:00 to 12:30, in room 340.Coordinated by:• Jost Reinecke - University of Bielefeld , Germany• Marn Spiess - Hamburg University, Department of Psychology, Germany• Krisan Kleinke - University of Bielefeld , Germany• Roel de Jong - University of Hamburg, Faculty of Psychology, Germany3.14.1 Mulple Imputaon using nonparametric regressionR. de Jong 11 University of Hamburg, Faculty of Psychology, GermanyMulple imputaon by Fully Condional Specificaon (FCS) has been very popular among praconers in recentyears because of its simple concept and straighorward applicaon through readily available soware.Predicve Mean Matching (PMM) and Passive Imputaon (PI) are marginal imputaon methods typically implementedin FCS soware to handle non-linear relaonships between variables. We will show that with respectto marginal imputaon models, linear relaons are the excepon, rather than the rule. Moreover, weshow that both PMM and PI can lead to bias or low coverage rates in the inference based on mulply imputeddata. We propose a new method for impung connuous data where both the condional mean andvariance of the variable to be imputed is esmated using nonparametric regression. In conducted simulaonexperiments, the method has beer stascal properes than exisng alternaves...3.14.2 Mulple Imputaon of Incomplete Count DataK. Kleinke 1 , J. Reinecke 11 University of Bielefeld , GermanyOver the last couple of years mulple imputaon has become a popular and widely accepted technique tohandle missing data properly. Although various mulple imputaon procedures have been implemented inall major stascal packages, currently available soware is sll highly limited regarding the imputaon ofincomplete count data. As count data analysis typically makes it necessary to fit stascal models that aresuited for count data like Poisson or negave binomial models, also imputaon procedures should be speciallytailored to the stascal specialies of count data. We present flexible and easy to use soware to createmulple imputaons of incomplete ordinary, overdispersed and/or zero-inflated count data under a Poisson,quasi-Poisson or hurdle model respecively. Our procedures work as add-ons for the popular and powerfulMICE soware (van Buuren& Groothuis-Oudshoorn, 2010)...3.14.3 An Assessment of Missing Data of Confirmatory Factor Analysis with Categorical QuesonnairesL. Tsai 1 , C. Yang 2

132 THURSDAY 21 JULY1 Graduate School of Educaonal Measurement& Stascs, Naonal Taichung University, Taiwan; 2 Graduate School of Educaonal Measurement& Stascs Naonal Taichung University, TaiwanThis arcle aims to propose the Learning Vector Quanzaon(LVQ) approach to impute missing informaonand sampling weights in inferring the accuracy of populaon parameters of CFA models with categorical ques-onnaires. Survey data with missing background informaon, such as gender, age, and race, are very familiar.However, the informaon is crical for calculang the stratum sampling weight for each respondent.Asparouhov (2005), Tsai and Yang (2008), and Yang and Tsai (2006) have described that appropriate imputa-on can further improve the precision of esmaon as the importance of using weights in confirmatory factoranalysis (CFA) model. The quesonnaires with categorical responses are not well established yet. In the currentarcle, a simulaon study was conducted to compare the LVQ with Non-weighted, List-wise deleon, andweighng-adjustment class method...3.14.4 Combining Disnct DatasetsD. Bell 2 , A. Bowes 11 University of Srling, United Kingdom; 2 University of St, United KingdomThis presentaon examines methods of fusing survey datasets probabiliscally. This is part of the DAMES(Data Management through e-Social Science) project at the University of Srling. The presentaon focuseson the condions under which inference on relaonships embedded within the ”global” data structure can beretrieved. The problem is presented as an applicaon of mulple imputaon and is implemented through theimputaon algorithms available within Stata 11. As well as Monte Carlo results, an applicaon to social caredata is discussed.The presentaon also describes how this methodology is implemented within the DAMES framework.3.15 Comparave research ITo be held on July 21, 2011 from: 11:00 to 12:30, in room 319.Coordinated by:• Willem E. Saris - Universitat Pompeu Fabra, Spain• Paolo Moncagaa - Universitat Pompeu Fabra, Spain3.15.1 Evaluaon of the quality of quesons across countries using Split ballot MTMM experimentsW. Saris 1 , M. Revilla 11 Universitat Pompeu Fabra, SpainIn the ESS the Split Ballot Mul-Trait Mul-Method (SB-MTMM) approach has been used in each round toevaluate the quality of some of the quesons. In the classical MTMM design each respondent has to answerthree different forms of the same queson for at least three different concepts. Because the repeon ofthree alternave forms for the same queson may lead to problems of memory effects or irritaon, Saris,Satorra and Coenders (2005) developed the SB-MTMM approach. In this approach each respondent gets onlytwice an alternave form of the same queson. The quality parameters of these quesons can be esmatednevertheless because two or more random samples of the populaon are used where one group gets forms 1and 2 and the other group forms 1 and 3...3.15.2 Cross-Naonal Survey Methods in Migraon Research: Lessons From the IS Academy Survey inBurundi and EthiopiaM. Siegel 1 , S. Fransen 1 , K. Kuschminder 11 Maastricht University, Netherlands

THURSDAY 21 JULY 133This paper deals with issues facing researchers when working across countries, specifically when working in thearea of migraon research using the specific cases of survey implementaon in Burundi and Ethiopia. The mainareas where we focus our aenon is on how definions are used, language and translaon, cultural sensiviesand diversity, praccalies (including: finding the right survey partner and the roles and responsibiliesof each party as well as infrastructure challenges and soluons) and safety and security.3.15.3 Incomplete data handling in cross-naonal longitudinal studiesH. Oud 1 , M. Delsing 21 Radboud Universiteit Nijmegen, Netherlands; 2 Prakkon, Radboud University, Nijmegen, NetherlandsData collecon in big cross-naonal longitudinal studies ideally requires exactly the same instruments to beadministered at exactly the same me points in all parcipang countries. Because of language problems,poor coordinaon between countries and many other problems, this ideal seldom is realized in pracce. As aresult, because of large quanes of missing data, the main purpose of the cross-naonal study, comparabilitybetween naons, oen is not achieved.3.15.4 Equivalence of the le right scale across countriesW. Weber 1 , W. Saris 11 Universitat Pompeu Fabra, SpainOne of the most frequently used scales in survey research is the 11 points le right self placement scale. Thisscale has also been used in comparave research even though it is not at all clear if the scale is comparableamong individuals and across countries. In order to test the comparability mulple indicators are necessary.By means of equivalence tesng different types of le-right scales can be compared among individuals andacross countries. In Round 4 of the ESS alternave forms of this le right scale have been presented to therespondents in all parcipang countries. The results of the test of equivalence of these scales across countrieswill be presented.3.16 Natural Experiments in Survey ResearchTo be held on July 21, 2011 from: 11:00 to 12:30, in room 318.Coordinated by:• Henning Best - University of Mannheim, Germany• Gerrit Bauer - University of Mannheim, Germany3.16.1 Agree-Disagree Response Format versus Importance JudgmentD. Krebs 11 University of Giessen, GermanyAbstractCharacteriscs of response scales are important factors in guiding cognive processes underlying the choice ofa response category in responding to the request for an answer on an atude item. This paper deals with twodifferent requests and two different response formats as well, yet addressing the same topic. For example,assessing the importance of an item to respondents one can ask directly for importance judgments (on a scaleranging from very important to unimportant) or one can formulate an item containing already the statementof importance and then offer a response scale ranging from agree to disagree. In both cases responses expressrespondents’ subjecve importance of idencal items and therefore responses can be expected to be similaron both response formats.

134 THURSDAY 21 JULY3.16.2 Ethnic Discriminaon in the German Housing Market. A Field Experiment on the Underlying Mechanisms.A. Horr 1 , C. Hunkler 2 , C. Kroneberg 21 Mannheim Centre for European Social Research MZES, University of Mannheim, Germany; 2 University ofMannheim, GermanyWe study ethnic discriminaon in the rental housing market of a German metropolitan area using a telephoneaudit design; i.e., randomly assigned testers called landlords who had adversed apartments for rent. Whileearlier field experiments focused on assessing the degree of discriminaon by varying the ethnicity of testers,we also try to idenfy the mechanisms behind landlords’ behavior. To disnguish between informaon- andtaste-based discriminaon, we added two addional experimental variaons: (1) whether or not at the beginningof the call testers disclosed informaon on having a stable job, and (2) whether or not testers with anethnic minority (Turkish) name spoke with an accent. Using fixed-effects models to account for unobservedheterogeneity in apartments and landlords, we find no significant discriminaon against callers with Turkishnames in invitaons for apartment viewings...3.16.3 Causal Effects of Suppression Policy in a Market for Heroin: A Natural Quasi-Experiment with DiD-EsmaonR. Berger 2 , N. Braun 11 LMU Muenchen, Germany; 2 University of Leipzig, GermanyWe focus on the market effects at the retail level of a more severe suppression of hard drug supply. For theempirical examinaon we use data from a natural quasi-experiment. This was possible aer surveys had beenconducted in the visible drug scenes in the cies of Bern and Zurich during the summer of 1997 (n=624). Aerthese surveys, a remarkable change in police suppression towards middlemen and dealers of hard drugs, whowere not using drugs themselves, started in the city of Bern in January of 1998 (campaign ”Citro”). By repeangthe surveys in summer 1998 in both cies (n=419) we used Bern as test group and Zurich - where no changein police suppression had happened - as control group.3.16.4 Curbside Collecon and Parcipaon in Household Waste Recycling: A Causal AnalysisH. Best 1 , T. Kneip 2 , G. Bauer 11 University of Mannheim, Germany; 2 MEA, University of Mannheim, GermanyThis paper tests hypotheses on the causal effect of a reducon of behavioral cost on parcipaon in householdwaste recycling. We use data from a natural experiment in Cologne, Germany and employ propensityscore matching and condional fixed effects regression to idenfy the treatment effect. The empirical datawas collected in a panel survey in three city districts; in one of the districts, the tradional bring-scheme wasreplaced with a curbside scheme between the two panel waves. Using this data we can idenfy the causaleffect of curbside collecon, its variaon between types of recyclables, and its elascity with regard to thedistance to collecon containers in the bring-scheme condion. First results indicate that a curbside schemeis most efficient for plascs, metal cans and packaging and less efficient for paper...3.17 Construcon of Answer Scale Formats in Quesonnaires ITo be held on July 21, 2011 from: 11:00 to 12:30, in room 412.Coordinated by: Natalja Menold - GESIS - Leibniz Instute for the Social Sciences, Germany3.17.1 Experiments on the Design of the Le-Right Self-Assessment ScaleC. Zuell 1 , E. Scholz 1 , D. Behr 11 GESIS - Leibniz Instute for the Social Sciences, Germany

THURSDAY 21 JULY 135The le-right self-placement is one of the most frequently used measures for ideological self-idenficaonin empirical polical science research. However, the respondents’ understanding of ”le” and ”right” is onlyrarely tested.3.17.2 Comparisons of ”Frequency” and ”Intensity” Based Perceived Cognive Funcon Using Item ResponseTheory ModelsJ. Lai 1 , Z. Bu 1 , D. Cella 11 Northwestern University, United StatesBACKGROUND. Cognive dysfuncon is a common concern for children with various chronic illnesses. Rouneassessment of cognion enables prompt referral for comprehensive tesng or intervenon. Scales of childandparent-perceived cognive funcon (PCF) may be ideal methods for repeated assessment given their easeof administraon, low cost and relevance to paents’ daily lives. The purpose of this analysis was to comparesymptom frequency vs intensity rang scales using Item Response Theory (IRT) models.3.17.3 Cognive process in answering quesons: Are verbal labels in rang scales aended to?N. Menold 1 , L. Kaczmirek 1 , T. Lenzner 11 GESIS - Leibniz Instute for the Social Sciences, GermanyTwo different formats of labelling in rang scales are commonly used in quesonnaires: verbal labels for theend categories only (END form), and verbal labels for each of the categories (ALL form). On the one hand, thereare a lot of research findings which recommend the usage of the ALL form, but on the other hand, there is noclear evidence whether this only holds true for short answer scales (e.g. with up to five categories) or also forlonger answer scales (e.g. with seven categories).3.17.4 Middle alternaves revisited: How the neither/nor response acts as a ’face-saving’ way of saying’I don’t know’P. Sturgis 1 , C. Roberts 21 University of Southampton, United Kingdom; 2 University of Lausanne, SwitzerlandIn this paper, we use follow-up probes administered to respondents who inially select the mid-point in bi-polaratude rang scales to determine whether they selected this alternave in order to indicate opinion neutrality,or to indicate that they do not have an opinion on the issue. We find the vast majority of responses turnout to be ’don’t knows’ and that reallocang these responses from the mid-point to the don’t know categorysignificantly alters descripve inferences. Our findings have important implicaons for the design and analysisof bipolar rang scales.3.18 Challenges in Educaonal Surveys ITo be held on July 21, 2011 from: 11:00 to 12:30, in room 315.1.Coordinated by:• Hans Kiesl - University of Applied Sciences Regensburg, Germany• Alina Matei - Instute of Stascs, University of Neuchatel; Instute of Pedagogical Research and Documentao,Switzerland• Leyla Mohadjer - Westat, United States3.18.1 Cross-Naonal Surveys of Educaonal Achievement of Students – the Case of PISA (Programme forInternaonal Student Assessment)K. Rust 2 , R. Adams 1

136 THURSDAY 21 JULY1 Australian Council for Educaonal Research and University of Melbourne, Australia; 2 Westat and Universityof Maryland, United StatesOver the past 20 years cross-naonal comparisons of educaonal achievement of school students have beenincreasingly frequent, and the results are now well-publicized. These studies constute a unique blend ofsurvey sampling and psychometric methodologies. By taking advantage of the fact that individual student resultsare not required, but rather data on the distribuon of achievement across the student populaon, thesestudies are able to provide rich detail while only requiring limited student me and moderate size samples ofschools and students.3.18.2 Preparing for the First Cycle of the Programme of Internaonal Assessment of Adult Competency(PIAAC)L. Mohadjer 11 Westat, United StatesAn extensive amount of research has been conducted and results have shown that there is a strong relaonshipbetween adult competency levels and economic success and social advancement of both individuals andcountries. Thus it comes as no surprise that European governments are increasingly interested in understandingthe level and distribuon of competency levels in their populaons and how they compare with othercountries in Europe or around the world.3.18.3 The Naonal Educaonal Panel Study in Germany: Purpose, Design and ChallengesJ. von Maurice 2 , V. Stocké 1 , H. Blossfeld 21 University of Bamberg, Germany; 2 Naonal Educaonal Panel Study (NEPS), GermanyEducaon is a key factor in modern sociees for individual life chances as well as for economic growth. Thereforethere is a high demand for high-quality data and in parcular for panel data in the area of educaonalresearch. Consequently the Naonal Educaonal Panel Study (NEPS) has been established in Germany. Paneldata have methodological advantages and allows to describe educaonal trajectories over the live course andto test different theorecal approaches about educaonal success and competence development. The NEPSdisnguishes eight stages in the live course that are integrated by five theorecal perspecves...3.18.4 Two-stage indirect sampling of Kindergarten children in the Naonal Educaonal Panel Study (NEPS)S. Koch 11 Naonal Educaonal Panel Study (NEPS), GermanyIn general sampling is based on a complete list. Complete lists of informaon concerning all units of the targetpopulaon enable using sampling strategies and geng – depending on the quality of list and strategy– unbiased esmators for parameters of the target populaon. Without a sampling frame direct sampling isnot possible. In this situaon Lavallée (2007) suggests to use an indirect sampling approach. If there existsa populaon with an available sampling frame and this populaon is connected to the target populaon ina uniquely defined manner, indirect sampling allows provision of a random sample of the target populaonand of weights perming an unbiased esmaon of populaon parameters. The connecon between bothpopulaons can be described by an indicator variable...3.19 Naonal minories in (inter)naonal surveys: Bias, Precision and data qualityTo be held on July 21, 2011 from: 11:00 to 12:30, in room 413.Coordinated by:

THURSDAY 21 JULY 137• Francesco Laganà - University of Lausanne, Switzerland• Lavinia Gianeoni - Misc - University of Lausanne, Switzerland• Oliver Lipps - FORS - Swiss Foundaon for Research in Social Sciences, Switzerland3.19.1 What standard surveys can learn from immigrant surveysJ. Dollmann 1 , F. Kalter 2 , I. Kogan 21 Mannheim Centre for European Social Research MZES, University of Mannheim, Germany; 2 University ofMannheim, GermanyThe problems when interviewing immigrants in standard populaon surveys are well documented. Their contactand cooperaon rates are usually lower compared to their nave counterparts. Moreover, higher arionrates aggravate this problem, leading to overall lower response rates in panel studies. Preliminary results fromthe German part of the Children of Immigrants Longitudinal Survey in Four European Countries (CILS4EU) showthat both set of problems can be solved by using a tailored design approach. In this contribuon, we focus onthe parental sub survey of CILS4EU consisng of a self-compleon interview with a telephone follow up in casethe quesonnaire was not returned by the parents...3.19.2 Naonal minories and their representaon in Swiss surveys. Analysis of causes and remediesF. Laganà 2 , G. Elcheroth 2 , O. Lipps 11 FORS - Swiss Foundaon for Research in Social Sciences, Switzerland; 2 University of Lausanne, SwitzerlandThe paper presents the main results obtained from an interdisciplinary Swiss working group that analyses therepresentaon of naonal minories in social surveys. We address three research quesons: Are naonalminories, overall, correctly represented in Swiss surveys? Which minority communies, or social subgroupswithin them, are under-represented? How effecve are soluons proposed in the methodological literatureto correct for such minority bias?3.19.3 Is it Possible to Measure Race and Ethnicity in the U.S. Census With Acceptable Accuracy?D. Dillman 11 Washington State University, United StatesIs it Possible to Measure Race and Ethnicity in the U.S. Census With Acceptable Accuracy?Don A. Dillman3.20 Preferences and Well-being in Europe ITo be held on July 21, 2011 from: 11:00 to 12:30, in room 303.Coordinated by: Mariya Aleksynska - CEPII, France3.20.1 Is habituaon an illusion ? Using ”natural vignees” to disnguish the sasfacon and the hedonictreadmillsX. Fontaine 11 Paris school of Economics - EHESS, FranceHabituaon to circumstances is a central topic for sciensts interested in the determinants of well-being. Unfortunately,any aempt to evaluate the extent of this phenomenon (named ”hedonic treadmill” by Brickman& Campbell, 1971) using subjecve well-being data is subject to some important methodological cricisms.The main of them arises from the plausible existence of what Kahneman called the ”sasfacon treadmill”(Kahneman, 1999).

138 THURSDAY 21 JULY3.20.2 Unemployment and well-being in Europe. The effect of country unemployment rate, work ethicsand family es.M. Mikucka 11 Centre d’Etudes de Populaons, de Pauvreté et de Poliques Socio-Economiques (CEPS / INSTEAD), LuxembourgSubjecve well-being literature shows that higher unemployment rate corresponds to lower psychological costof own unemployment. The goal of the paper is to deepen the understanding of this regularity by invesgangthe role played by the work ethics and the strength of family es. I analyze the European Values Study data(2008) for 36 countries using mullevel regression methodology.3.21 Mullevel analysis in comparave research ITo be held on July 21, 2011 from: 11:00 to 12:30, in room 315.Coordinated by:• Bart Meuleman - University of Leuven, Belgium• Elmar Schlueter - University of Cologne, Research Instute for Sociology, Germany3.21.1 How many countries do you need to do mullevel modeling? A Monte Carlo Experiment comparingFrequenst and Bayesian approaches.D. Stegmueller 11 University of Mannheim, GermanyResearchers in comparave research increasingly use mullevel models to test effects of country level factorson individual behavior. An underlying assumpon of those models is asymptoc normality of countrylevel random effects. However, applicaons in comparave research rounely involve only a small number ofcountries, which has led some researchers to queson the ulity of mullevel modeling. However, just howmany countries constute too small a sample is unclear and different author give widely different ’rules ofthumb’. The key quesons are: How many countries do we need to achieve reliable inferences? Does switchingto a Bayesian approach, which does not rely on asymptoc jusficaons, lead to beer results? To answerthese, I present results from a Monte Carlo experiment, comparing esmaon strategies for mullevel modelsbased on Maximum Likelihood and Bayesian approaches...3.21.2 Inequality in student achievement: a mullevel means and variance regression modelH. van de Werorst 1 , R. Koçer 11 University of Amsterdam, NetherlandsThis paper presents a mullevel means and variance regression model that assesses to what extent individualand contextual variables predict the mean score on dependent variables and the deviance from the mean.The model takes account of heteroskedascity that results from deviances being related to predictor variablesat the individual or contextual level. We apply the model to the well-known trade-off hypothesis that statesthat a country’s level of between-school tracking has a posive effect on the mean student performance andmagnifies its dispersion. We use the PISA 2006 data on science achievement. Similar to most earlier studiesthat relied on aggregate data, our findings refuted the trade-off hypothesis. Between-school tracking magnifiesthe variance in performance, but does not affect its mean.3.21.3 Advantages and Disadvantages of Bayesian Mulnomial Mullevel Analysis in Comparave SurveyResearchJ. Mewes 11 University of Bremen, Germany

THURSDAY 21 JULY 139Sll, applicaons of mulnomial mullevel techniques are scarce to find in the survey research literature. Giventhe sheer numbers of discrete quesonnaire items with more than binary response categories available incomparave survey data, this shortcoming appears to be quite astonishing. One might argue that this is dueto the lack of appropriate textbooks that could advice scholars how to conduct related analyses. Yet, this maynot be the only barrier to mulnomial mullevel modeling in comparave survey research. Given that thecommon frequenst approach does not yield useful model fit criterions in the case of mulnomial mullevelanalysis (like the common AIC or BIC measures in hierarchical linear models), the according results oen lackvaluable informaon. A possible soluon to this problem seems to be offered by applying Markov Chain MonteCarlo (MCMC for short) methods...3.21.4 Cross-validang availability and random samples on the country-level in mullevel modelingT. Beckers 11 University of Düsseldorf, GermanyMullevel analysis has become a popular tool to analyze cross-seconal cross-naonal data sets with individuals(level 1) nested in countries (level 2). While the samples on level 1 can reasonably be assumed tobe proper random samples or strafied random samples, the selecon of countries is oen less systemac.Most commonly the researcher is confronted with data sets which do not provide a full sample of countriesof larger supranaonal enes. Thus a selecon of countries is chosen which uses availability or conveniencesamples based on external criteria, e.g. exisng research funding is the basis for the selecon of a country.Although this aspect is not necessarily correlated with the research queson, the randomness of the sampleis undermined and the meaning of standard errors and significance tests of effect coefficients on level 2 isquesoned...3.22 European Values Study - methodological and substanve applicaons ITo be held on July 21, 2011 from: 11:00 to 12:30, in room 410.Coordinated by:• Wolfgang Jagodzinski - GESIS, Germany• Ruud Luijkx - Tilburg University, Netherlands3.22.1 Value orientaons and le-right self-idenficaonO. Knutsen 11 University of Oslo, NorwayThe le-right semancs are central in polics, at the elite level as well as on the level of mass polics. The lerightscale is oen used to tap a general ”le-right” orientaon in polics and it is considered as expressions ofdifferent phenomena in mulvariate analyses. According to the approach in this paper le-right idenficaonis considered as an expression of value orientaons. ”Le” and ”right” are ideological labels and self-locaonon a le-right scale is ideological self-idenficaon because values movate people to place themselves in thele-right scale.3.22.2 Measuring support for democracy and non-democrac alternaves in EuropeC. Tufis 11 Research Instute for Quality of Life, Romanian Academy of Science, RomaniaThe paper focuses on atudes towards democracy, comparing how European cizens react to the principlesof the polical system in which they live. Using data from the European Values Survey (1990-2008, with focuson the 2008 wave), the paper aempts to understand the mechanisms of support for democracy and toidenfy both individual- and societal-level factors that affect people’s atudes towards democracy. Special

140 THURSDAY 21 JULYaenon is given to former communist countries in Central and Eastern Europe, tesng the hypothesis thatthe mechanisms of support in these countries are different from those exisng in older democracies.3.22.3 Simulang Polical Cleavages with Survey Data: On the Meaning of Being ”Le” in Different EuropeanCountriesG. Müller 11 Université de Fribourg, SwitzerlandLe-wing ideology has in different European countries different polical meanings. State ownership in industry,redistribuon of wealth, social security, or universalisc social cizenship are some of the major ideologicalelements, which vary however from country to country with regard to their importance for those, who definethemselves as being le.3.22.4 An-liberal atudes in Western Europe : comparing the effects of inequalies referring to individualsituaons and to internaonal indicatorsF. Gonthier 11 Instut d’Etudes Poliques - Université de Grenoble, FranceThe EVS gives the opportunity to analyze how the atudes of Western Europeans towards economy havechanged between 1990 and 2008. Confidence in major companies and liberal preferences concerning economyhave strongly declined since 1990 ; whereas confidence in states economic or social funcons and need forreforms have connuously increased. Moreover, this gradual erosion of the beliefs in market refers to differentdimensions of an-liberalism, that we can qualify as ”intervenonism”, ”egalitarism” and ”radicality”.3.23 Informed pursuit of high response rates IITo be held on July 21, 2011 from: 11:00 to 12:30, in room 414.Coordinated by:• Achim Koch - GESIS - Leibniz Instute for the Social Sciences, Germany• Andy Peytchev - RTI Internaonal, United States• Ineke Stoop - The Netherlands Instute for Social Research/SCP, Netherlands3.23.1 Nonresponse strategies in the European Social Survey (ESS)I. Stoop 1 , A. Koch 21 The Netherlands Instute for Social Research/SCP, Netherlands; 2 GESIS - Leibniz Instute for the Social Sciences,GermanyThe objecve of the ESS is to design, develop and run a conceptually well-anchored and methodologicallybullet-proof study of changing social atudes and values. Achieving these aims in a cross-naonal contextrequires ’opmal comparability’ in the operaonalisaon of the study within all parcipang countries. Withregard to nonresponse this would mean equal response rates, and a balanced response composion in everycountry and every round. In order to aain this, a target response rate of 70% is prescribed in the centralspecificaons, and countries are expected to allocate funds and deploy efforts to pursue this target.3.23.2 Response rates: can we predict success?A. Ramos 2 , S. Lavado 11 Instuto de Ciências Sociais - University of Lisbon (ICS) , Portugal; 2 University of Lisboa - ICS, PortugalThe complaint is universal: it is increasingly difficult to achieve high response rates. Interviewers and respondentscontribute to the problem in different ways (e.g. Hox, J., de Leeuw, E.,

THURSDAY 21 JULY 141& Kre, I. 1991; Hox, J.,& de Leeuw, E. 2002; Stoop 2005). Many reasons are given: people do not open the door to strangers; theyare not even a lile bit interested in listening to what the interviewer has to say; they do not trust in the interviewer;they believe that these studies are pointless; they cannot spend so much of their me answeringa quesonnaire…and so on. What can then be done to ensure a good sample and a reliable set of answers?With this presentaon we aim to show the work that the Portuguese research team has been doing in order toachieve the requested response rates of the European Social Survey; the procedures that have been introducedin each round to improve results; the consequences of those improvements...3.23.3 What do respondents and non respondents think of incenves? The ESS experience in PolandF. Sztabinski 1 , P. Sztabinski 11 Polish Academy of Sciences, PolandBackground: Much like other countries, Poland has seen a systemacally declining response rate in surveys.Incenves are recommended as one way to drive it higher (Dillman 2000, Biemer& Lyberg 2003, Groves& Couper 1998). On the other hand, ESS data analysis shows that in countries where incenves were used theresponse rate is somewhat lower in comparison with countries where no incenves were applied (Stoop et al.,2010). Based on authors’ interpretaon, this may stem from the fact that incenves were used in countrieswhich struggle to achieve a sasfactory response rate. But it may also arise from non-adherence to generaland country-specific rules of using incenves. In Poland, we have used incenves since Round 3 and monitoredtheir effecveness. In addion, we have conducted dedicated qualitave studies on the percepon ofincenves...3.23.4 Are you really that different: The effect of tailor made response enhancing measures on the measurementof outcome variablesJ. Kappelhof 11 Social and Cultural Planning office (SCP), NetherlandsIn recent years the Netherlands instute for Social Research/SCP has conducted several surveys among minorityethnic populaons. In order to achieve a high and balanced response general and tailor made measureswere employed such as increased number of contact aempts, translated quesonnaires, re-issuing refusalsand bi-lingual interviewers.3.24 The Quality of Longitudinal Survey Data: Measurement Error & Arion IIITo be held on July 21, 2011 from: 11:00 to 12:30, in room 415.Coordinated by:• Ulrich Krieger - SHARE MEA University of Mannheim, Germany• Peter Lugg - Utrecht University, Netherlands• Emanuela Sala - University of Milano Bicocca, Italy• Noah Uhrig - University of Essex, United Kingdom3.24.1 The Impact of Arion on Healthy Life Expectancies Esmated Using Longitudinal Survey DataM. Molla 11 Naonal Center for Health Stascs , United StatesAIMS: Summary measures of populaon health (SMPH) is used to summarize the health of a populaon takinginto account both mortality and morbidity. One such measure is the remaining years of healthy life. Healthylife is esmated using data from cross-seconal or longitudinal health surveys. Models that use data from

142 THURSDAY 21 JULYcross-seconal surveys use prevalence rates whereas models based on data from longitudinal surveys useincidence. While models that use longitudinal survey data are superior in terms of methodology, they alsohave challenges. One such challenge is esmaon bias caused by arion.3.24.2 Imputaon of rounded duraon dataJ. van der Laan 1 , L. Kuijvenhoven 11 Stascs Netherlands (CBS), NetherlandsOne type of recall errors occurring in surveys are rounding errors. For example, in the Dutch Labour ForceSurvey where unemployed persons are asked how long they have been unemployed, there is clearly a tendencyto give answers that are rounded off to years or half years. As rounding distorts the distribuon of the variableof interest, esmates may be biased when these are dependent on the shape of the distribuon. This canhappen, for example, when performing regression analyses, or when the shape of the distribuon itself is thestasc of interest. At Stascs Netherlands, there is an interest in publishing stascs on the distribuon ofunemployment duraons. It is for example of interest to see whether certain policies mainly affect personswith long unemployment spells or mainly persons with short unemployment spells...3.24.3 The effect of event history calendar on dang accuracy in an online surveyA. Neusar 1 , S. Ježek 11 Masaryk University, Czech RepublicTo increase the accuracy of dang various calendar methods like the event history calendar (EHC) have beenused (e.g. Belli, Stafford,& Alwin, 2009). Calendar methods incorporate aids into the common structured research interview that facilitatethe respondents’ recall. Studies have found small to substanal effect of calendar methods on dangaccuracy. In our exploratory study we were interested if calendar methods can bring the some posive effectin a self-administered online quesonnaire. The sample consists of 150 university students - volunteers. Thedang quesons in the online survey were both personal (days of illness; trip abroad) and public (joining EU;Tsunami in Indonesia). Random half were instructed to work out a self-instructed version of the EHC beforegoing on with the survey items. We measured calendar compleon and survey compleon mes as well as amanipulaon check...3.24.4 UNDERSTANDING AND ADJUSTING FOR WAVE NON-RESPONSE IN LONGITUDINAL STUDIESI. Plewis 11 Social Stascs, University of Manchester, United KingdomPromote the virtues of longitudinal studies and you will soon be challenged about the problem of missing data.Data are missing for a number of reasons and in a variety of ways. This talk will focus on the missingness thatarises from wave non-response – cases that are missing at wave t but observed again at one or more laterwaves t + k. The prevalence of wave non-response will be described for a number of well-known longitudinalstudies and the characteriscs of wave non-respondents will be compared with arion cases and consistentresponders. The fact that wave non-respondents provide informaon not only before but also aer a missingoccasion offers opportunies for a beer understanding of missingness mechanisms, including the possibilitythat data at wave t might be missing not at random in the Lile and Rubin classificaon. The data measuredaer wave t give scope for improving imputaon models...3.25 Linking Survey and Administrave Data: A Methodological Perspecve ITo be held on July 21, 2011 from: 14:00 to 15:30, in room 321.Coordinated by:

THURSDAY 21 JULY 143• Julie Korbmacher - SHARE, MEA, University of Mannheim, Germany• Annelies Blom - Survex - Survey Methods Consulng, Germany• Joe Sakshaug - JPSM, University of Michigan, United States• Mark Trappmann - Instute for Employment Research, Germany3.25.1 Extending the Empirical Basis for Wealth Inequality Research Using Stascal Matching of Administraveand Survey DataA. Rasner 1 , J. Frick 1 , M. Grabka 11 DIW Berlin, GermanySocial security entlements are a substanal source of wealth that grows in importance over the individual’slifecycle. Despite its quantave relevance, social security wealth has been thus far omied from wealthinequality analyses. In Germany, it is the lack of adequate micro data that accounts for this shortcoming. Thetwo main contribuons of this paper are: First, to elaborate a stascal matching approach that complementsinformaon on net worth as surveyed in the German Socio-Economic Panel (SOEP), a populaon representavepanel study, with informaon on social security wealth from the Sample of Acve Pension Accounts (SAPA),a large-scale administrave dataset maintained by the German Statutory Pension Insurance. In order to findout the best data fusion technique for the data at hand, we apply four stascal matching and imputaontechniques to the populaon of rerees...3.25.2 Nonresponse when linking survey data with administrave dataJ. Korbmacher 1 , M. Schroeder 21 SHARE, MEA, University of Mannheim, Germany; 2 DIW, GermanyThe Survey of Health, Ageing and Rerement in Europe (SHARE) has collected retrospecve life history data inits third wave (2008/2009) in thirteen European countries. In addion, in Germany, about 900 cases have beenlinked with the respondents consent to records of the German Pension Fund (Deutsche Rentenversicherung,DRV) in order to collect more exact and reliable measurements of income data and job spell informaon.3.25.3 Correlates of obtaining informed consent to data linkage: Respondent, interview and interviewercharacteriscsJ. Burton 3 , E. Sala 1 , G. Knies 21 University of Milano Bicocca, Italy; 2 Instute for Social and Economic Research, United Kingdom; 3 Universityof Essex, United KingdomIn the UK, in order to link individual-level administrave records to survey responses, a respondent needs togive their consent. This presentaon explores whether characteriscs of the respondent, the interviewer orsurvey design features influence consent. We use the Brish Household Panel Survey (BHPS) combined with asurvey of interviewers to model the probability that respondents consent to adding health and social securityrecords to their survey responses.3.25.4 A new cryptographic long-term stable key for generang panels based on survey and administravedataR. Schnell 1 , T. Bachteler 1 , J. Reiher 11 University of Duisburg-Essen, GermanyLongitudinal micro data are a rich source of informaon on important research topics all through the socialsciences. To obtain longitudinal data individuals must however be tracked over me. For example, in epidemiologicalresearch, a naonal cohort may be tracked lifelong in databases of health care providers. Incriminological research, the identy of offenders has to be known for compung individual risk of recidivism.

144 THURSDAY 21 JULY3.26 Spaal approaches to cross-contextual analysesTo be held on July 21, 2011 from: 14:00 to 15:30, in room 318.Coordinated by: Guy Elcheroth - University of Lausanne, Switzerland3.26.1 Spaal approaches to cross-contextual analyses: promises, challenges and new developmentsG. Elcheroth 11 University of Lausanne, SwitzerlandIn this introductory paper, the main issues outlined in the overall session descripon will be developed. Firstof all - why should comparave survey researchers be interested at all in spaal approaches? Scale and zoningeffects, conceptualised inially as a problem for comparave research, can similarly be viewed as an opportunityto learn more about the way social contexts shape human behaviour and social representaons. Thecrical impact of variable definions of contextual units on survey outcomes can rather easily be illustratedwith findings from the European Social Survey. The introducon of a connuous distance funcon into suchanalyses allows then to study scale effects more systemacally, and to clarify empirically the structure andboundaries of social systems within which contextual influences operate...3.26.2 Esmang Constuency Preferences from Sparse Survey Data Using Auxiliary Geographic Informa-onP. Selb 1 , S. Munzert 11 University of Konstanz, GermanyMeasures of constuency preferences are of vital importance for the study of polical representaon andother research areas. Yet, such measures are oen difficult to obtain. Previous survey-based esmates frequentlylack precision and coverage due to small samples, rely on quesonable assumpons, or require detailedauxiliary informaon about the constuencies’ populaon characteriscs. We propose an alternaveBayesian hierarchical approach that exploits minimal geographic informaon readily available from digitalizedconstuency maps. If at hand, social background data is easily integrated. To validate the method, we usenaonal polls and district-level results from the 2009 German Bundestag elecon, an empirical case for whichdetailed structural informaon is missing.3.26.3 The group threat hypothesis revisited with geographically weighted regressionC. Teney 11 Social Science Research Centre Berlin (WZB), GermanyThe group threat hypothesis has been extensively tested on an-immigrant vote with cross-naonal and na-onal mullevel analyses, but with mixed results. The group threat hypothesis states that the percepon bymembers of the majority group that an outside group threatens their group’s prerogave is posively associatedwith prejudice against the outgroup (Quillian, 1995). This perceived collecve threat is usually measuredwith macro-social condions that are presumed to imply larger degrees of compeon between the majorityand ethnic minories and thus to a greater perceived ethnic threat among the majority, namely both the immigrantand unemployment rates. This contribuon aims at challenging the state of the art literature on thegroup threat hypothesis in explaining an-immigrant vote by invesgang the spaal variaon of the associa-on between group threat and an-immigrant vote in Germany...3.26.4 Beyond physical space: Examining the impact of contextual indicators weighted by spaal andsocial proximity in mullevel analyses of survey dataS. Penic 2 , T. Junge 11 EPFL , Switzerland; 2 University of Lausanne, Switzerland

THURSDAY 21 JULY 145In mullevel models, geographic space is conceptualized as a disconnuous phenomenon that is divided intodiscrete, oen arbitrarily defined units with fixed boundaries. Such a pracce ignores the fact that usuallythere are important social, economic, and instuonal es that link residents from different units and thatthese links are stronger as proximity between the units is higher. In this study we argue that interdependencebetween the spaal units does not only result from physical proximity between them, but also from their socialproximity. We illustrate this assumpon with findings from comparave survey research conducted in post-warcommunies across the former Yugoslavia...3.27 Methodological and technical challenges in surveys of immigrant-backgroundand ethnic minority groups ITo be held on July 21, 2011 from: 14:00 to 15:30, in room 413.Coordinated by:• Laura Morales - Instute for Social Change, University of Manchester, United Kingdom• Mónica Méndez - Centro de Invesgaciones Sociológicas (CIS), Spain3.27.1 The role of funconal equivalence in the applicaon of psychometric instruments to migrant populaonsP. Brzoska 1 , O. Razum 11 Bielefeld University, School of Public Health, Dept. of Epidemiology& Int. Public Health, GermanyBackground:In quantave migrant health research, researchers oen apply instruments validated only for the source populaonsof the migrants. However, aer years in the host country, migrants may develop language usage andresponse behavior different from that of the source populaon. This may jeopardize the funconal equivalenceof instruments, i.e. comparability in their semanc, operaonal, item and measurement properes.These limitaons are usually not considered. We apply the Turkish version of the Revised Illness PerceponQuesonnaire (IPQ-R) to Turkish migrants residing in Germany to examine to what degree this pracce mayproduce invalid results.Methods:3.27.2 Challenges and soluons in implemenng a survey covering minories across 27 European countriesG. Hideg 1 , A. Illyes 1 , R. Manchin 21 Gallup, Belgium; 2 Gallup Europe, BelgiumThe European Union Minories and Discriminaon Survey (EU-MIDIS) was the first ever systemac large-scaleaempt to address vulnerable immigrant and ethnic minority groups with a scienfic survey in all 27 MemberStates of the EU, in 2008. The praccal design and implementaon were done by the authors.3.27.3 Sampling of ethnic Germans (Spät-Aussiedler): The quality of data drawn using onomasc methodsbased on telephone registry informaonE. Liebau 11 SOEP - DIW Berlin, GermanySince the fall of the Iron Curtain in 1990, around 2.5 million ethnic Germans have immigrated to Germany.This immigrant group has raised a number of interesng new theorecal quesons in the research on labormarket integraon. In contrast to the labor migrants who dominated the immigraon to Germany from the1960s to the 1980s, ethnic Germans are quite well educated, but sll have difficules entering the labor market.

146 THURSDAY 21 JULYAnalyses of ethnic Germans are scarce due to the difficules of drawing a random sample. There is no completeand easily accessible sampling frame for this immigrant group since the majority received German cizenshipimmediately aer immigraon and therefore ”disappeared” into the large pool of German naonals. Severalmethods have been used to construct a sampling frame using different sources, e.g...3.27.4 Sample frame and coverage issues for Indigenous populaon surveys in AustraliaR. Watmuff 11 Australian Bureau of Stascs, AustraliaNaonal surveys of Indigenous Australians (the Aboriginal and Torres Strait Islander peoples) are a core componentof the ABS household survey program. Producing appropriate sample designs for these surveys presentsunique challenges. The target populaon is rare (less than 3% of all Australians) and has two disnct composions:Indigenous people highly clustered in remote Indigenous communies and those dispersed thinlyacross highly populated areas. To idenfy the target populaon for the laer interviewers screen householdsin the field.3.28 Informed pursuit of high response rates IIITo be held on July 21, 2011 from: 14:00 to 15:30, in room 414.Coordinated by:• Achim Koch - GESIS - Leibniz Instute for the Social Sciences, Germany• Andy Peytchev - RTI Internaonal, United States• Ineke Stoop - The Netherlands Instute for Social Research/SCP, Netherlands3.28.1 Ancipatory Survey Design: Reducon of Nonresponse Bias through Bias Predicon ModelsA. Peytchev 1 , S. Riley 2 , J. Rosen 1 , J. Murphy 1 , M. Lindblad 2 , P. Biemer 11 RTI Internaonal, United States; 2 University of North Carolina at Chapel Hill, United StatesMethods aimed at maximizing response rates without consideraon of survey esmates may fail to reducenonresponse bias. Conversely, direcng survey effort to increase parcipaon in a manner that targets nonresponsebias reducon should be a prime objecve in probability-based surveys. Use of auxiliary informaon,such as available from the sampling frame, prior survey administraons, and including survey variables in addionto collected demographic characteriscs can be highly informave of predicng survey parcipaon.Targeng sample members who are least likely to parcipate rather than most likely may seem inefficient inreducing nonresponse, yet raonal if the goal is nonresponse bias reducon...3.28.2 Assessing Differences in Esmates Between a Compulsory and Voluntary Response American CommunitySurveya. navarro 1 , D. Griffin 21 US Census Bureau, United States; 2 U.S. Census Bureau, United StatesThe American Community Survey(ACS)is, and has always been, a compulsory response survey. The ACS usesthree modes of data collecon: mail, telephone, and personal visit. In 2003, the Census Bureau conductedresearch to assess whether the American Community Survey could be implemented as a voluntary, ratherthan a compulsory, survey. A test, not a randomized experiment, was designed to collect informaon to beable to answer key quesons about the impact, if any, that a change to a voluntary response survey would haveon mail response, cost, and some elements of data quality. Results from the test showed a dramac decreasein mail and and final response rate,reliability, and a significant increase in implementaon cost. This inialstudy did not assess if esmates from a voluntary ACS would differ from esmates from a compulsory ACS...

THURSDAY 21 JULY 1473.28.3 The design of the Dutch Parliamentary Elecon Study: Promising or not?H. Schmeets 11 Stascs Netherlands (CBS), NetherlandsSpecific measures to enhance response rates used in the Dutch Parliamentary Elecon Study (DPES) of the 22November 2006 elecons proved to be very successful. A specific fieldwork strategy at Stascs Netherlands,including new advance leers, incenves, interviewer training sessions, and re-approaching non-contacts andrefusals resulted in a 72% response rate (first wave) and 64% (second wave). Moreover, the distribuons ofparty choice in the DPES fit perfectly with the official elecon results. However, the 93.1% reported turnoutwas substanally higher than the official 80.1% turnout rate. Five percent of the 13% gap in reported andofficial turnout in the DPES is assigned to smulus effects and the remaining 8% to a mixture of nonresponsebias and social desirability effects.3.28.4 Maximizing response rates: following the line of least resistance?K. Beullens 1 , G. Loosveldt 11 Katholieke Universiteit Leuven, BelgiumCurrent ESS guidelines with respect to unit-(non)response involves predominantly the maximizaon of responserates (70% benchmark), although in recent ESS round more focus is devoted to the representavenessof the obtained sample. In this contribuon, arguments are collected to encourage a strategic shi fromresponse rate maximizaon towards nonresponse bias minimizaon.3.29 The Quality of Longitudinal Survey Data: Measurement Error & Arion IVTo be held on July 21, 2011 from: 14:00 to 15:30, in room 415.Coordinated by:• Ulrich Krieger - SHARE MEA University of Mannheim, Germany• Peter Lugg - Utrecht University, Netherlands• Emanuela Sala - University of Milano Bicocca, Italy• Noah Uhrig - University of Essex, United Kingdom3.29.1 Panel Condioning and in the Relaonship between Body-Mass and Economic OutcomesN. Uhrig 11 University of Essex, United KingdomHumans are reflexive beings; the act of observaon can potenally transform the atudes and behaviour ofthose being observed (Kalton& Citro 2000). This study examines whether ’panel condioning’ – exposure to panel survey content in earlierwaves – alters response behaviour at subsequent waves. Depending on the type of data obtained and thecondions under which condioning effects are examined, some evidence suggests that condioning mayenhance data quality. Lile research invesgates the extent to which panel condioning can affect substanveanalyses, however. I examine experimental longitudinal data from the Understanding Society Innovaon Panelon self-reported height and weight, and the associated effect of body-mass on a range of economic outcomes...3.29.2 Influences of Respondent and Interviewer Incenves on Survey Responses. An Empirical Analysisusing PASSY. Kosyakova 1 , S. Eckman 21 Faculty for Social and Economic Sciences / Chair of Sociology I / Oo-Friedrich-University Bamberg, Germany;2 Instute for Employment Research, Germany

148 THURSDAY 21 JULYResearchers want the highest quality data possible from their surveys, and oen employ interviewers to collectthese data from respondents. However, all three pares may have different incenves, with importanteffects on measurement error in survey data. This paper will use economic theories of incenves, in parcularasymmetric informaon and moral hazards, to examine the impact of respondent and interviewer incenveson data quality.3.29.3 Can We Use Public Landmarks as Recall Aids in Survey research?A. Neusar 1 , W. van der Vaart 21 Masaryk University, Czech Republic; 2 University for Humaniscs / GGZ inGeest Instute of Psychiatry andMental Health, NetherlandsOne of the methods that researchers use to aid recall in surveys consists of employing landmark life events.These are very salient events that aim to facilitate recall of other events and parcularly their dates. Generallypersonal landmarks are used, however, the dates of these events are very prone to recall error themselves.Using public events as landmarks would solve this problem since researchers can provide these events, includingtheir dates, to the respondent. However it is unclear what type of public events will be effecve to whichrespondents and under which condions. This paper aims to examine what kind of public events can be usedas effecve landmarks in survey research. One core indicator for this effecveness is whether respondents candate the public landmark accurately. If so, then this event will be of importance to the respondent and maybe able to cue other events...3.29.4 Interdisciplinary understanding of youth: Development and test of a longitudinal survey for theSwiss Federal Survey of AdolescentsS. Huber 2 , I. Lussi 1 , U. Moser 3 , F. Keller 3 , T. Abel 4 , S. Bucher 4 , K. Hofmann 4 , D. Schori 4 , S. Caacin 5 , J. Eberhard 51 Instut für Bildungsmanagement und Bildungsökonomie (IBB), Pädagogische Hochschule Zentralschweiz (,Switzerland; 2 Instut für Bildungsmanagement und Bildungsökonomie (IBB), Pädagogische Hochschule Zentralschweiz, Switzerland; 3 Instut für Bildungsevaluaon, assoziiertes Instut der Universität Zürich, Switzerland;4 Instut für Sozial- und Prävenvmedizin der Universität Bern, Switzerland; 5 Département de Sociologieder Universität Genf, SwitzerlandGeneral descripon on research quesons, objecves and theorecal framework: The project ”Swiss FederalSurvey of Adolescents, Main Indicators ch-x” iniates a long-term monitoring of Swiss youth in the fields ofeducaon, polics, and health behaviour. The target populaon are Swiss adolescents who are 19 years old,and it is planned that the survey will be conducted every four years in order to permit a longitudinal analysis ofthe Swiss youth. The theorecal framework is interdisciplinary, applying an adapted capability approach (Sen1985; 1987; 1993; 1999), which is a normave paradigm for the evaluaon of individual well-being.. In thisproject, the capability approach is enriched with concepts of self-percepon, values, and acons in differentcontexts. Methods/methodology: The survey comprises established survey items as well as newly developedones. The quesonnaire has undergone mulple pre-tesng...3.30 Mulple imputaon of complex data designs IITo be held on July 21, 2011 from: 14:00 to 15:30, in room 340.Coordinated by:• Roel de Jong - University of Hamburg, Faculty of Psychology, Germany• Krisan Kleinke - University of Bielefeld , Germany• Jost Reinecke - University of Bielefeld , Germany• Marn Spiess - Hamburg University, Department of Psychology, Germany

THURSDAY 21 JULY 1493.30.1 Bayesian Analysis of Binary Probit Models: The Case of Measurement Error and Sequenal RegressionModeling for Missing Explaining FactorsB. Schönberger 1 , C. Aßmann 11 Naonal Educaonal Panel Study, GermanySince large panel data sets on educaonal or epidemiological issues are despite tremendous efforts in fieldwork almost inevitably plagued by missing data, for example, due to item non response, development of appropriatemulple imputaon techniques is necessary to allow for valid inferences incorporang the uncertaintyabout missing values. Sequenal regression modeling is a widespread approach to generate mulple imputeddata sets. The structure of mulple imputaon algorithms is well suited for incorporaon in MCMC samplingalgorithms providing the analysis of primary interest. This paper implements two approaches to approximatethe full condional distribuon of missing values within a sequenal regression setup. In the context of a paneldata set of bone ages with missing data, simple parametric models are chosen to provide an approximaon ofthe full condional distribuon...3.30.2 Double standards regarding life-course events: a matching approach for split-ballot design surveysB. Arpino 11 Università Bocconi, ItalyMan and women can be evaluated differently regarding the same behavior. In this case we say that there is adouble standard. Past studies on double standards relied on quesonnaires with split-ballot designs, where arandom sub-sample is assigned to items regarding women and another to items regarding men. This approachdoes not allow to idenfy double standards at the individual level because each respondent either answer thequeson regarding men or women, while the other answer is missing. The soluon is to impute the missinganswers. The simplest approach uses regression models where among the covariates it is included a dummyfor the gender of target. This approach has some limitaons. First, a methodological drawback is that theimputaon relies on the goodness of the specified regression model assumpons (e.g., linearity)...3.30.3 Mulple Imputaon in Educaonal Survey Research with MICET. Stubbe 11 Instute for School Development Research, GermanyThe impact of missing data on results of mulvariate analyses is oen disregarded in educaonal survey research.Although there is strong empirical evidence that the missing data mechanism in these data is notMCAR (e.g. more missing data from parents with an immigrant background) ’listwise deleon’ is the rouneprocedure (as this is the default seng of most stascal soware packages) leading to systemacally biasedresults (Schafer,& Graham, 2002).Only in the last years mulple imputaon – as the state-of-the-art procedure – is becoming more establishedin educaonal research. Two general approaches for impung mulvariate data can be disnguished: jointmodelling (JM) and fully condional specificaon (FCS) (van Buuren, 2007).3.30.4 Imputaon of missing disposable household income in the SOEP datasetJ. Goebel 11 German Instute for Economic Research (DIW), GermanyThis talk will discuss mulple imputaon as a strategy for dealing with missing data due to item-non-responseon income quesons in large panel surveys.The German Socio-Economic Panel Study (SOEP) started in 1984 and is a wide-ranging representave longitudinalstudy of private households. Every year, nearly 11,000 households are surveyed, and the data provide

150 THURSDAY 21 JULYinformaon on all household members. In contrast to cross-seconal surveys, panel data make it possible toimpute missing values based on longitudinal informaon on the very same observaon units at different pointsin me, although at the cost of a more complex data design.3.31 Challenges in Educaonal Surveys IITo be held on July 21, 2011 from: 14:00 to 15:30, in room 315.1.Coordinated by:• Hans Kiesl - University of Applied Sciences Regensburg, Germany• Alina Matei - Instute of Stascs, University of Neuchatel; Instute of Pedagogical Research and Documentao,Switzerland• Leyla Mohadjer - Westat, United States3.31.1 Simulaon based determinaon of first stage’s sample size using mulstage sampling in educaonalsurveysH. Steinhauer 11 Naonal Educaonal Panel Study (NEPS), GermanyIn educaonal research, students in classes within schools are oen sampled via mulstage sampling respectingthe hierarchical structure of the educaonal system. In Germany class size strongly varies between differentschool types as well as between schools. Since students are located at the ulmate stage and there’s a fixednumber of clusters sampled, the resulng sample size of ulmate stage’s units is not under control. Besidesthat, nonresponse is a severe problem to most voluntary surveys and makes it difficult to determine the firststages sample size, when organizaonal or financial restricons are given.3.31.2 How to use the sampling weights in esmang of Item Parameters in Educaonal SurveysA. Uzaheta 1 , V. Cervantes 21 Colombian Instute for Educaonal Evaluaon (ICFES), Colombia; 2 Colombian Instute for Educaonal Evaluaon, ColombiaOne of the most important goals in educaonal surveys is measuring students’ proficiency and skills. Educaonalsurveys are usually carried out by applying a test to a sample of students in order to esmate theirability and the items’ parameters through an Item Response Theory model. These surveys are usually usedwith complex sampling designs to select the students. This way of selecon has associated a sampling weightfor each student, which is an important role in the item parameter and standard error esmaon. However,there has been some concern about the proper use of sampling weights in this context (e.g. Cyr, A.& Davies, 2005) and there is not large enough literature with descripon on how to incorporate the weightsin the item parameter esmaon...3.31.3 Item Nonresponse in Group Administered Quesonnaire with pupils/studentsM. Jelonek 1 , A. Szczucka 21 Cracow Uniwersity of Economics, Poland; 2 Jagiellonian University, PolandItem nonresponse is a problem to the extent that it biases significantly results of survey. Unfortunately, lilehas been published on the subject of item nonresponse in a specific field - pupil/students research. However,its prevalence makes it essenal that researchers be familiar with the purposes of item nonresponses and withits implicaons.In my presentaon I analyze the effect of item and person characteriscs on item nonresponse for wrienquesonnaire (the group administered quesonnaire) used comparavely with students and pupil. I disnguishinteracons between item and personal characteriscs and propose specific approach to examine incompletedata.

THURSDAY 21 JULY 1513.31.4 Mul-stage tesng in longitudinal designsS. Pohl 1 , C. Carstensen 01 Oo-Friedrich-Universität Bamberg, Naonales Bildungspanel, GermanyThe Naonal Educaonal Panel Study (NEPS) is a large-scale study with a longitudinal design that aims at inves-gang the development of competencies across the whole life span. An aim of the NEPS is (a) to measure thewhole ability range of the examinees (b) in a given me (c) as accurate as possible (d) without increasing panelmortality due to too difficult or too easy tests. In order to meet all these requirements mul-stage tesng isdiscussed.3.32 Comparave research IITo be held on July 21, 2011 from: 14:00 to 15:30, in room 319.Coordinated by:• Paolo Moncagaa - Universitat Pompeu Fabra, Spain• Willem E. Saris - Universitat Pompeu Fabra, Spain3.32.1 Measurement of SupranaonalismL. Coromina 1 , W. Saris 21 University of Girona, Spain; 2 Universitat Pompeu Fabra, SpainA crucial issue in the European Union (EU) is which policies should be regulated by EU and which ones bynaonal governments. The level of polical decision power, supranaonal or naonal/regional level, has beendecided without much influence of the cizens in the member states. Given this situaon it is interesng tostudy the supra-naonalism of the people in the different countries. Therefore the issues studied here are:- How can one determine the supra-naonalism of the European cizens?- Is the homogeneity of these opinions such that one scale can be used in the whole of Europe?3.32.2 Comparave Survey Research in the GCC RegionD. Al-Emadi 2 , D. Howell 1 , A. Diop 21 University of Michigan, United States; 2 Social and Economic Survey Research Instute (SESRI), Qatar University,QatarThe GCC (Gulf Cooperaon Council) states, consisng of Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, and theUnited Arab Emirates, are at a very unique period in their history. The naons and their cizens find themselvesadapng to newfound wealth, majority expatriate populaons, globalizaon, and accompanying social change.All this while the sociees are seeking to balance and preserve long-standing idenes, values and tradionsthat are tribal, religious and cultural in nature. This is a period worthy of careful study, with one valuable toolbeing survey research. However, survey research and comparave research (especially academically-based)are less widely pracced and capacity less developed in the GCC states than in much of the Western world...3.32.3 Esmang Effects of Macro- and Micro Variables on Individual Characteriscs in a Cross-CountryFramework. RE- and FE- Techniques under ReviewM. Giesselmann 2 , H. Lohmann 3 , H. Hans-Jürgen Andreß 11 University of Cologne, Germany; 2 SOEP - DIW Berlin and University of Cologne, Germany; 3 German Instutefor Economic Research (DIW), GermanyAn important task of empirical social science is to determine the impact of macro-economical, instuonaland cultural characteriscs on individual living condions and behavior. As a major instrument to examinesuch macro-micro hypotheses, the country-comparave approach has well established. This approach regards

152 THURSDAY 21 JULYcountries as representaves of a set of macro characteriscs. According to this perspecve on countries,variaons in individual outcomes are related to variaons in the country characterisc of interest, usually byapplying random effects esmaon. In this paper, the comparave design with random effects esmaon iscrically reviewed. Addionally, we will discuss alternave techniques.3.32.4 How to deal with aberrant response paerns in cross-naonal data: Atudes towards progressivetaxaon in ISSP countriesI. Bechert 2 , J. Edlund 11 Department of Sociology, University of Umea, Sweden; 2 GESIS - Leibniz Instute for the Social Sciences, GermanyIn internaonal comparave survey research, scholars are oen confronted with unexpected findings. Whenfinding response paerns that do not fit theorecal expectaons, researchers face a dilemma: (1) is the findinga substanve and true finding, for which we lack any theory explaining it (a theory problem), or (2) is thefinding a research artefact generated by the survey instrument itself (a data problem)? In the ISSP data of1996 and 2006 dealing with ”The Role of Government” such an unexpected response paern is found in somecountries in a variable baery concerned with the issue of progressive taxaon. While in the majority of countriesprogressive taxaon is supported, in some less affluent countries a majority of respondents would likethe poor to carry a higher tax burden than the rich. This is a prime example of a response paern that mayeither be regarded as a problem of theory or data...3.33 Preferences and Well-being in Europe IITo be held on July 21, 2011 from: 14:00 to 15:30, in room 303.Coordinated by: Mariya Aleksynska - CEPII, France3.33.1 Secularizaon or religious revival? Measuring church aendance trend in Italy using mullevelanalysisF. Biolca Rinaldi 1 , C. Vezzoni 11 University of Milan, ItalyThe debate over secularizaon in Italy is far from having reached unanimous conclusions: some scholars underlinean unbroken trend toward a decrease of religiosity, while other scholars highlight the signs of a religiousrevival especially in younger generaons. Beside difficules of definions, such different conclusions arealso due to the lack of informaon on a sufficiently long period of me. In the paper this problem is tackleddeveloping a joint analysis of different repeated cross-secons: Eurobarometer (1973-1998, 18 waves), Issp(1987-2001, 11 waves), Evs (1980-2008, 4 waves), Ess (2003-2006, 2 waves), Italian Naonal Elecon Studies(Itanes) (1968-2008, 9 waves), Istat Mulpurpose Household Survey (1993-2008, 15 waves)...3.33.2 Preferences for redistribuon and the percepon of a fair societyE. Guillaud 1 , B. Françon 11 Université Paris 1 Panthéon-Sorbonne, FranceUsing ISSP micro survey data for 25 OECD countries, we show that there is no link between individuals’ preferencesfor government redistribuon and their desired type of society in terms of inequality. In line with recenttheorecal literature, this lead us to consider the role of individual beliefs about the incenves cost (or disincenveeffect) of redistribuon. Different beliefs about the role of effort and luck in individual success underlietheir preferred level of taxes, which in turn could explain observed naonal differences in the development ofwelfare states.

THURSDAY 21 JULY 1533.33.3 Religiosity of the Czech populaon in generaonal perspecveM. Laudátová 11 Masaryk University, Czech RepublicThe goal of the paper is to analyze religiosity of Czech populaon from a generaonal perspecve (Mannheim1928) and answer two quesons: how does the religiosity differ considering generaon persistence in meand how important is the process of socializaon of generaons with respect to religion. Data from EuropeanValues Study surveys from the years 1991, 1999 and 2008 will be used and most frequently used indicatorsassociated with tradional religiosity will be chosen for the stascal analysis. It is because the socializaonprocess is related primarily to transmission of this type of beliefs and values. Special emphasis will be puton the possibility of disnguishing of three effects – life-cycle, generaon and period effect (Riley 1972) – onreligiosity of generaons in the Czech Republic.3.33.4 Alcohol consumpon and happiness: An empirical analysis using Russian panel dataS. Massin 1 , P. Kopp 11 Centre d’Economie de la Sorbonne, FranceThis paper is based on the idea that the consumpon of addicve goods in general, and of alcohol in parcular,is very likely to rely on an imperfect decision process and gains being analyzed through subjecve sasfacondata.3.34 European Values Study - methodological and substanve applicaons IITo be held on July 21, 2011 from: 14:00 to 15:30, in room 410.Coordinated by:• Wolfgang Jagodzinski - GESIS, Germany• Ruud Luijkx - Tilburg University, Netherlands3.34.1 Trust in instuons and jusficaon of fraud from a comparave perspecveC. Abreu Lopes 2 , E. Bartolomé Peral 11 University of Deusto, Spain; 2 London School of Economics, United KingdomThe main objecve of this paper is to explore the individual and structural factors that support cizens’ jusficaonof fraudulent pracces such as cheang on taxes, falsely claiming state benefits and paying cash with noreceipt. At the individual level, jusficaon of fraud is explained through trust in instuons such as the parliament,government, polical pares, and social security system. Countries differences on levels of jusficaonof fraud are framed in the structural context of some European countries in a comparave perspecve.3.34.2 The effect of communist secularizaon on church aendance in Romania: An Age - Period- CohortAnalysisM. Voicu 11 European Data Laboratory for Comparave Social Research, GESIS, Cologne, GermanyThe paper focuses on the dynamic of religious pracce in post-communist Romania. Changes in religious beliefsand behaviors are usually aributed to three different causes: cohort effects, contextual effects and age affects.Romania, as a former communist society has experienced a dual process of secularizaon, a self-induced,generated by modernizaon, as well as an enforced one imposed by the totalitarian regime (Meulermann,2000, 2004; Need, Evans, 2001; Pollack 2001). Modernizaon produces a fast self-induced secularizaon andyounger cohorts are more likely to be less religious then the older ones. The effect of enforced secularizaongenerated a different relaon between age and religiosity (Need, Evans, 2001). Like in self-induces secularized

154 THURSDAY 21 JULYsociees, younger generaon were less exposed to religious beliefs and ideologies and a generaonal declinemost likely occurred...3.34.3 The impact of social values on school achievementB. Voicu 11 Romanian Academy of Science, Research Instute for Quality of Life, RomaniaThis paper tests for the effect of society’s average value orientaons on school achievement. I follow Fensham’s(2007) argument that there is an explicit need for contextualizing the explanaon of the average TIMSS and PISAdifferences across countries, parcularly considering some ”more fundamental values and complex mores”.The main hypothesis is that when controlling for various other determinants, the school achievement willdepend on what society thinks that school should teach children.3.34.4 Theories and Researches in Western Sociology of Religion: Otherness in the Study of ReligionK. Manabe 11 Aoyama Gakuin University, JapanIn this presentaon, first, based on a literature survey, I try to do an overview of concepts, proposions andtheories in Western Sociology of religion, namely secularizaon, religious pluralism, and religions market.Then, based on a search for code-books, I examine the queson items, wordings and scales of large-scalemul-naonal comparave surveys, namely EVS, ISSP, and RAMP. One problem of above-menoned scienficworks is that theories and researches are based on observaons, measurements and generalizaons madefrom Western perspecves. Lastly, I suggest a future direcon of the scienfic study in this field, which isdescribed as the codificaon of theories and researches from comparave perspecves.3.35 Mullevel analysis in comparave research IITo be held on July 21, 2011 from: 14:00 to 15:30, in room 315.Coordinated by:• Bart Meuleman - University of Leuven, Belgium• Elmar Schlueter - University of Cologne, Research Instute for Sociology, Germany3.35.1 The impact of rereporng on children’s developmental trajectories from early childhood to earlyadolescence: A mullevel approachF. Li 2 , M. Godinet 1 , P. Arnsberger 21 University of Hawaiii, United States; 2 University of Hawaii, United StatesUsing a mullevel growth model, this study analyzed individual developmental trajectories of behavioral problems,such as anxiety and depression, among 758 mul-cultural children aged 4 to 12 with early child maltreatmentexperiences. These children were recruited from five study sites across America. In parcular, this studyexamined (1) whether the trajectories of these problems differ by child’s gender and ethnicity, controlling forstudy sites; (2) whether the trajectories are influenced by reports of repeated maltreatment, aer controllingfor child’s gender, ethnicity and study sites; and (3) whether the impact of rereporng is moderated bychild’s gender and ethnicity, controlling for study sites. The results indicated that on average, these problemsincreased significantly from age 4 to age 12 (p

THURSDAY 21 JULY 1551 Universität zu Köln, GermanyAccording to Inglehart (1997) social change follows a sequence of two successive processes called Moderniza-on and Postmodernizaon. During the Modernizaon process tradional, usually religious norms are supplantedby achievement-oriented, increasingly secular-raonal norms. Aer World War II, advanced industrialsociees have aained unprecedentedly high levels in physiological and economic security. Due to a diminishingmarginal ulity of further economic accumulaon, a Postmodern shi took place in Western Europeand North America. As a consequence, maximizing economic gains gradually fades from top priority of an increasingnumber of cizens: subjecve well-being and self-expression values are becoming more crucial for agrowing segment of the populaon...3.35.3 Pung latent variable contextual models to pracce, or: novel answers to the queson why na-onal idenficaon promotes prejudiceE. Schlueter 11 University of Cologne, Research Instute for Sociology, GermanyThis presentaon demonstrates three key advantages of applying latent variable contextual models for examiningmeasurement- and structural models within- and between groups (Marsh et al. 2009). This class of modelsnot only helps researchers to adequately account for measurement error at different levels of analysis and tocorrect for sampling error when the number of observaons within groups for forming group-level constructsis small. As an added benefit, this methodology also offers an opportunity to examine cross-level interaconsbased on latent constructs within- and between groups. These features are illustrated by tesng a novel setof contextual-, individual- and cross-level explanaons for answering the queson why naonal idenficaonenhances an-immigrant prejudice.3.35.4 The applicaon of mullevel modeling to the me-series vong dataD. Poznyak 1 , K. Abts 1 , M. Swyngedouw 11 Katholieke Universiteit Leuven, BelgiumThe topic of vong behavior has been usually analyzed using single-shot cross-seconal data, which eliminatesthe dynamic aspect of the problem, i.e. the changes in vong paerns following the change in contextualpredictors across me. As a result, such studies can only provide a ”here and now” view on the problem,which may not be sufficient. Oen, however, the research interest is to invesgate how the vong behaviorchanges across me following the changes in the magnitude of context-level predictors, like unemployment,inflaon, crime, income inequality or proporon of foreign populaon.3.36 Construcon of Answer Scale Formats in Quesonnaires IITo be held on July 21, 2011 from: 14:00 to 15:30, in room 412.Coordinated by: Natalja Menold - GESIS - Leibniz Instute for the Social Sciences, Germany3.36.1 Different funconing of rang scale formats – results from psychometric and physiological experimentsM. Koller 1 , T. Salzberger 11 WU Vienna, AustriaThe type of response scale may compromise the psychometric quality of a survey instrument even though itsqualitave underpinning is excellent and the wording of the items is appropriate. A series of four experiments(one paper-and-pencil, two online, one eye-tracking experiment) were conducted to examine the funconingof response scales of different direcon (agree-disagree versus disagree-agree). The effects of the direconof the response scale are exemplified by two different constructs, based on the Rasch measurement model.

156 THURSDAY 21 JULYAs a key result, the agree-disagree format performs beer (in terms of fit and the unit of measurement) thanthe disagree-agree variant. Since a difference in the unit of measurement (a mulplicave bias) may lead tospurious mean differences, extreme cauon has to be exercised when data sets based on different modes ofcollecon or different response scales are merged...3.36.2 Explaining more variance with visual analogue scales: A Web experimentF. Funke 11 , GermanyScope: This study focuses on measurement error, one component of error of observaon in the frameworkof the total survey error (Groves et al., 2009). More precisely, this research is about is formang error thatoccurs if a rang scale does not provide a perfectly matching response opon (see Schwarz& Oyserman, 2001). Therefore data collected with two different closed-ended rang scales - convenonal 5-point scales and graphical visual analogue scales (VASs) - were checked against each other. About VASs: Thegeneral advantages of VASs are (1) great sensivity because of a great range, (2) data are less affected by error,leading to more stascal power (see Funke, 2010), and (3) there are far more possibilies for data analysis(e.g., recoding into odd and even number of categories, as well as into any empirical quanle)...3.36.3 How easy is a text that is not difficult? Comparing answers to posive, negave, and bipolar ques-ons.N. Kamoen 2 , B. Holleman 2 , H. van den Bergh 11 University of Amsterdam, Netherlands; 2 University of Utrecht, NetherlandsRespondents give different answers to posive (X is good. Agree-Disagree), negave (X is bad. Agree-Disagree)and bipolar quesons (X is bad-good). This makes survey answers hard to interpret, especially since conflicngresults are reported on how the answers are affected. The current study focuses on the generalizability of suchresponse effects. In twelve studies (each N=about 200) the effect of the same thirteen contrasve word pairswas examined. In each study, three survey versions were constructed: quesons worded posively in the oneversion, were negavely worded in the other and on a bipolar scale in the third.3.37 What is meant by mode effect on measurement? ITo be held on July 21, 2011 from: 16:00 to 17:30, in room 319.Coordinated by:• Steven Hope - Naonal Centre for Social Research - NatCen, United Kingdom• Gerry Nicolaas - Naonal Centre for Social Research - NatCen, United Kingdom3.37.1 Causes of Mode Effects on Survey MeasurementA. Jäckle 2 , P. Lynn 2 , P. Campanelli 1 , G. Nicolaas 3 , S. Hope 3 , A. Nandi 21 The Survey Coach, United Kingdom; 2 University of Essex, United Kingdom; 3 Naonal Centre for Social Research- NatCen, United KingdomThere are hundreds of studies which experimentally compare the effects of different modes of data colleconon measurement. Some consistent findings and theories have emerged from these studies, notably regardingthe measurement of sensive atudes and behaviours. Whether or not respondents will answer a parcularsurvey queson differently in one mode than in another is, nonetheless, sll mostly unpredictable. This is inpart because many experimental tests focus on descripve comparisons of response distribuons and do notaempt to idenfy generalisable features of the queson or the modes that might cause the differences inmeasurement.

THURSDAY 21 JULY 1573.37.2 The role of visual and aural smuli in producing mode effects on answers to survey quesonsP. Lynn 2 , S. Hope 3 , A. Jäckle 2 , P. Campanelli 1 , G. Nicolaas 3 , A. Nandi 21 The Survey Coach, United Kingdom; 2 University of Essex, United Kingdom; 3 Naonal Centre for Social Research- NatCen, United KingdomJäckle et al (2011, this session) argue that differences in the characteriscs of the queson administraonmethod are a cause of mode effects on measurement. One such characterisc is the communicaon channelsby which quesons and answers are transmied between researcher/interviewer and respondent. For anyparcular survey queson, a combinaon of channels may be used. There are three pieces of informaonto be communicated – the queson, the response opons, and the answer – and there are mulple possiblechannels for each. A respondent may perceive a survey queson either visually or aurally, but the responseopons need not be perceived through the same channel (e.g. if an interviewer reads the queson but therespondent has to read the response opons on a card)...3.37.3 The role of the interviewer in producing mode effects: results from a mixed modes experimentS. Hope 2 , P. Campanelli 1 , G. Nicolaas 2 , P. Lynn 3 , A. Jäckle 3 , A. Nandi 31 The Survey Coach, United Kingdom; 2 Naonal Centre for Social Research - NatCen, United Kingdom; 3 Universityof Essex, United KingdomIn this presentaon we focus on the role of the interviewer in causing mode effects, contrasng modes inwhich an interviewer is used (face-to-face and telephone) with a mode without an interviewer (web). Otherdifferences between modes, such as aural versus visual transmission of informaon, are held constant wherepossible.3.37.4 Is it a good idea to opmise queson format for mode of data collecon? Results from a mixedmodes experimentG. Nicolaas 2 , P. Campanelli 1 , S. Hope 2 , A. Jäckle 3 , P. Lynn 3 , A. Nandi 31 The Survey Coach, United Kingdom; 2 Naonal Centre for Social Research - NatCen, United Kingdom; 3 Universityof Essex, United KingdomIt is common pracce for survey designers to change how quesons are asked and answered to beer fit thesurvey mode. For example, the visual presentaon of response lists in self-compleon modes and face-to-faceinterviews with show cards allows the survey designer to use long lists. However, such lists are not feasiblein telephone interviews relying solely on verbal communicaon, and this oen leads to the use of radicallydifferent queson formats between modes.3.38 Innovaons in survey sampling ITo be held on July 21, 2011 from: 16:00 to 17:30, in room 414.Coordinated by:• Seppo Laaksonen - University of Helsinki, Finland• Siegfried Gabler - GESIS - Leibniz Instute for the Social Sciences, Germany3.38.1 Generalized calibraon with latent variablesA. Matei 11 Instute of Stascs, University of Neuchatel; Instute of Pedagogical Research and Documentao, SwitzerlandNonresponse is defined as the failure to provide the required informaon by a unit selected in a sample. Dealingwith nonresponse is an important topic, since nonresponse is present almost in all surveys, and can bias

158 THURSDAY 21 JULYesmaon. We focus on unit non-response and propose a method to reduce non-response bias in the case ofnon-ignorable nonresponse based on latent variable models.3.38.2 Designing minimum cost mul-stage sample designsM. Ganninger 1 , S. Gabler 11 GESIS - Leibniz Instute for the Social Sciences, GermanyIn cross-naonal sample surveys like the ESS, a huge variety of sample designs is oen applied in parcipangcountries. In order to achieve esmates of comparable precision, the samples drawn according to these differentsampling schemes must be of equivalent effecve sample sizes, n_eff = n/deff, where n is the net sample size and deff is the design effect. As deff, among another parameter,depends on the average cluster size b, increasing the number of sampled clusters, ceteris paribus, decreasesthe design effect and hence increases n_eff. The presentaon will show that, at a given linear cost structure (costs per interview and costs per sampledcluster), there exists an opmal number of clusters to sample so that a pre-defined effecve sample sizeis exactly achieved - at minimum total costs.3.38.3 Indirect Sampling using Dual FramesM. Maia 2 , P. Vicente 11 ISCTE-Lisbon University Instute, Portugal; 2 Portuguese Catholic University, PortugalIn any survey, the random selecon of the sample requires that a sampling frame is available. There are manypopulaons, though, for which a sampling frame that lists the individual elements are not readily available.When available, one central stascal concern for the survey researcher is how well the sampling frame actuallycovers the target populaon. In pracce, perfect frames seldom exist; there are always problems that canoccur, namely: (a) undercoverage, (b) duplicaon and (c) overcoverage.3.38.4 Dealing with Household Nonresponse using Generalized CalibraonG. Osier 11 Service Central de la Stasque et des Etudes Economiques (STATEC), LuxembourgIn most cases, in spite of all efforts dedicated to prevent it, Household Nonresponse cannot be eradicated andreweighng techniques must be used in order to compensate for nonresponse bias, that is, the bias caused bydifferences in esmates between responding and nonresponding households. Powerful auxiliary informaonis needed to make a notable impact on bias reducon. On the other hand, such informaon is rarely availablein many praccal situaons. For instance, when nonresponse is said to be non-ignorable, that is, the responsepropensity of a household depends on the target variables of the survey, it is parcularly difficult fing anonresponse model to the data since some of the explanatory variables are observed over the respondinghouseholds only.3.39 Analyzing social change with repeated cross-secons ITo be held on July 21, 2011 from: 16:00 to 17:30, in room 303.Coordinated by:• Christof Wolf - GESIS - Leibniz Instute for the Social Sciences, Germany• Tilo Beckers - University of Düsseldorf, Germany3.39.1 Lessons on Developing Laws for Studying Societal ChangeT. Smith 11 University of Chicago, United States

THURSDAY 21 JULY 159The Laws of Studying Societal Change might go as follows:First Law: The way to measure change is not to change the measure.Second Law: When constant measures produce non-constant measurement, change the measure to measurechange.3.39.2 How to apply mullevel models to analyze social change using repeated cross-seconal dataM. Boehle 1 , C. Wolf 11 GESIS - Leibniz Instute for the Social Sciences, GermanyFrom a methodological and sociological perspecve, analyzing social change is best done by (1) using repeatedcross-seconal data and (2) by including individual level variables, me and me-dependent macro variables.Furthermore, (3) interest very oen focuses on whether the effects of explanatory variables change over me,e.g. the impact of educaon on one’s social posion. We argue that mullevel trend models are best suited tofulfill these requirements. However, methods for applying the mullevel approach to study me-dependentprocesses are comparavely poorly elaborated. Most analyses of repeated cross-seconal data study changesof individual level effects neglecng changes at the macro level. In contrast, pooled me series analysis focusesexclusively on macro level variables...3.39.3 On the Mulple Ways of Using Mullevel Models to Study Social ChangeM. Fairbrother 11 University of Bristol, United KingdomAnalyses of repeated cross-seconal survey data have relied increasingly on mullevel/random effects models,in two ways. First, mullevel models have been used to disnguish age, period, and cohort effects, wherethe goal is to understand the mechanism by which some social change is occurring. Second, models of surveyrespondents nested within social units (typically countries or states) have been used to examine the effectsof society-level condions on individual-level outcomes. Both approaches, however, provide limited insightsinto the drivers of change over me. The former approach does not exploit differences among sociees experiencingmore or less change, and the laer does not disnguish longitudinal from cross-seconal variaon...3.39.4 Disentangling contextual and individual change using Bayesian Hierarchical Generalized AddiveModelsD. Stegmueller 11 University of Mannheim, GermanyIn the absence of panel data, using repeated cross-seconal data is the most suitable way to disentangle effectsof contextual and individual-level change on atudes and behavior. By approximang a panel structure,researchers can follow cohorts through me and examine their reacon to contextual and instuonal change.However, the linear dependence between age, cohort and year of observaon makes this task difficult. Buildingon recent innovaons in biostascs and sociology, I show how to simultaneously esmate the influenceof age, period and cohort membership on individual behavior. By employing Bayesian hierarchical generalizedaddive models, one can break the linear dependence and model the mullevel nature of the data, whereindividuals are simultaneously cross-classified in cohorts and me points...3.40 Mobilise! Incorporang Mobile Phones in Survey Research ITo be held on July 21, 2011 from: 16:00 to 17:30, in room 318.Coordinated by:• Robert Manchin - Gallup Europe, Belgium• Marek Fuchs - Darmstadt University of Technology, Germany• Femke De Keulenaer - Gallup Europe, Belgium

160 THURSDAY 21 JULY3.40.1 Mobile-only – Persistent status or passage in the life course? Results from a Mobile Phone PanelStudyB. Busse 1 , C. Neuert 1 , M. Fuchs 11 Darmstadt University of Technology, GermanyIn recent years the mobile-only populaon has gained more and more importance for survey research since ithas increased consistently in most industrialized countries. In some European countries like Finland, Swedenor the Czech Republic mobile phone penetraon rates have exceeded 90 percent; at the same me landlinepenetraon rates have declined. Thus, mobile-only rates have surpassed 30 percent in many countries. Asknown from the literature (e.g. Blumberg and Luke 2010; Fuchs, 2008; Ehlen and Ehlen, 2007) mobile-onlysdiffer from the landline populaon as they are young, single, male and either belonging to a low-income or ahigh-income segmentaon. Consequently it is necessary to complement tradional landline telephone surveymethods with mobile phone survey in order to cover the mobile-onlys in telephone surveys...3.40.2 Survey Response via Mobile Phone: A Total Survey Error PerspecveP. Lynn 1 , O. Kaminska 11 University of Essex, United KingdomThe search for cost effecve survey designs with high coverage and high response increasingly leads surveypraconers to consider data collecon via mobile phones, typically as part of a mixed-mode design. However,quality of measurement is also important. To conduct high quality surveys via mobile phone one needs tounderstand what factors may influence responses to survey quesons in this mode and how, if at all, thesefactors may differ from other modes.3.40.3 Are dual users from landline and mobile samples the same?A. Alanya 1 , F. De Keulenaer 1 , R. Manchin 11 Gallup Europe, BelgiumDual frame (landline and mobile phone) samples are increasingly used in telephone surveys; there is, however,no consensus on the best design for such surveys. Integrang mobile and landline samples may beaccomplished in two ways: (1) segmented designs which screen out respondents with landlines from the mobilesample (i.e. only mobile-only are interviewed by mobile phone) and (2) overlapping or ”full” dual framedesigns which include dual users from the mobile and landline samples.3.40.4 Reaching the Young and the Mobile - Reducing the CATI Landline Telephone Bias using the Dual-Frame Approach in a large Naonal Survey in SwitzerlandS. Klug 2 , D. Müller 11 IBSF Instut für Begleit- und Sozialforschung Zürich, Switzerland; 2 DemoSCOPE, SwitzerlandWhile the availability of people through households with landline telephones declined over the last years theusage of randomly generated mobile telephone numbers to cover up for this lack of representaveness hasnot yet been fully implemented as a standard approach in social and market research surveys.DemoSCOPE has been awarded the contract for a mul-year large naonal survey (Connuous Rolling Surveyof Addicve Behaviours and Related Risks, CoRolAR) for the Schweizerische Eidgenossenscha, Bundesamt fürGesundheit (BAG), Bern, on behalf of the IBSF Instut für Begleit- und Sozialforschung with 11’000 interviewsannually using the dual-frame approach.3.41 The Quality of Longitudinal Survey Data: Measurement Error & Arion VTo be held on July 21, 2011 from: 16:00 to 17:30, in room 415.Coordinated by:

THURSDAY 21 JULY 161• Ulrich Krieger - SHARE MEA University of Mannheim, Germany• Peter Lugg - Utrecht University, Netherlands• Emanuela Sala - University of Milano Bicocca, Italy• Noah Uhrig - University of Essex, United Kingdom3.41.1 Response Bias in Global Online Panels: Arion EffectsE. Trimarchi 11 Mktg, Inc., United StatesIn the online commercial sphere panels provide the lion share of respondents. These databases of surveyparcipants have double-opted in to complete surveys. There has been considerable debate that the hyperacverespondents, those who complete many surveys and belong to a large number of panels, are sourcesof variability and bias in the online data collected from them. Clearly, if they are different from new respondents,then their shiing numbers provide a dominant force that serves to increase variability in the onlinepanel universe. An 18 minute online quesonnaire was fielded among over 200 commercial panels in thirtyfive countries returning over 121,000 responses. New panel members are found to differ from long term andhyperacve respondents when measured against buying behavior and media segmentaons. The need forstandardized metrics to control fluctuaons in these panels is argued...3.41.2 Survey Error and the Theory of Liking: Evidence based on a Household Panel’s Interviewer SurveyM. Weinhardt 1 , F. Kreuter 21 DIW Berlin, Germany; 2 University of Maryland, United StatesThis paper invesgates how similarity between survey respondents and interviewers regarding their atudesand personality traits impacts on panel arion and data quality. The ’theory of liking’ is a social-psychologicalmodel of response behavior which rests on the assumpon that likeness induces likability, i.e. that the morepeople are alike, the more they will also like each other. While the theory has been formulated to explainsurvey parcipaon, we extend the theory to explain measurement error also. Our lead hypothesis is that themore interviewer and respondent resemble each other on the key characteriscs of atudes and personality,the more likely respondents will be persuaded to parcipate in the survey and to deliver complete and accurateresponses. This study uses data from an exceponally detailed interviewer survey linked to respondent datato invesgate the theory of liking...3.41.3 Panel arion. Seperang stayers, sleepers and lurkersP. Lugg 3 , J. Hox 1 , E. De Leeuw 1 , A. Scherpenzeel 21 University of Utrecht, Netherlands; 2 CentERdata, Netherlands; 3 Utrecht University, NetherlandsErrors stemming from panel dropout (arion) are one of the largest sources of survey error in panel surveys.Earlier studies into the determinants of arion have mostly made a disncon between respondents sll inthe survey, and those who arited in any given wave of data collecon. The difference between the two groupscan yield informaon on arion bias and the causes of arion. Addionally, survival or hazard-rate modelsare used to study when arion takes place. In many panel surveys, the process of arion is however moresubtle than being either in or out of the study. Respondents oen miss out on one or more waves, but mightreturn aer that. Others start responding infrequently, but respond more frequently later in the study. Usingcurrent models, it is difficult to incorporate such non-montone response paerns in analyses of arion. Wepropose to study arion in a Latent Class framework...3.41.4 Interdisciplinary understanding of youth: Development and test of a longitudinal survey for theSwiss Federal Survey of AdolescentsI. Lussi 1 , U. Moser 2 , F. Keller 2 , T. Abel 3 , S. Bucher 3 , K. Hofmann 3 , D. Schori 3 , S. Villiger 3 , S. Caacin 4 , J.Eberhard 4

162 THURSDAY 21 JULY1 Instut für Bildungsmanagement und Bildungsökonomie (IBB), Pädagogische Hochschule Zentralschweiz (,Switzerland; 2 Instut für Bildungsevaluaon, assoziiertes Instut der Universität Zürich, Switzerland; 3 Instutfür Sozial- und Prävenvmedizin der Universität Bern, Switzerland; 4 Département de Sociologie der UniversitätGenf, SwitzerlandGeneral descripon on research quesons, objecves and theorecal framework:The project ”Swiss Federal Survey of Adolescents, Main Indicators ch-x” iniates a long-term monitoring ofSwiss youth in the fields of educaon, polics, and health behaviour. The target populaon are Swiss adolescentswho are 19 years old, and it is planned that the survey will be conducted every four years in order topermit a longitudinal analysis of the Swiss youth.The theorecal framework is interdisciplinary, applying an adapted capability approach (Sen 1985; 1987; 1993;1999), which is a normave paradigm for the evaluaon of individual well-being.. In this project, the capabilityapproach is enriched with concepts of self-percepon, values, and acons in different contexts.Methods/methodology:3.42 Linking Survey and Administrave Data: A Methodological Perspecve IITo be held on July 21, 2011 from: 16:00 to 17:30, in room 321.Coordinated by:• Annelies Blom - Survex - Survey Methods Consulng, Germany• Julie Korbmacher - SHARE, MEA, University of Mannheim, Germany• Joe Sakshaug - JPSM, University of Michigan, United States• Mark Trappmann - Instute for Employment Research, Germany3.42.1 Longitudinal consent paerns and predictors of consent to linkage of survey data in the MillenniumCohort StudyS. Ketende 1 , L. Calderwood 1 , J. W McDonald 11 Centre for Longitudinal Studies, Instute of Educaon, University of London, United KingdomAt each sweep of the four sweeps of the Millennium Cohort Study (MCS), respondents have been asked forconsent to linkage of their survey data to their health, educaon or economic records. Addionally, at wave 2when cohort children were 3 years old, consent was sought from parents or guardians of the cohort children totake a sample of cohort children’s oral fluids. Tate, Calderwood and Dezateux (2006) studied predictors of consentfor data linkage for MCS sweep 1. While consent for linkage was high (92%), the proporons consenngdiffered by mother’s country of residence, age and educaon, with consent less likely among minority ethnicgroup mothers, lone parents and those with higher degrees or no qualificaons. Barngton et. al. (2009)studied the feasibility of collecng oral fluids samples in the home seng...3.42.2 Assessing the Magnitude of Administrave Non-Consent Biases in the German PASS StudyJ. Sakshaug 1 , F. Kreuter 21 JPSM, University of Michigan, United States; 2 University of Maryland, United StatesAdministrave records are increasingly being linked to survey records to enhance the survey data and increaseresearch opportunies. A usual prerequisite to performing exact record linkage is obtaining informed consentfrom respondents. Respondent consent is not universal, however, and several studies have found systemacdifferences between consenng and non-consenng cases across socio-demographic characteriscs. A naturalqueson arises as to whether the selecvity of consent introduces significant biases on key survey andadministrave esmates. Esmang non-consent biases for administrave esmates is complicated by thefact that administrave records are typically not available for the non-consenng cases. We overcome thislimitaon by ulizing data from the first wave of the German Labour Market and Social Security Study (PASS)...

THURSDAY 21 JULY 1633.42.3 Requesng Consent to Link Survey Data to Administrave RecordsJ. Pascale 11 US Census Bureau, United StatesAdministrave records have begun to play a key role in survey research and, while policies regarding consentare sll in flux, there is general agreement that research is needed on how to request consent from respondentsto link their survey data with administrave records. Previous research found that 26 percent of those iniallyopposed to data sharing shied their posion when prompted with arguments about potenal improvementsin accuracy and reducons in cost (Singer and Presser, 1996). In order to take these findings further, in thespring of 2010 a field experiment was carried out by the US Census Bureau which included three panels, eachpresenng a different raonale to the respondent for data linkage: improved accuracy, reduced costs, and reducedrespondent burden. Somewhat contrary to expectaons, there was no stascally significant differencein consent rates across the three versions of the consent queson...3.42.4 Stascal Matching of the German Ageing Survey and the Sample of Acve Pension Accounts as aSource for Analysing Life Courses and Old Age IncomesJ. Simonson 1 , L. Romeu Gordo 1 , N. Titova 11 German Centre of Gerontology, GermanyThe paper examines the combinaon of the German Ageing Survey with the Sample of Acve Pension Accountsvia stascal matching as an example for linking survey and administrave data. Stascal matching isa technique increasingly applied for combining informaon from different data sources when a record linkagebased on a unique idenfier is not possible due to confidenality restricons or aempts to avoid the arionrate linked to the informed consent requirements.3.43 Challenges in Educaonal Surveys IIITo be held on July 21, 2011 from: 16:00 to 17:30, in room 315.1.Coordinated by:• Hans Kiesl - University of Applied Sciences Regensburg, Germany• Alina Matei - Instute of Stascs, University of Neuchatel; Instute of Pedagogical Research and Documentao,Switzerland• Leyla Mohadjer - Westat, United States3.43.1 What is the appropriate me frame for measuring the frequency of educaonal acvies?S. Porter 1 , A. Ryder 11 Iowa State University, United StatesQuesons about the frequency of behaviors are common in educaonal surveys, as researchers seek to understandhow oen students engage in posive educaonal behaviors. Surprisingly lile research has beenconducted on the appropriate me frame to use, in either the educaonal literature or the broader surveymethods literature (Chang& Krosnick, 2003). This is due to the difficulty of validang the frequency quesons; generally, data externalto the survey are needed.3.43.2 Measuring the quality of teaching – A crical analysis of students’ evaluaons of teaching effec-venessS. Schief 1 , C. Young 11 University of Fribourg, Switzerland

164 THURSDAY 21 JULYThe evaluaon of teaching has a long history. Since the 1960s universies in the USA have measured so-calledteaching effecveness by using student evaluaon. Student evaluaon had its breakthrough in Switzerland andin most of Europe with the implementaon of the so-called Bologna system. In the Bologna declaraon of 1999,the European ministers of educaon agreed on a ”promoon of European co-operaon in quality assurancewith a view to developing comparable criteria and methodologies”. Since then, almost every university hasdeveloped ways of measuring the quality of teaching by means of students’ evaluaons. Most of these shortquesonnaires are based on assessments of the skills and the behaviour of the lecturer. Teaching outcomesare oen measured by simply asking for the overall sasfacon with the course or a similar queson...3.43.3 Addional samples with balancing or overlapping condions and given inclusion probabilies inthe framework of PISA.M. Chrisne 2 , T. ROCHER 11 Depp (Assessment, forecasng and performance directorate, Ministry of educaon), , France; 2 INSEE, FranceThis paper provides a theorecal frame and methods to solve a problem which occurs as soon as a first samplehas been drawn at a given me and that one intends later to draw a 2nd sample in an updated samplingframe, linked in a way with the 1st one, but without any possibility of changing the condions or results ofthe drawing of the former sample. The origin of this issue lies in PISA surveys (Programme for internaonalstudent assessment): the next cycle 2012 will be on the same main topic as in 2003 and it will be necessary tomake comparisons between both surveys. One of the ways to perform it is to build the 2012 sample of schoolswith overlapping condions with the 2003 sample. But it is also necessary to have the best representavenessfor the new sample. This one can be met introducing balancing condions when the new sample is drawn.Other constraints should be prescribed (fixed size, given inclusion probabilies…)...3.43.4 How Best to Collect Secondary Educaon Course-Taking Data: Student Surveys vs. Transcript AnalysisS. Kramer 1 , R. Bernotsky 3 , J. Osgood 3 , N. Wolff 21 Math Science Partnership of Greater Philadelphia, United States; 2 Arcadia University, United States; 3 WestChester University, United StatesOne of the crical issues for understanding student success in STEM (Science, Technology, Engineering andMath) disciplines in college is understanding the impact of their secondary school preparaon for college.Researchers have found that transcript analysis is a useful tool for analyzing student experiences (Adelman1999; Lee, Croninger, and Smith 1997), however, transcript data can be difficult to obtain due to privacy laws.Transcripts can be obtained with student permission, but it is difficult to get such permission since it requiresstudents to provide a social security number. Surveys are an alternave method of collecon, but there areconcerns about the accuracy of survey responses that are, by definion, self-reported data (Cole and Gonyea2010).3.44 Methodological and technical challenges in surveys of immigrant-backgroundand ethnic minority groups IITo be held on July 21, 2011 from: 16:00 to 17:30, in room 413.Coordinated by:• Mónica Méndez - Centro de Invesgaciones Sociológicas (CIS), Spain• Laura Morales - Instute for Social Change, University of Manchester, United Kingdom3.44.1 The EDCA-Survey: How Important are Sampling Technique, Survey Language and Interviewer Backgroundfor Surveying Migrants?M. Schaeffer 1

THURSDAY 21 JULY 1651 Wissenschaszentrum Berlin für Sozialforschung (WZB), GermanySurveys on migrants face methodological, logisc and financial challenges. The Ethnic Diversity and ColecveAcon Survey (EDCAS) includes two oversamples of migrants, one of migrants in general (26%) and one of Turkishmigrants in parcular (14%). In addion, it has a cross-naonal design since it was conducted in Germany(n=7500), France (n=1000) and the Netherlands (n=900). From the EDCA-Survey we can learn about many ofthe challenges of surveying migrants. First, the EDCA-Survey was conduced in the three host-country languages(German, French and Dutch) but also in Turkish. Does the survey language affect the populaons sampled anddoes it affect important atudinal outcome variables? Second, migrants in general were oversampled via thecostly but assumably more representave random digit dialing in combinaon with an extensive screeningprocedure...3.44.2 Survey research among ethnic minority groups in the absence of a sampling frameD. Jacobs 1 , B. Herman 1 , N. Vanparys 11 Université Libre de Bruxelles (ULB), BelgiumIn this paper we wish to compare two strategies for survey research among ethnic minority groups in the absenceof a sampling frame: a) random walk and focused enumeraon for face-to-face surveys and b) namerecognion for telephone based interviewing. Both methods have been recently used in Brussels focussingon the same targeted ethnic minority groups (cizens of Moroccan and Turkish origin). We wish to assessadvantages and drawbacks of both strategies (also focussing on cost effecveness) in trying to achieve representavity.3.44.3 Onomasc Methods: an approach to avoid under-representaon of immigrants in surveysE. Liebau 11 SOEP - DIW Berlin, GermanyFor longitudinal studies, covering immigrants in a representave way and providing a sufficient number ofcases poses major challenges. The reasons for this are diverse: connuous immigraon that changes the underlyingpopulaon, lower response and higher arion rates among immigrants in comparison to the restof the populaon. To counter this under-representaon several methods were applied in the SOEP. Immigrantsampling via informaon from the registry office (Sample B) and the screening procedure (Sample D) arethe most important ones. However, for the latest sub-Sample I the onomasc method was employed. Theonomasc method derives peoples’ origin from their names...3.44.4 Surveying Romanian migrants in Madrid areaV. Grigoras 2 , A. Rusu 2 , M. Serban 1 , P. Tufis 11 The Research Instute for Quality of Life, Romanian Academy, Romania; 2 University of Bucharest, Departmentof Sociology, RomaniaBecoming an immigraon space in the mid 80’s, Spain is now one of the top ten world’s countries in terms ofinternaonal migrant stock. At the end of 2008, according to the Spanish populaon register, Romanians werethe populaon to contribute most to the immigrant stock.3.45 European Values Study - methodological and substanve applicaons IIITo be held on July 21, 2011 from: 16:00 to 17:30, in room 410.Coordinated by:• Wolfgang Jagodzinski - GESIS, Germany• Ruud Luijkx - Tilburg University, Netherlands

166 THURSDAY 21 JULY3.45.1 To what extent are values and atudes shaped by religion?Y. Esmer 2 , W. Jagodzinski 11 GESIS, Germany; 2 Bogazici University, TurkeyIn recent public discussions values and atudes are oen seen as a result of cultural or, more specifically,religious influences. Accordingly, polical and social orientaons are interpreted as an outcome of Islamic,Orthodix, Catholic, or Protestant cultural tradions. The European Values Survey 2008 offers a unique oppurtunityfor tesng this hypothesis against the alternave view that those orientaons are much more influencedby the societal development and the funconal requirements of sociees at given levels of human development.3.45.2 Life Sasfacon of Women in the Stress Field of Social Norms and Personal Atudes towards GenderRolesK. Harzeneer 11 GESIS – Leibniz Instute for the Social Sciences , GermanyAlthough in the course of the last decades prosperity within European countries increased and gender inequalitydecreased, a gender gap in reported global life sasfacon remains. However, counter-intuively,the discrepancy between women and men in relaon to life sasfacon is not narrowing with rising economicperformance.3.45.3 ’Atheist-Proporon-Development’-Typology. Generaon of a Cross-European Typology based onAPC and Cluster AnalysisJ. Kampmann 11 GESIS - Leibniz Instute for the Social Sciences, GermanyRecent empirical surveys suggest that Europeans claiming no religious or spiritual preference, i.e. Atheists,constute a substanal minority in today’s European sociees, yet this group has not been extensively studiedby social sciences.3.45.4 Motherhood atudes in European countries - conflict of woman roles in different culturesI. Rimac 1 , J. Ogresta 1 , L. Zorec 11 Faculty of Law University of Zagreb, CroaaThe main purpose of this paper is to determine impact of different roles in woman’s life as compeng roles tomotherhood. The focus is put on impact of economic status, educaonal goals and professional aspiraons asparally interfering acvies to motherhood. According to goals of analysis the female subsample of EVS datais analyzed taking into account variety of countries covered by EVS.3.46 Mullevel analysis in comparave research IIITo be held on July 21, 2011 from: 16:00 to 17:30, in room 315.Coordinated by:• Bart Meuleman - University of Leuven, Belgium• Elmar Schlueter - University of Cologne, Research Instute for Sociology, Germany3.46.1 Applying a Mullevel Perspecve in Persuasion ResearchF. Mangold 1 , M. Schenk 11 University of Hohenheim, Germany

THURSDAY 21 JULY 167The presentaon discusses how mullevel designs can contribute to overcoming shortcomings in persuasionresearch. Persuasion researchers agree that the effects of persuasive smuli on recipients’ atudes heavilyvary depending on numerous boundary condions, wherefore different effect paerns occur in differentpopulaon segments (comparave perspecve). Yet, predicons of persuasion effects are possible only to alimited extend, because there is a substanal lack of evidence on the strength of their boundary condions’ influencesand interacon effects. In order to overcome this deficit, we introduce a mullevel design that differsfrom the prevailing mullevel modeling techniques. These are typically applied to outcomes at a person-level,whereby individuals (level 1) are nested in social enes (level 2)...3.46.2 Esmang the Impact of Expected Unemployment Benefit Profiles on the Duraon of UnemploymentSpells: a mullevel analysis of nine EU CountriesI. Salagean 1 , J. Ray 21 CEPS/INSTEAD Luxembourg, Luxembourg; 2 Nancy 2 University, FranceThis study examines the role that expected unemployment insurance benefits play in explaining the duraon ofindividual unemployment spells. Based on a detailed legal review of benefit rules applicable in nine Europeancountries during the second part of the 1990s, we define profiles of expected unemployment benefits whichindicate what levels and duraons of unemployment insurance benefits are expected by workers enteringunemployment in each of these countries. The impact of the expected benefit profiles on the hazard of exingunemployment is esmated by applying a discrete-me two-level event history model to the data in the eightwaves of the European Community Household Panel. Our key explanatory variable is an innovave indicatorof the generosity of unemployment benefits, inially proposed by Ray et al (1986), which sums the discountedmonthly replacement rates expected over the whole compensaon period...3.46.3 Analyzing Contextual-Level Outcomes in Mullevel ModelsD. Becker 1 , W. Breustedt 2 , C. Zuber 31 Cologne Graduate School in Management, Economics and Social Sciences, Germany; 2 University of Duisburg-Essen, Germany; 3 Chair in Comparave Polics, University of Cologne, GermanyIn contemporary social sciences, mullevel analysis has become a commonly used stascal technique forregressing a level-one variable on both level-one and level-two predictors. However, if the dependent variableis located on a contextual-level such as schools or naons, no similarly established procedure is at hand.While many studies in comparave research with level-two outcomes ’solve’ this problem by aggregang alllevel-one predictors to their respecve group means and use these group means as predictors for the level-twodependent variable in convenonal OLS regression (Luna/Zechmeister 2005; Lim/Bond/Bond 2005), Croon andvan Veldhoven (2007) have shown that this method leads to biased esmates.3.46.4 Mind the Gap! Geographic Transferability of Economic Evaluaon in HealthC. Boehler 1 , J. Lord 1 , M. Buxton 11 Health Economics Research Group (Brunel University), United KingdomTransferring cost-effecveness informaon from one domain to another offers the potenal to invest analyticalresources more efficiently. However, it is difficult for decision-makers to know when they can rely oncost-effecveness evidence produced for another context. This paper explores the transferability of economicevaluaon results produced for one geographic area to another locaon of interest, and outlines an approachto idenfy factors to predict when this is appropriate.3.47 Construcon of Answer Scale Formats in Quesonnaires IIITo be held on July 21, 2011 from: 16:00 to 17:30, in room 412.Coordinated by: Natalja Menold - GESIS - Leibniz Instute for the Social Sciences, Germany

168 THURSDAY 21 JULY3.47.1 A Comparison of Branching Response Formats with Single Response FormatsR. Thomas 11 ICF Internaonal, United StatesUnlike single response formats that use a single dimension to measure atudes, the branching technique ofatude measurement separates the rang task into two different sequenal tasks for measuring bipolar atudes:direcon of atude (posive versus negave) and intensity of atude (strength). Some research (Malhotra,Krosnick, and Thomas, 2009, POQ) has indicated that superior outcomes are obtained with branchingthat has more arculaon (3 categories rather than 2 categories) on the endpoints rather than in the middle.However, no comparisons were made with single response formats and, further, the nature of the midpointmay have affected the results. We report here 2 studies comparing a number of branching alternaves, includinga consideraon of alternave middle responses, with a series of single response measures to determinerelave efficacy of the scales...3.47.2 Cross-country Comparisons: Effects of Scale Type and Response Style DifferencesR. Thomas 11 ICF Internaonal, United StatesResponse categories may be used differently as a result of ethnic background or country of residence. Whenmaking comparisons between countries, there are a number of factors (e.g. mode, demographics, etc.) thatmust be equated before we can aribute differences due to culture and not other factors. Scale polarity isone issue that can cause some differences between countries – bipolar scales may somemes be inappropriatelyrendered as unipolar scales (and vice versa) in translaons. In this study, we compared scale variants(e.g. unipolar versus bipolar) and extent of semanc anchoring (fully anchored versus end anchored scales) inleading to differences between countries. This experiment had 36,938 respondents from 6 different Europeancountries complete a web-based survey on atudes and behaviors toward different acvies...3.47.3 Forced choice scales and Likert scales: a study about two different ways to ask quesonsD. La Sala 11 Università degli Studi di Firenze , ItalyThe paper presents the results of part of a broader study, aimed at evaluang the capacity of two differenttechniques, Likert scale and forced choice scale, in measuring the same traits (atude towards violaon oflaw, environment respect and the death penalty.3.48 Analyzing Survey Data - New Methodologies ITo be held on July 21, 2011 from: 16:00 to 17:30, in room 340.Coordinated by: Jean-Marie Le Goff - University of Lausanne, Switzerland3.48.1 Sequenal Logit Models with Unobserved Heterogeneity: Latent Class Esmators for Large andSmall SamplesH. Tam 11 Chinese University of Hong Kong, ChinaThe sequenal logit model or the sequenal response model has long been an influenal framework for modelingsequenal transions, decisions, or any other nested events. For instance, it has been the dominantmodeling framework for the study of inequality of educaonal opportunity ever since the seminal works ofMare (1980, 1981). But convenonal applicaons of the model to esmate causal effects or structural parametersare known to be biased by the ubiquitous presence of unobserved heterogeneity. In a widely citedpaper, Cameron and Heckman (1998) show that sequenal logit models are generally biased and unidenfied.

THURSDAY 21 JULY 169Researchers oen unknowingly achieve idenficaon by making strong and arbitrary assumpons, especiallywhen the data are based on cross-seconal non-experimental surveys in which me-varying covariates areunavailable...3.48.2 Specificaon and Esmaon of Rang Scale Models – with an Applicaon to the Determinants ofLife SasfaconR. Studer 1 , R. Winkelmann 01 University of Zürich, SwitzerlandA rang variable represents the extent to which a quality (e.g. health, risk, accordance) is present. A leadingexample in economics is self-rated well-being. Empirical research into the determinants of rang variablesoen uses data from household (panel) surveys that include a single-item 11-point queson. The linear regressionmodel or ordered latent models are employed for regression analysis of rang variables. However,both esmaon methods are not sasfactory.3.48.3 A Simple Methodology to Increase Discriminang Power of Test Scores with an Applicaon in MentalHealth ResearchJ. Kampen 2 , H. Tobi 3 , J. Barre 11 Wageningen University and Research Centre, Netherlands; 2 Wageningen University and research Center,Netherlands; 3 University of Wageningen, NetherlandsRecently, discriminang power of health indices and tests has received increased aenon. The discriminangpower of a test refers to the possibility to make meaningful differenaons between individuals based on theirtest scores. The renewed aenon for discriminang power is the result of the work of Hankins on Ferguson’sCoefficient of Test Discriminaon delta. A Likert score ignores all informaon on response paerns for individualitems. For instance, in a four item test, the response paerns ”0011” and ”0101” both produce the sameLikert score 2 (in this four item example, 6 different response paerns produce Likert score 2). Consequently,the discriminang power of Likert scores is low. To increase the number of possible outcome values of a test,the informaon contained in the specific answering paern may be used, whereas Likert scores do not usethis informaon...

170Day 4Friday 22 July4.1 Usability of Web-Based Surveys ITo be held on July 22, 2011 from: 09:00 to 10:30, in room 315.Coordinated by: Emily Geisen - RTI Internaonal, United States4.1.1 Detecng the unconscious: Eye Tracking in pretesng quesonnairesS. Tries 11 Federal Stascal Office, GermanyFor several years, the Federal Stascal Office (FSO) has been working on the systemac implementaon ofquesonnaire tesng. A pretest laboratory was established in 2007 and complemented by an eye tracker in2009. Quesonnaires of online surveys are now increasingly evaluated by qualitave tesng methods andredesigned to reduce the burden for respondents and to increase data quality of official stascs.4.1.2 Usability tesng of market research quesonnairesA. Johnson 2 , R. Coombe 2 , C. Jarre 11 Effortmark Limited, United Kingdom; 2 Kantar Operaons, United KingdomKantar is one of the world’s largest insight, informaon and consultancy networks. One of its core tools isthe online market research quesonnaire administered to a panel of respondents. Ensuring that those respondentshave an enjoyable experience is a key Kantar objecve, and usability tesng with respondents hasgrown in its importance for driving improvements in the consumer experience of online surveys.Usability tesng to explore ways of asking quesons4.1.3 Image effects on online survey respondentsB. Marinica 11 University of Bucharest, RomaniaWEB surveys through online panels have become an important data collecon mode in survey research. AlthoughInternet is rapidly penetrang more and more households, online panels face a delicate problem, ofpanel arion as well as turning respondents into professionals. Consequently, online panel providers striveto reduce as much as possible these phenomena using different approaches: implemenng fraud deteconalgorithms (straight lining, digital fingerprinng), data quality modules, web layout enrichment (web designenhancements though using various HTML and web2.0 elements: background pictures, fonts& colors, dynamic web pages, etc). On one hand, all these together aim to detect bad respondents in orderto be excluded from further invites and on the other hand they strive to increase good respondent’s loyalty asmaking web quesonnaires more aracve to them...

FRIDAY 22 JULY 1714.1.4 Le feels right! The opmal posion of answer boxes in Web surveysT. Lenzner 2 , L. Kaczmirek 2 , M. Galesic 1 , M. Merkert 11 Max Planck Instute for Human Development, Germany; 2 GESIS - Leibniz Instute for the Social Sciences,GermanyThe literature on human computer interacon consistently stresses the importance of reducing the cogniveeffort required by users who interact with a computer in order to improve the experience and enhance usabilityand comprehension (e.g., Shneiderman, 1998). Applying this perspecve to Web surveys, quesonnairedesigners are advised to strive for layouts that facilitate the response process and reduce the effort requiredto select an answer. In this paper, we examine whether placing the input fields (i.e., radio buons or checkboxes) to the le or to the right of the answer opons in closed-ended quesons enhances usability and facilitatesresponding. First, we discuss two opposing principles of how respondents may process these quesonsin Web surveys, one suggesng placing the answer boxes to the le and the other suggesng placing them tothe right side of the answer categories...4.2 Weighng of Household Panel SurveysTo be held on July 22, 2011 from: 09:00 to 10:30, in room 315.1.Coordinated by: Mahias Schonlau - RAND, United States4.2.1 Cross Seconal Weights in Household Panel Surveys: Approaches and Condions for their EquivalenceM. Schonlau 1 , M. Kroh 21 RAND, United States; 2 German Instute for Economic Research (DIW) and Humboldt-Universität zu Berlin,GermanyThe computaon of cross seconal weights in household panels is challenging because household composionschange over me. Sampling probabilies of new household entrants are generally not known andassigning them zero weight is not sasfying. Two common approaches to cross seconal weighng addressthis issue: (1) ”shared weights” and (2) modeling or esmang unobserved sampling probabilies based onperson-level characteriscs. We survey how several well-known naonal household panels address cross sec-onal weights for different groups of respondents (including immigrants and births) and in different situaons(including household mergers and splits). We show that for certain esmated sampling probabilies the modelingapproach gives the same weights as ”fair shares”, the most common of the shared weights approaches...4.2.2 Propensity Score Approach in Reducing Bias: Marginal Mean Weighng through Straficaon in aStudy of First-Year Maternal Employment ImpactsY. Im 11 University of Chicago, United StatesMany empirical quesons in social science depend on the causal effects of intervenons or treatments. Thesuperlave model for social science research is the experiment in which individuals are randomly assigned totreatment and control groups. However, many of the causal factors important to theory and pracce are notenrely amenable to experimentaon for ethical, praccal, technical, or financial reasons. When randomiza-on is not feasible in social experiments, the propensity score method is a viable tool for esmang unbiasedtreatment effects. This paper presents a method, marginal mean weighng through straficaon (MMW-S),to address the issue of selecon bias in drawing causal inferences from non-experimental data and to highlightthe usefulness of its applicaon in survey research. MMW-S is a nonparametric adjustment method in whichcausal effects are defined as comparisons between marginal populaon means...

172 FRIDAY 22 JULY4.2.3 The representave research of households based on the data of individual interviewing.A. Veykher 11 Higher School of Economics in St-Petersburg, RussiaThe diverse techniques of household research are usually rather expensive. For many problems we use simplifiedmethod to study households parameters without developing a household panel. 4-5 quesons, added toquesonnaires designed for different studies of general populaon, and a special method of weighng surveydata allow us to get representave data on households as a supplemental result. We use this methodology insurvey in St. Petersburg for more than 8 years.4.3 Experimental Methods in Survey Research ITo be held on July 22, 2011 from: 09:00 to 10:30, in room 415.Coordinated by:• Heiko Rauhut - ETH Zürich, Switzerland• Ben Jann - University of Bern, Switzerland• Thomas Hinz - University of Konstanz, Germany• Prof. Dr. Stefanie Eifler - University of Halle-Wienberg, Germany4.3.1 The Impact of Queson Context on the Measurement of An-Semic Atudes: An ExperimentalSurveyH. Beyer 2 , I. Krumpal 11 Universität Leipzig, Germany; 2 Georg-August-Universität, Göngen, GermanyRegarding the sensive character of an-Semic atudes, it is uncertain whether they are reported truthfullyin surveys. It is assumed that respondents answer in a socially desirable manner, i.e. underreport an-Semicatudes. Using an experimental survey, we demonstrate that the cognive acvaon of an an-Semic groupnorm elicits more socially undesirable answers, thus increasing the prevalence of self-reported an-Semism.To acvate group norms, we make use of context effects resulng from an experimental manipulaon of thequeson order. Our research combines methodological experiments and applied research problems. Wedemonstrate that experimental survey methods can be used to address substanve research quesons, i.e.to invesgate the causal impact of group norms on the communicaon of socially undesirable an-Semicatudes in a survey situaon...4.3.2 A Validated Module to Measure Risk, Time, and Social PreferencesA. Becker 2 , A. Falk 11 University of Bonn, Germany; 2 Bonn Graduate School of Economics, GermanyIn contrast to obtaining preference measures by looking at data generated in economic experiments, surveydata lacks incenve compable decision-making. Problems arising from this fact include self-serving and hypothecalbias, strategic moves, or simple lack of aenon. Thus, there is concern regarding the behavioralvalidity and reliability of survey measures. Our goal is to construct a preference module, which includes itemsthat are shown to predict behavior in incenvized laboratory experiments. We observe data from more than300 subjects, who take part in real-stakes experiments and answer a set of survey items. Incenve compableexperimental measures were obtained by conducng standard economic experiments, including the trustgame and choices between safe opons and loeries...4.3.3 Combining experimental and survey data. An examinaon of potenal bias.G. Garcia Albacete 1 , T. Dannwolf 2 , C. Schmidt 11 University of Mannheim, Germany; 2 MZES, University of Mannheim , Germany

FRIDAY 22 JULY 173Behavioral quesons in survey research do not control the context while economic lab experiments pose ques-ons regarding external validity. Our study combines both methodologies in order to answer substanve ques-ons. Polical behavior research usually relies in survey quesons measuring past behavior. This makes it difficultto address theorecal quesons regarding the effect of the specific context and individuals ’movaons.Therefore we designed an economic experiment that permied to control the environment and induce themovaons. Main independent variables were measured by a short version of the European Social Survey.4.4 Mobilise! Incorporang Mobile Phones in Survey Research IITo be held on July 22, 2011 from: 09:00 to 10:30, in room 318.Coordinated by:• Marek Fuchs - Darmstadt University of Technology, Germany• Femke De Keulenaer - Gallup Europe, Belgium• Robert Manchin - Gallup Europe, Belgium4.4.1 The effect of field period on response in a mobile phone surveyP. Vicente 1 , E. Reis 11 ISCTE-Lisbon University Instute, PortugalMobile phone ownership is approaching saturaon in Portugal as in many other European countries. Not surprisingly,survey companies are starng to make use of mobiles phones to collect survey data. Mobile phoneshave specific characteriscs that are disnct from other modes of data collecon – they are personal devices,carried at all mes and to all places – and as such can influence the representaon of various respondentsgroups and the quality of responses in a mobile phone survey.4.4.2 Pre-Call Validaon of Random Digit Dialing Cell Phone Numbers: A Field ExperimentT. Kunz 1 , M. Fuchs 11 Darmstadt University of Technology, GermanyThe present experiment is designed to test mechanisms which can be used to validate random digit cell phonenumbers prior to field work: (1) Cell phone number validaon services provide verificaon in real me byperforming a Home Locaon Register (HLR) lookup. (2) Text messaging services can also be used for cell phonenumber validaon by using a text message gateway and sending bulk text messages from the Internet to cellphones.4.4.3 Differenal call scheduling in dual frame telephone surveysF. De Keulenaer 1 , R. Manchin 11 Gallup Europe, BelgiumIt is well established that me-of-day and day-of-the-week are important factors to consider when calls arescheduled in telephone surveys. Calling schedules for landline surveys are generally skewed towards eveningand weekend calling since this is when most people are available at home. Research shows that many peoplehave their mobile phones on all day; as such, mobile phones provide a direct communicaon link to respondentsat virtually any me. Nevertheless, most mobile phone surveys use the same ”skewed” calling schedulesthat are typical of landline surveys. In this study, we want to explore how call scheduling in mobile phonesurveys can be opmized in order to increase contact and compleon rates. We will endeavor to answer thisqueson by analyzing contact history data of several landline, mobile phone and dual frame surveys conductedby Gallup in the past five years in many European countries...

174 FRIDAY 22 JULY4.4.4 How to ask about fixed/mobile telephone usage? Impact of queson wording and analysis of correspondingpopulaon segmentsA. Slavec 1 , V. Vehovar 21 University of Ljubljana, Slovenia; 2 University of Ljublana, SloveniaDue to non-coverage in fixed telephone samples the dual frames combining mobile and fixed telephone numbersare increasingly used in telephone surveys of general populaon. However, with dual frames we need toesmate the domains related to single and mixed telephone usage. In the simplest case, we usually have threesegments: fixed-telephone-only users, mobile-phone-only users and those in the overlap. The exact allocaonheavily depends on phone use definion. There, at least two criteria can be used: possession (person has amobile/fixed phone) and availability (can be reached on a mobile/fixed phone). In pracce, various wordingscan be applied for possession and for availability queson. In this paper, we first explored contemporary dualframe research. Based on that, we generated a meta-overview of queson wordings. Next, we address theproblem of the corresponding sample and populaon shares for these segments...4.4.5 Opmal allocaon in dual frame telephone samples. Modeling costs and differenal response patternsfor sample allocaon in strafied samples.S. Pasadas 1 , M. Trujillo Carmona 11 Instute for Advanced Social Studies, SpainCosts are one of the main concerns when dealing with dual frame samples of landline and mobile telephonelines. As a result of a survey conducted among U.S. survey organizaons known to have conducted dual frametelephone surveys, AAPOR Cell-phone Task Force found that ”the cost per compleon in a US RDD cell phonesurvey is most oen at least twice that of a compleon in a US RDD landline survey, and under certain designcondions can be three or four mes as expensive” (AAPOR, 2010: pg. 11). Although in Europe we don’tface most of the legal and commercial constraints exisng in the US, mobile telephone surveys are sll moreexpensive than landline surveys. Kuusela et al. report an increase of 42% in telephone costs of interviewingin Stascs Finland as a result of including mobile telephone lines in telephone samples (Kuusela et al. 2008:pg. 104)...4.5 What is meant by mode effect on measurement? IITo be held on July 22, 2011 from: 09:00 to 10:30, in room 319.Coordinated by:• Steven Hope - Naonal Centre for Social Research - NatCen, United Kingdom• Gerry Nicolaas - Naonal Centre for Social Research - NatCen, United Kingdom4.5.1 Telephone vs computer assisted self-interviewing in surveys on drug useF. Beck 1 , S. Legleye 2 , R. Guignard 11 INPES, France; 2 INED, FranceThe literature on data collecon modes proved that self-administered quesonnaires provide higher prevalencesthan face-to-face or telephone interviews in surveys on drug use. Although there are important genderand age differences in drug use reports, it seems crucial to check whether this result is true in all age andgender groups.4.5.2 Effects of elapsed interview me in Mixed-Mode SurveysS. Bathelt 1 , J. Bauknecht 21 Universität Koblenz-Landau, Germany; 2 University of Stugart, Germany

FRIDAY 22 JULY 175Due to declining response rates and, concomitantly, rising costs in face-to-face (CAPI) surveys, other methodsgain in aracveness. Most of these methods are coupled with severe disadvantages, like ”digital divide”(online surveys), uncertainty concerning the me of the interview (postal surveys) or uncertainty about whogave the answers (in both). By contrast, telephone surveys (CATI) limit costs and fieldwork duraons, andclosely resemble face-to-face interviews, apart from limited interviewer control over respondents´ parallelacvies and the impossibility of issuing show cards. Apart from sampling issues, interviewing mode effectscould yield different results in CATI surveys than in CAPI surveys...4.6 Linking Survey and Administrave Data: A Methodological Perspecve IIITo be held on July 22, 2011 from: 09:00 to 10:30, in room 321.Coordinated by: Jonathan Burton - University of Essex, United Kingdom4.6.1 Pandata Systems: Its Merits and Downsides of Turning Administrave Data into InformaonY. Chun 11 NORC at the University of Chicago, United StatesJoe Waksberg long ago forecast that administrave records are always the answer for the next round of astascal program (Citro, 2010). Administrave data help control rising data collecon costs, reduce surveyrespondent burden, and refine the esmaon process with parcular aenon to nonresponse adjustmentwith auxiliary variables, yet without a rigorous theorecal navigator.4.6.2 Consenng to Health Record Linkage: Evidence from the Brish Household Panel SurveyG. Knies 2 , E. Sala 1 , J. Burton 31 University of Milano Bicocca, Italy; 2 Instute for Social and Economic Research, United Kingdom; 3 Universityof Essex, United KingdomThe Brish Household Panel Survey (BHPS) is the first long-running longitudinal survey with a non-medical focusand a sample covering the whole age range to have asked for permission to link to a range of administravehealth records. This presentaon determines whether informed consent led to selecon bias on the BHPS andreflects on the value of the BHPS linked with health records for epidemiological research. Mulvariate stas-cal analysis is performed on informaon from the BHPS. Permission to add health records was asked fromadult parcipants (aged 16+), and pro persona of adults responsible for children, at the end of the personalinterviews to the 18th round of annual interviews of the BHPS (2008). The consent rate across the BHPS Wave18 sample is 40 percent, i.e., much lower than typically is the case on previous medical studies. There are fewdifferences between those who consent and refuse...4.6.3 Exploing Administrave Data to explore Job Churn in the Irish Labour MarketJ. Dunne 11 Central Stascs Office, IrelandThe paper will cover experiences from the Job Churn Explorer project at CSO. The project adapts and developsthe underlying methodology outlined to date to the situaon in Ireland to provide a detailed insight into thedynamics of job churn and its components as Ireland entered the current recessionary period. The analysisdatasets used are derived from linking the following three sources: business register, employer tax returns,social protecon records. The comprehensiveness of the resulng analysis dataset containing aributes onboth workers and enterprises provides for significant new opportunies to inform policy and decision makingwith respect to the labour market. The presentaon will also graphically present some of the analysis at botha worker and enterprise level to demonstrate the potenal of this new informaon...

176 FRIDAY 22 JULY4.7 Analyzing social change with repeated cross-secons IITo be held on July 22, 2011 from: 09:00 to 10:30, in room 303.Coordinated by:• Tilo Beckers - University of Düsseldorf, Germany• Christof Wolf - GESIS - Leibniz Instute for the Social Sciences, Germany4.7.1 The Impact of Past Events on Current Electoral Behaviour. Age-Period-Cohort Analysis of Czech CommunistParty VotersL. Linek 11 Czech Academy of Sciences, Czech RepublicThe paper analyses the effect of polical generaons on current Czech Communist Party (KSČM) voter support.The goal is to show that past events have had an effect on current vong behaviour without having to bereproduced in me and strengthened through repeated voter behaviour in compeve elecons. The datafrom repeated monthly surveys between 1990 and 2010 are used. The cohort effects are esmated usingcross-classified random-effects models (Yang, Land 2006). The findings regarding the various generaon effectson KSČM support confirm this goal. KSČM support is largely affected by (1.) the Second World War and thesubsequent communist coup, (2.) the military invasion of Warsaw Pact armies under the leadership of theSoviet Union in 1968 and (3.) the transion to democracy in 1989...4.7.2 Strategies to Overcome the APC Conundrum. The Age-Stability Hypothesis in Postmaterialist Values,Atudes to Homosexuality and Religious PraccesR. Tormos 11 Universitat Autónoma de Barcelona, SpainResearch on values and atudes tends to underline the importance of generaonal factors, following the agestabilityhypothesis and the impressionable-years model of learning. Formave experiences shape each agecohort,and social change takes place progressively due to the force of generaonal replacement. Examples ofit are the ”socializaon hypothesis” in Inglehart’s postmaterialist values and the ”age-stability hypothesis” inAlwin and Krosnick’s atudes research. However, if we carefully analyze repeated cross-secon data, we canidenfy relevant value change within each generaon over me as well. Some values and atudes follow alifeme model of learning, instead of the pure impressionable-years one. To test the impact of the historicalperiod in comparison to generaon and life cycle effects is necessary to face the APC conundrum...4.7.3 Generaon, Period and the Decline in Trust among Black and White AmericansR. Wilkes 11 University of Brish Columbia, CanadaAmericans have less generalized faith in others than ever before. Explanaons of this decline draw aenon tothe role of generaonal replacement and to period effects stemming from macro-level economic and policalchanges. In this paper I use cross-classified random effects models to test whether the decline thesis appliesequally to black and to white American’s trust levels. The data come from 23 repeated cross-secons of U.S.General Social Survey data. The models used in the paper allow the researcher to simultaneously assess theeffects of generaon and period by considering them as random level 2 effects within a mul-level framework.The results show that while the war baby generaon of whites (1935-39, 1940-44) was more trusng thanother generaons, trust does not differ significantly across generaons of blacks...4.7.4 Analyzing social and media change with cross-classified random-effects APC modelsD. Becker 1 , J. Hagenah 2

FRIDAY 22 JULY 1771 Cologne Graduate School in Management, Economics and Social Sciences, Germany; 2 University of Köln, GermanyAlthough communicaon research found a whole bunch of posive effects of newspaper usage on individualcharacteriscs such as verbal abilies or social capital (Glenn 1994, Schmi-Beck 2008), a longitudinal analysisof the determinants of a parcular range of media products based on individual-level data appears to bemissing.4.8 Methodological and technical challenges in surveys of immigrant-backgroundand ethnic minority groups IIITo be held on July 22, 2011 from: 09:00 to 10:30, in room 413.Coordinated by:• Mónica Méndez - Centro de Invesgaciones Sociológicas (CIS), Spain• Laura Morales - Instute for Social Change, University of Manchester, United Kingdom4.8.1 Measurement errors in a quantave survey on migrants. A case study from SalzburgW. Aschauer 11 University of Salzburg, AustriaQuantave surveys in migraon research have to deal with special parcularies and biases. Problems mayarise regarding the definion of the analysed populaon, the sampling procedure and several potenal measurementerrors (e.g. translaon issues, formulang quesons and scale use). Till now there is only lilescienfic research about potenal errors during fieldwork and with regard to the interview condions. Thesepotenal biases are analysed in a case study in Salzburg, where a survey with 187 migrants aending an agencywhich assist foreigners experiencing difficules locang work was conducted. To control for methodologicalproblems, an innovave strategy of fieldwork was adopted. Consultants of the instute filled in a short introductoryquesonnaire together with the migrants (trying to produce a leap of faith)...4.8.2 Survey co-operaon of non-autochtonous populaon in Spain: a comparison of the 2008 and 2010ISSP surveysM. Méndez 11 Centro de Invesgaciones Sociológicas (CIS), SpainSurvey research has pointed out at the generally lower response rate among immigrant/non- autochthonouspopulaon, though this has been challenged by cases in which specific fieldwork strategies addressed at nonautochthonouspopulaon have been put in place.4.8.3 Surveying first- and second-generaon immigrants across European countries: Experiences from twocomparave surveysC. Teney 1 , L. Lessard-Phillips 2 , F. Fleischmann 0 , E. Ersanilli 31 Social Science Research Centre Berlin (WZB), Germany; 2 Instute for Social Change, University of Manchester,United Kingdom; 3 University of Oxford, United KingdomTwo large-scale surveys have recently aimed to fill the void of comparave European survey data on firstandsecond-generaon immigrants. TIES (’The Integraon of the European Second generaon’) surveyed thechildren of immigrants from Turkey, Morocco and/or the former Yugoslavia in major cies of eight Europeancountries. SCIICS (’Six Country Immigrant Integraon Comparave Survey’) surveyed immigrants from Turkeyand Morocco and their descendants in six European countries. Both surveys also included a comparison groupof persons of ’nave descent’, i.e., persons born in the survey country with both parents born in the survey

178 FRIDAY 22 JULYcountry. While the two surveys target similar populaons and mostly cover the same countries, they useddifferent sampling frames and survey methodology...4.8.4 Czech Surveys of Non-EU Labour Immigrants: Tackling the Challenge of Incipient ResearchY. Leonyeva 11 Instute of Sociology, Academy of Sciences of the Czech Republic, Czech RepublicThe presentaon aims to give an image of present state of the art of migraon research in Czech context.Special aenon will be given to two unique surveys of third country naonals. The first survey conducted in2006 was supported by the Ministry of Labour and Social Affairs and was targeng non-EU foreigners (1,011respondents), who stayed in the Czech Republic on the basis of long-term residence permit for the purpose ofemployment. The aim of this survey was to explore the qualificaon and educaonal background of immigrantsand their posion on the Czech labor market. The second survey financially supported by the Czech StascalOffice was realised in late 2010 and was focused on economically acve immigrants from selected countries(1,004 respondents), their incomes, expenditures and remiances...4.9 European Values Study - methodological and substanve applicaons IVTo be held on July 22, 2011 from: 09:00 to 10:30, in room 410.Coordinated by:• Wolfgang Jagodzinski - GESIS, Germany• Ruud Luijkx - Tilburg University, Netherlands4.9.1 The locaon of European Countries on Schwartz’ Value CircleS. Huismans 1 , W. van Schuur 21 Vrije Universiteit Amsterdam, Netherlands; 2 University of Groningen, NetherlandsUsing the Basic Human Values items from the European Social Survey, we apply a new method for measuringboth items and respondents on an ordinal circular scale.4.9.2 Conflict dimension in the European party systems. A comparave study based on European ValuesStudy 2008.O. Knutsen 11 University of Oslo, NorwayIn this paper the European Values Study 2008 will be used to examine conflict dimensions in the party systemsaccording to a policy-defined space approach. Basic social and polical values are used to tap the central policyorientaons. Discriminant analysis will be used to delineate various conflict dimensions on the party system...4.9.3 Can we combine the EVS and ESS to create a me series of health inequalies?R. Jutz 11 GESIS - Leibniz Instute for the Social Sciences, GermanyThe past two decades have seen a rising interest in the cross-naonal study of health inequalies. In contrastto most of the exisng work which has been cross-seconal, uncovering the underlying causes of health inequaliesrequires an analysis of changes over me. For this purpose, the European Value Study (EVS) presentsa unique data source. In three of its four rounds, the EVS contains one central indicator of health: subjecvehealth status. Unfortunately, EVS 1999 misses subjecve health and thus, the cross-naonal me series is

FRIDAY 22 JULY 179interrupted. The European Social Survey (ESS) also offers the opportunity to study health inequalies – however,it covers only a short me period since its first round took place as recently as in the year 2002. Canresearchers use data from both programs to create a longer me series?4.9.4 Value change in Eastern Europe: what is happening there and why?M. Comsa 2 , H. Rusu 11 Lucian Blaga University of Sibiu, Romania; 2 Babes-Bolyai University of Cluj Napoca, RomaniaThere are not many comparave studies dedicated to value change in Eastern European countries. In ourpaper, we employ both longitudinal and cross- naonal analyses to describe and explore whether and, ifso, how and why Eastern European countries have changed their values on the conservasm-openness tochange axis. We considered the support for religiosity, tradional family type, instrumental qualies of work(work ethos) and authority, as value orientaons of a conservave/normave type. Sexual permissiveness,tolerance, orientaons against a non-democrac polical system (democrac orientaons) were regarded asopenness to change values. We are using the available EVS/WVS 1990-2009 data sets from Albania, Belarus,Bosnia-Herzegovina, Bulgaria, Czech Republic, Estonia, East Germany, Hungary, Latvia, Lithuania, Moldavia,Montenegro, Poland, Romania, Russian Federaon, Serbia, Slovakia, Slovenia and Ukraine...4.10 Construcon of Answer Scale Formats in Quesonnaires IVTo be held on July 22, 2011 from: 09:00 to 10:30, in room 412.Coordinated by: Natalja Menold - GESIS - Leibniz Instute for the Social Sciences, Germany4.10.1 The Impact of Direcon and Polarity in Response Scales on Response BehaviorD. Krebs 11 University of Giessen, GermanyAbstractThe applicaon of cognive theory to survey methodology uncovered that answering survey quesons is acognive process consisng basically of four tasks: queson interpretaon, memory retrieval, judgment formaon,and response eding. This paper deals with the laer two tasks in examining the effect of polarity(uni- versus bipolar response scales) within answering categories running either from negave to posive orfrom posive to negave.4.10.2 Choosing the range for a rang scale: an empirical research on the effects of using different scalesD. La Sala 11 Università degli Studi di Firenze , ItalyThe paper concerns some aspects of rang scales used in social research. In parcular, aenon is paid to theinfluence on responses by scales with different ranges.A series of rang scales on different topics (opinions, values, atudes towards legality, democracy, parcipa-on, public policies) were administered to a sample of university students. 7-steps scales (1-7) and 11-stepsscales (0-10) were used for evaluang the same items, by dividing the sample into two comparable sub-groups.The analysis of the distribuon structure of responses in the two sub-groups was conducted by comparingasymmetry, kurtosis, discriminaon capacity of the techniques and by operang a graphical analysis of thedistribuon structure of the responses.4.10.3 Does the direcon of Likert-type scales influence response behavior in web surveys?F. Keusch 11 Instute for Adversing and Markeng Research, Vienna University of Economics and Business, Austria

180 FRIDAY 22 JULYAtude measurement in web surveys mostly relies on requiring respondents to indicate their agreement ordisagreement with each of several items under the same Likert-type scale in a grid format. As it is known thatrespondents do not only aend to the words that convey the quesons but also to the visual language of aquesonnaire (i.e., format and shape of response scales, verbal and numerical labels of scale points, spacing,posioning, and order of response opons) it is essenal to understand how this effects the response process.Although there is no conclusive evidence about the influence of the direcon of extreme point labelling (e.g.,Belson, 1966; Friedman et al., 1993; Salzberger& Koller, 2010; Weng& Cheng, 2000), applying the ”near means related” heurisc (Tourangeau et al...4.10.4 Knowing or guessing? Queson format effects on the esmated level of polical knowledgeD. Johann 1 , T. Schübel 21 Department of Methods in the Social Sciences, University of Vienna, Austria; 2 Department of Polics andPublic Administraon, University of Konstanz, GermanyMuch of the scienfic interest in polical knowledge stems from the debates over the capacity of ordinarypeople to parcipate in democrac life. Polical knowledge has become increasingly central to the study ofpublic opinion, media effects, and polical behaviour more generally (e.g. Bartels 1996; Delli Carpini/Keeter1996; Jerit et al. 2006). However, since praccal and theorecal uncertaines sll surround the measurementof polical knowledge, such conceptual and theorecal quesons will be answered only through careful empiricalstudy. We want to contribute to the methodological debate on esmang levels of polical knowledgeby examining the influence of three survey design factors: (1) guessing-discouragements (Nadeau/Niemi 1995;Mondak/Creel Davis 2001; Krosnick et al...4.11 Innovaons in survey sampling IITo be held on July 22, 2011 from: 09:00 to 10:30, in room 414.Coordinated by:• Siegfried Gabler - GESIS - Leibniz Instute for the Social Sciences, Germany• Seppo Laaksonen - University of Helsinki, Finland4.11.1 Design effect of randomized systemac samplingS. Gabler 11 GESIS - Leibniz Instute for the Social Sciences, GermanyIn stascal pracce, systemac sampling (SYS) is used in many modificaons due to its simple handling. Inaddion, SYS may provide efficiency gains if it is well adjusted to the structure of the populaon under study.However, if SYS is based on an inappropriate picture of the populaon a high decrease of efficiency, i.e. ahigh increase in variance may result by changing from simple random sampling to SYS. In the context of twostagedesigns SYS so far seems oen in use for subsampling within the primary units. As an alternave to thispracce, we propose to randomize the order of the primary units, then to select systemacally a number ofprimary units and, thereaer, to draw secondary units by simple random sampling without replacement withinthe primary units selected. This procedure is more efficient than simple random sampling with replacementfrom the whole populaon of all secondary units, i.e...4.11.2 Possibilies of Sampling Methods in case of Hidden GroupsZ. Szendrő 11 Univesity of Miskolc, HungarySince there is no adequate data of hidden groups’ distribuon it is quite hopeless to carrying out a quantavesurvey even with mulstage strafied probability sampling which involves proporonately these populaons.

FRIDAY 22 JULY 181Thanks to large samples we can reach a good coverage of the country, thus we can expect gaining a representavesample regarding a rare populaon, but this kind of sample seems to fail in most cases. The mulstagestrafied probability sampling is not appropriate for finding non-uniformly distributed minories.4.11.3 Tesng a single mode vs a mixed mode designS. Laaksonen 11 University of Helsinki, FinlandEncement to use web surveys is becoming more common, since this mode is less expensive than the alternavemodes. On the other hand, no-one trusts in web as the only mode, since non-response is expectedto be too high. Hence, a mixed mode strategy is proposed. There are different approach-es to mixed modedesign. We test such an approach that the data collecon will start via web and aer a certain short period,CATI will complete data. Finally, we compare the successfulness of this mixed mode strategy against pure CATI.The reason is that this survey has been compiled using CATI but if a new mixed mode strategy is good, this willbe used in future. So, we have to compare the successfulness of both approaches in a good way. This alsorequires to design a sample well. So, our target populaon is the same as used in the regular survey that doesnot cover the full populaon but the people with telephone access only...4.11.4 Selecon Errors in Random Route SamplesJ. Bauer 11 LMU Muenchen, GermanyRandom Routes are a type of widely used survey. Like most sampling methods they are applied to achieveequally distributed selecon probabilies. The preliminary approach to assess the quality is to compare thedata collected by random route samples with data from reliable sources, such as the German Microcensus.Typically, in these studies the distribuons of demographic variables are compared because they are availablein nearly all surveys. As this concentrates only on a specific set of variables, it is conceivable that other variablesof interest are biased, even if the conducted comparisons are not significantly different.4.12 Analyzing Survey Data - New Methodologies IITo be held on July 22, 2011 from: 09:00 to 10:30, in room 340.Coordinated by: Oriane Sarrasin - Misc - University of Lausanne, Switzerland4.12.1 Tesng a formal theory of images of straficaon: a pilot studyZ. Karpinski 11 Instute of Philosophy and Sociology - Polish Academy of Sciences, PolandThe study reported in this presentaon may be thought of as an exercise in theory-driven research. A studyis said to be theory-driven if empirical procedures used in the study reproduce properes of a theorecalmodel, or replicate specific condions postulated by a theory. Because theory-driven research requires thatcondions specified by a theory are recreated in the research design, theory-driven research usually takesthe form of experiment. In my presentaon, I am trying to show that the idea of theory-driven study can beapplied to survey research as well. As an example, I am using a formal theory of images of social structuresdeveloped by the sociologists Thomas Fararo and Kenji Kosaka. The theory’s focus is on issues that have longbeen invesgated by sociologists, such as percepon of social inequality, self-placement in the social hierarchy,or class idenficaon...4.12.2 Causal Analysis and Social SurveysK. Yang 1

182 FRIDAY 22 JULY1 University of Durham, United KingdomSocial sciensts are constantly striving to achieve causal explanaons for a variety of movaons even when thecondions for making such explanaons are poorly idenfied or sasfied. During the past decade or so bothsocial sciensts and methodologists have called for specific idenficaon of underlying causal mechanismsand cricized the use of linear stascal models for causal explanaons, especially when such models areused on data collected from sample surveys. However, different types of social surveys have produced a hugeamount of data, and more data will be produced in the coming years. These two developments lead to thefollowing queson: what kinds of contribuon can these data make to causal analysis? I answer this quesonby discussing the role of social surveys and the subsequent stascal analyses in two tasks of causal analysis:theorizing causal relaons and measuring causal effects...4.12.3 Validity of images from binary coding tables. Some evidence through Textual Data AnalysisK. Fernández-Aguirre 0 , M. Garín-Marn 1 , J. Modroño-Herrán 11 University of the Basque Country, SpainIn recent years, the main objecve for most praconers is to idenfy interesng structures in the data sets,such as clusters of observaons, or relaonships among the variables. Principal axes methods such as PrincipalComponent Analysis (PCA) and Correspondence Analysis (CA)4.13 General human values, personality characteriscs and response stylesTo be held on July 22, 2011 from: 11:00 to 12:30, in room 318.Coordinated by:• Michael Braun - GESIS - Leibniz Instute for the Social Sciences, Germany• Timothy Johnson - University of Illinois at Chicago, United States4.13.1 Basic Human Values and Response StylesT. Johnson 1 , M. Braun 21 University of Illinois at Chicago, United States; 2 GESIS - Leibniz Instute for the Social Sciences, GermanyOn the basis of the first round of the European Social Survey Johnson et al. (2010) found associaons betweenthe individual- and country-level measures of some of the Schwartz scales and response styles. For example,individual conservaon values were associated posively with middling responses and negavely with extremeresponse styles. The associaon between individual level Conservaon and acquiescence was also posive, ashypothesized. Mean country level Conservaon, however, modified this relaonship in an unexpected direc-on, suggesng that the personal level effects of Conservaon on acquiescence are greater within countrieslower rather than higher in mean Conservaon.4.13.2 Personality, cultural, and socio-demographic predictors of value importance: Invesgang the roleof acquiescent response style in value rangsC. Beierlein 2 , E. Davidov 1 , S. Bokredenghel 2 , B. Rammstedt 21 University of Zürich, Switzerland; 2 GESIS - Leibniz Instute for the Social Sciences, GermanySchwartz’ basic human values explain various atudes and behaviors across cultures. Values are usually assessedusing self-report measures such as the Portrait Value Quesonnaire (PVQ; Schwartz et al., 2001). Likeother self-report quesonnaires, the PVQ is prone to the influence of response styles. For example, socialdesirability was found to be associated with higher rangs for parcular values (Schwartz et al., 1997).

FRIDAY 22 JULY 1834.13.3 Pialls of internaonal comparave research: Taking acquiescence into account.A. Franzen 1 , D. Vogl 11 University of Bern, SwitzerlandAcquiescence can be the source of a serious response bias in internaonal comparave research. We demonstratethis by referring to an example taken from environmental sociology. The effect of wealth on individuals´willingness to pay for environmental protecon is controversially discussed in the literature. Studies analyzingthe Internaonal Social Survey Programme (ISSP) report that individuals in wealthier naons are more concernedabout the environment, while studies using the World Values Survey (WVS) or the European ValuesStudy (EVS) come up with the opposite finding. The puzzle is resolved when the different levels of acquiescenceare taken into consideraon. As it turns out, respondents in poorer naons in Asia and Eastern Europehave higher levels of acquiescence than respondents in richer Western naons...4.13.4 Do people understand the difference between issue extremity and intensity quesons? A behaviourgenec analysis of individual differencesZ. Fazekas 2 , L. Livay 11 Central European University, Hungary; 2 Department of Methods in the Social Sciences, University of Vienna,AustriaProximity theorists argue that the distance between voter and candidate atudes is responsible for votechoice. Also, from a social psychological perspecve, it is expected that proximity of polical atudes betweencandidate and voters weighs more in the electoral decision if the issue is considered to be important.To study these issues, polical sciensts devised survey instruments that enquire about the persons’ posionson various issues and how strongly they feel about these issues. Quickly, a debate emerged concerning theappropriate survey measurement of issue extremity and issue importance. One side argued that the two arehighly correlated and the addional importance queson is a waste of survey space (Niemi and Bartels, 1984).The other side argues that importance and extremity are substanvely different concepts and need to be keptseparate (Krosnick, 1988; Krosnick et al, 1993)...4.14 Separang selecon from mode effects in mixed-mode surveysTo be held on July 22, 2011 from: 11:00 to 12:30, in room 412.Coordinated by:• Barry Schouten - Stascs Netherlands (CBS), Netherlands• Joop Hox - University of Utrecht, Netherlands4.14.1 Using propensity score matching to separate mode- and selecon effectsP. Lugg 2 , G. Lensvelt-Mulders 11 University for Humaniscs, Netherlands; 2 Utrecht University, NetherlandsMode effects are difficult to separate from nonresponse effects in mixed-mode surveys. Mode effects includeacquiescence and social desirability effects due to the presence of an interviewer, recency effects in audiosurveys and primacy effects in visual surveys. However, in everyday mixed-mode surveys we oen cannotdisnguish mode-effects from effects due to differences in nonresponse and coverage that coincide with theuse of mulple survey modes.4.14.2 Evaluang Relave Mode Effects in Mixed-Mode Surveys: Three methods to disentangle seleconand measurement effectsJ. Vannieuwenhuyze 1 , G. Loosveldt 11 Katholieke Universiteit Leuven, Belgium

184 FRIDAY 22 JULYIn order to invesgate the advantage of mixed-mode surveys, selecon effects between the modes should beevaluated. Selecon effects refer to differences in respondent composions on the target variables betweenthe modes. However, esmaon of selecon effects is not an easy task because they may be completelyconfounded with measurement effects between the modes (differences in measurement error). Publicaonsconcerning the esmaon of these mode effects are scarce. This paper provides and compares three methodsthat enable evaluang measurement effects and selecon effects separately. The first method starts from exisngpublicaons where the confounding problem is usually avoided by introducing a set of mode-insensivevariables into the analysis model. However, this paper will show that this method precludes exact esmaonof the effects...4.14.3 Tesng between-mode measurement invariance under controlled selecvity condionsT. Klausch 11 Utrecht University, NetherlandsAs more survey modes have become available, it has ever been an aspiraon in survey methodology researchto test and assure comparability of measurements obtained in different modes. Such research has oen found’mode effects’, which are measurement differenals between modes beyond sampling variaon. Moreover,with the rise of the mixed-mode survey, which combines more than one survey mode in the data collecondesign , the queson of comparability has become even more pressuring.4.14.4 Separang selecon from mode effects when switching from single (CATI) to mixed mode design(CATI / Web) in a health surveyJ. Carstensen 1 , P. Kriwy 1 , G. Krug 3 , C. Lange 21 University of Erlangen-Nuremberg, Germany; 2 Robert Koch Instute, Germany; 3 Instute for EmploymentResearch, GermanyAsking sensive quesons in surveys is prone to mode effects. To analyze the consequences of switching fromsingle to mixed mode designs, we randomly assign survey parcipants to a single mode (CATI) and a mixedmode (CATI and Web-based) survey. Both surveys contain the same set of items based on already establishedinstruments from German health surveys conducted by the Robert Koch Instute. The instruments cover sensiveitems such as subjecve well-being, health behavior and self-reported illness, augmented with quesonsabout confidenality concerns. Because of randomizaon differences in means are informave on the net impactof switching to mixed mode. Applying a Blinder/Oaxaca decomposion method from econometrics to thefield of survey methodology, we are able to separate (1) selecon effects due to differences in non responserates from (2) mode effects due to differences in item non response and measurement error...4.15 Potenals and constraints of weighng to improve survey quality ITo be held on July 22, 2011 from: 11:00 to 12:30, in room 303.Coordinated by:• Stephanie Steinmetz - University of Amsterdam, Netherlands• Kea Tijdens - Erasmus University, Netherlands4.15.1 Propensity Score Weighng: Methods and ApplicaonsM. Bergmann 11 University of Mannheim, GermanyDespite the increasing use of online surveys in polical opinion research, the field sll lacks thorough researchmapping the effects of this polling tool on achieving representaveness. For cause, the two most salient problemsassociated with online research – limited internet coverage and selecvity of respondents – have not yet

FRIDAY 22 JULY 185been convincingly addressed. To compensate for distorons produced by online surveys, weighng oen isseen as a promising correcve soluon. However, the impacts of this technique are controversial since findingsfrom previous research are ambiguous at best (e.g. Taylor 2005; Malhotra&Krosnick 2007). The demonstraon proposed in his paper rests on a two-step approach, using the GermanLongitudinal Elecon Study (hp://www.dgfw.info/gles.php?lang=en) since this data source offers the opportunityto test the impact of different weighng schemes on models of party choice and turnout...4.15.2 Imputaon Models in Panel Surveys: Is Weighng for Sampling Design and Unit Non-ResponseImportant? (with Che Nicole)L. Fumagalli 11 University of Essex, United KingdomIn this paper we consider for the first me a study to evaluate whether imputaon models in panel surveysshould use weights to correct for unit non-response and sampling design. Using the Brish Household PanelSurvey (BHPS), we find that weighng does not seem important when unit and item non-response rates areas low as in the BHPS. On the contrary, when we simulate a doubling of the arion and item non-responserates, we find stascally significant differences between earnings computed using weighted and unweightedimputaon. Although stascally significant, these differences do not seem huge. Hence, the substanalimportance of weights remains sll disputable.4.15.3 Propensity Score Adjustment for a Web-panelG. Wiegand 1 , B. Hulliger 11 University of Applied Sciences Northwestern Switzerland (FHNW), SwitzerlandWhen conducng surveys via the Internet to esmate populaoncharacteriscs one can expect a bias for different reasons. Besides all the typical bias inducingmechanisms found in any survey mode web surveys are suscepble to undercoveragesince the Internet penetraon of the populaon is not exhausve. The bias potenal is even aggravated whena panel of respondents is recruited to be surveyed on several occasions via the web.One way to reduce bias is to apply Propensity Scores Adjustment (PSA). The basic idea is to esmate theresponse propensity with respect to some covariates. To do so allows the deducon ofweights to reduce the bias of a survey. The praccal problem is to idenfy suitablecovariates and to find a good model for the response behavior.4.15.4 Assessing the problem of selecon bias and solving it through weighng - an analysis of selectedWageIndicator countriesS. Steinmetz 1 , P. de Pedraza 3 , D. Raess 2 , K. Tijdens 41 University of Amsterdam, Netherlands; 2 University of Geneva, Switzerland; 3 Universidad de Salamanca,Spain; 4 Erasmus University, NetherlandsThe increasing popularity of web surveys triggered a heated debate about the quality of web surveys for scienficuse. The most obvious disadvantage of web surveys is that they may not be representave becausethe sub-populaon with Internet access is quite specific. Therefore, different weighng techniques, like poststraficaonand propensity score adjustment (PSA) have been proposed, parcularly with regard to nonprobability-basedweb surveys. As parcularly PSA has not been applied tradionally in the field of surveys,the implicaons of these methods sll need to be studied more extensively.4.16 Usability of Web-Based Surveys IITo be held on July 22, 2011 from: 11:00 to 12:30, in room 315.Coordinated by: Caroline Jarre - Effortmark Limited, United Kingdom

186 FRIDAY 22 JULY4.16.1 Results from the Iterave Usability Tesng of the American Community Survey (ACS)K. Ashenfelter 2 , T. Hughes 11 US Census Bureau, United States; 2 U.S. Census Bureau, United StatesThe American Community Survey collects important housing, social, economic data from a sample of approximatelythree million households each year using a sequenal mul-mode approach of mail, telephone, andpersonal visit interviewing. The U.S. Census Bureau conducted a field test in 2011 to assess the potenal implementaonof a fourth data collecon mode: an internet response opon. In preparaon for that field test,in 2010 and 2011 the Census Bureau’s Human Factors and Usability Research Group conducted five roundsof pre-tesng on the new Web version of the American Community Survey (ACS) and one round of tesngon the Puerto Rico Community Survey (PRCS). We will discuss major findings, including parcipant usage ofthe progress indicator and duplicaon of household members on the roster, as well as differences discoveredbetween the findings from the ACS and PRCS tests. We will also present quantave data (e.g...4.16.2 Reliability and Validity of self-reported Burnout in college students: A cross randomized comparisonof pencil-and-paper vs. online administraonJ. Maroco 1 , M. Lucindo Zucoloto 3 , F. Salloume Sampaio Bonafé 3 , P. Crisna Jordani 3 , J. Alvares Duarte BoniniCampos 21 ISPA - Instuto Universitário, Portugal; 2 Universidade Estadual Paulista - UNESP, Brazil; 3 Faculdade de Odontologiade Araraquara. Universidade Estadual Paulista - UNESP, BrazilInternet data collecon is becoming increasingly popular in all research fields dealing with human percepons,behaviors and opinions. Advantages of internet data collecon, when compared to the tradional paper-andpencilformat, include reduced costs, automac database creaon, and the absence of researcher-related biaseffects, such as availability and complete anonymity. However, the validity and reliability of internet gathereddata must be established, in comparison to the usual paper-and-pencil accepted formats, before an inferenalanalysis can be done. In this study, we compared quesonnaire data gathered from the internet with that fromthe tradional paper-and-pencil in a sample of college students. The quesonnaires used were the MaslachBurnout Inventory – Student Survey (MBI-SS), the Oldenburg Burnout Inventory (OBI-SS) and the CopenhagenBurnout Inventory (CBI-SS)...4.16.3 Model for incorporang usability-tesng as a survey pretesng methodE. Geisen 1 , S. Cook 11 RTI Internaonal, United StatesWeb-based surveys are unique in that they are both websites and surveys. As a result, pretesng methodsshould address both the website aspects and the survey aspects. Website usability professionals recommendthat usability tesng should be an iterave process that starts at the design phase and connues through systemdevelopment (Nielsen 1994; Dumas and Redish, 1993). However, usability tesng for web-surveys is oenconducted in a step-wise fashion, implemented aer the quesonnaire has been developed and cognivelytested and as a final step prior to launching data collecon. As a result, usability tesng findings for web-basedsurveys are frequently used to fine-tune survey design rather than to guide it. In this paper, we present aconnuum model for usability tesng for web-based surveys that integrates the methods for tesng websitesand surveys...4.17 Linking Survey and Administrave Data: A Methodological Perspecve IVTo be held on July 22, 2011 from: 11:00 to 12:30, in room 321.Coordinated by: Joe Sakshaug - JPSM, University of Michigan, United States

FRIDAY 22 JULY 1874.17.1 Stascal Matching of process generated data from the GDR and contemporary German panel data:An applicaon in social structure research.A. Salheiser 11 University of Jena, GermanyProcess-generated mass data from the vanished East German socialist society offer an in-depth picture of eliterecruitment, change of social structure, and societal differenaon. Biographies of three generaons of EastGerman cizens can be analyzed in order to trace the how’s and why’s of career mobility, and to research intothe impact of factors such as family background, social milieu, qualificaons and polical commitment. However,with the dissoluon of the East German state and the German reunificaon, the instuonal frameworkof careers in East Germany was drascally reshaped. In and aer the polical turnover, many of the old assetswere challenged or devaluated, others became more important. The career systems of a multude of societalsectors saw survivors, elite newcomers and losers of different types who faced complex condions of adapon,persistence, and social success...4.17.2 A Praccal Approach for Combining Surveys: An Applicaon using Egypt DHS 2005 and 2008 DataS. Abdelmageed 1 , H. Auda 11 IDSC, EgyptThis paper discusses the concept of combining surveys, the main different combining approaches, and thepossibility of using this opon to combine periodically conducted surveys in Egypt. Combining surveys is aninexpensive tool for increasing sample sizes and improving the precision of esmates. The paper presents asimple weighng approach that can be used for combining two or more surveys. This approach is then appliedto combine the two datasets of Egypt DHS 2005 and 2008 into a single dataset, where rescaling the weightsis based on minimizing the variances of esmates. Similar applicaons can be implemented using repeatedsurveys in Egypt such as the Household Income, Expenditure, and Consumpon Survey (HIECS) and Egypt LaborMarket Panel Survey (ELMPS)...4.17.3 Linking survey data with administrave employment data: The case of the IAB-ALWA surveyM. Antoni 11 Instute for Employment Research (IAB), GermanyFor many research quesons and methods of inference in sociology or economics, rich data sets are required.Since survey and administrave data sets have their respecve comparave advantages, a combinaon of bothdata sources enhances the informaon at hand.4.18 Experimental Methods in Survey Research IITo be held on July 22, 2011 from: 11:00 to 12:30, in room 415.Coordinated by:• Prof. Dr. Stefanie Eifler - University of Halle-Wienberg, Germany• Thomas Hinz - University of Konstanz, Germany• Ben Jann - University of Bern, Switzerland• Heiko Rauhut - ETH Zürich, Switzerland4.18.1 Germans’ segregaon preferences and immigrant group size: A factorial survey approachE. Schlueter 3 , J. Ullrich 1 , P. Schmidt 21 Goethe-Universität Frankfurt, Germany; 2 University of Giessen, Germany; 3 University of Cologne, ResearchInstute for Sociology, Germany

188 FRIDAY 22 JULYAlthough immigrant group size has long been hypothesized to exert negave effects on majority members’preferences for residenal and educaonal sengs, empirical research on this topic remains sparse. One plausiblereason for this gap in the literature are the methodological and substanve complexies created by thestrong associaon between immigrant group size and further correlates of preferences, e.g. neighbourhoodSES or quality of educaon. In this study, we use factorial survey methodology to address this problem. In twowithin-subjects experiments conducted over the internet, N = 1032 German parcipants evaluated schools orresidenal areas with different levels of immigrant group size. In the vignees describing schools and areas,we addionally varied factors that are thought to be ecologically related to immigrant group size (i.e...4.18.2 Measuring the percepon of gender income gaps using a vignee experimentC. Atzmüller 1 , P. Steiner 21 University of Applied Sciences, Vienna, Austria; 2 University of Wisconsin - Madison, United StatesVignee studies use experimentally varied descripons of situaons or persons (vignees) that are shownto respondents within surveys in order to elicit their judgments about these scenarios. In combinaon withquantave research techniques they are a powerful research tool with high internal and external validity andhigh potenal for causal invesgaons. We demonstrate the strength of vignee experiments as opposed tostandard quesonnaires using a study on the percepon of gender gaps in income. In this study which wasconducted in Vienna (Austria), each of the 972 respondents had to assess the income of 9 employee-vigneeswhich listed the employee’s sex, age, educaon, occupaon and job descripon, occupaonal experience, andmaternity/paternity leave...4.18.3 ”Many”, ”half” or ”one of two”? Assessing counter-biasing technique to reduce the self-reportedturnoutM. Comsa 2 , A. Gheorghita 11 ”Lucian Blaga” University of Sibiu, Romania; 2 Babes-Bolyai University of Cluj Napoca, RomaniaRegardless of the cultural area in which it is performed or of the survey type, electoral research shows thatself-reports in surveys over-esmate voter turnout. Over-reporng can be the (combined) result of manyfactors: social desirability bias, memory related errors, acquiescence, and unrepresentave sample. Socialdesirability seems to be one of the most influencing factors. In order to reduce social desirability effect onself-reported turnout many techniques are used (indirect queson, self-administered quesonnaire, counterbiasing,weighng the responses based on a social desirability scale, randomized response, list experimentdesign (or item / unmatched count technique), etc.). In this paper, we focus on the ”counter-biasing” technique.Offering the respondents counter-biasing informaon lowers the norm associated with the turnoutbehavior which in return should reduce the self-reported turnout...4.19 Innovaons in survey sampling IIITo be held on July 22, 2011 from: 11:00 to 12:30, in room 414.Coordinated by:• Siegfried Gabler - GESIS - Leibniz Instute for the Social Sciences, Germany• Seppo Laaksonen - University of Helsinki, Finland4.19.1 Student pilot group characteriscs and its relevance for final version tests resultsA. Slegrova 1 , I. Smetankova 11 www.scio.cz, Czech RepublicThe paper presents a comparison of several sets of pilot tests and their final versions. Students aged 11 weretested in their final year of lower primary stage (grade 5). The tests were taken for the purposes of entrance

FRIDAY 22 JULY 189examinaons to grammar schools in the Czech Republic in 2008-2010. The aim of the test is to idenfy pupils’general scholasc aptude. In the paper we deal with characteriscs of student groups tested in pilots: we lookinto the size of the group, the grade and the type of school. Selected test items stascs is presented for bothpilot and final versions. We discuss the difficulty of test quesons in relaon to the pilot group characteriscs.4.19.2 GIS support in survey samplingA. Illyes 2 , G. Hideg 2 , T. Prajczer 11 GeoX, Hungary; 2 Gallup, BelgiumThe presentaon looks at how recent innovaons in publicy available web-based and API-enabled GIS developmentsmay be applied in survey sampling. The authors have developed applicaons for supporng samplingacvies (typically in mul-country surveys) with GIS tools. Discussion will cover the role of such support applicaonsin addressing (a) coverage error, via random coordinate sampling on Google map (b) sampling error,via GIS support in full quality assessment of sampling implementaon (i.e. applicaon of random routes), and(c) nonresponse, using GSP-supported interviewer tracking to verify interviewer acvity against the survey protocol(presence at locaon at the required number of revisits, control for respecng vising schedule, etc.).The paper will report on the experiences of using such tools, i.e...4.19.3 On Computaonal Aspects of Simulaon Methods in the Sample Allocaon FrameworkP. Chiodini 3 , G. Manzi 1 , B. Martelli 4 , F. Verrecchia 21 Università di Milano, Italy; 2 ESeC, Italy; 3 University of Milano-Bicocca, Italy; 4 Italian Naonal Instute of Stas-cs (ISTAT), ItalyThe empirical validaon of the analycal properes of sampling allocaon methods is based on simulaontechniques either for the whole populaon or for domain analysis, or for other fields of stascs. At almostnegligible costs, these techniques allow for checking general properes (even asymptoc) of esmators orstascal models. The aim of this paper is to propose a simulaon sampling technique – namely a strafiedsampling with an ex-post ordered selecon - for the detecon of the most effecve sampling allocaon in termsof precision of esmates. More specifically, in order to compare several allocaon methods, it is necessary tomake unique and efficient the sampling experiment (from a computaonal point of view)...4.19.4 First evidences on the quality of the new register sampling frame in SwitzerlandM. Ernst Stähli 1 , A. Pollien 1 , M. Sapin 1 , D. Joye 21 FORS - Swiss Foundaon for Research in Social Sciences, Switzerland; 2 University of Lausanne, SwitzerlandThe sampling frame of high quality surveys in Switzerland historically moved from register data of municipaliesin the sevenes to fixed telephone number directories since the eighes. Recently, the coverage of theregistered fixed phone numbers is connually deteriorang, pushing to find new soluons. For Face to Facestudies such as the ESS and the ISSP, an intermediate soluon has been found with the complete databaseof mailboxes in buildings over the country. Since the end of 2010, a register sampling frame of individuals isavailable, supposed to cover the whole populaon with nearly real me informaon. The ESS 2010 is the veryfirst survey using this new sampling frame. In this paper we analyze data from 3 rounds of the ESS (2006, 2008and 2010) - each of them being based on a different sampling frame - and the non-response survey of 2006and 2010...4.20 Surveys and polical opinions and acviesTo be held on July 22, 2011 from: 11:00 to 12:30, in room 410.Coordinated by: Sandra Penic - University of Lausanne, Switzerland

190 FRIDAY 22 JULY4.20.1 German parliamentary elecons 2009 from the viewpoint of public surveys on policy issuesA. Tangian 11 Hans Boeckler Foundaon, GermanyFive German leading pares and their coalions are evaluated from the viewpoint of direct democracy. Forthis purpose, the posions of the pares on over 30 topical issues are compared with the results of polls ofpublic opinion. The outcomes are summarized in the indices of popularity and universality of the pares. Theselecon of policy issues and the informaon on the party posions are given as in the Wahl-O-Mat(2010) forthe last Bundestag (German parliamentary) elecons 2009.4.20.2 Does polical values distribuon in Russia match with German, French and American ones? (Crossculturalanalysis)T. Khavenson 1 , E. Migol 21 State University - Higher School of Economics, Russia; 2 State University Higher School of Economics, RussiaNowadays the role of non-instuonal society development factors is becoming more and more significant.That is why the system of the polical values is being in the centre of the debate of the academic community. Inthe paper comparave analyses of the polical values of three sociees: Russian, German, Franch and the USAis provided, as well as the reasoning of the analyses on the bases of socio-professional straficaon. (Evans,1993; Weakliem, 1992; Manza, Hout, Brooks, 1995).4.20.3 Voter overrepresentaon, vote misreporng, and turnout bias in postelecon surveys. A systemacreview of vote validaon studiesP. Selb 1 , S. Munzert 11 University of Konstanz, GermanyFigures from postelecon surveys oen grossly overesmate actual elecon turnout. Two disnct phenomenaaccount for this bias: overrepresentaon of actual voters in realized survey samples, and vote misreporngmostly by nonvoters parcipang in surveys. Vote validaon studies, while rarely conducted due to data privacyand economic constraints, offer invaluable informaon about both sources of bias. As yet, however,vote validaon studies have almost exclusively been used in isolaon to idenfy individual-level correlates ofself-selecon into survey samples and vote misreporng. Such approaches are ill-suited to shed light on thepuzzling observaon that turnout bias in postelecon surveys varies tremendously across countries and elec-ons. In this paper, we present a meta analysis of 47 vote validaon studies to determine why turnout bias(and its components) vary in magnitude across elecon surveys...4.20.4 The development of polical parcipaon: A 20-year longitudinal examinaon of American youthJ. Miller 11 Instute for Social Research, University of Michigan, United StatesThe first concerns about the development of polical atudes and a willingness to parcipate in public affairscan be traced to the democrac sociees in early Greece. Over the last century, a strong and growing socialscience literature has developed on this issue. The impact of the electronic revoluon in communicaondemands a re-examinaon of these issues.4.21 Special Issues in Surveys: Ecological BehaviorTo be held on July 22, 2011 from: 11:00 to 12:30, in room 319.Coordinated by: Kathrin Kissau - FORS - Swiss Foundaon for Research in Social Sciences, Switzerland

FRIDAY 22 JULY 1914.21.1 Naonal and Household Income and People’s Concern about the EnvironmentM. Fairbrother 11 University of Bristol, United KingdomWho cares about the environment? Work published by Franzen and Meyer (2010) in the European SociologicalReview, based on two waves of ISSP data, suggests that environmental concern is significantly greater inwealthier sociees, and among wealthier people within a given society. This paper challenges both of theseclaims, using data from the European and World Values Surveys, which together cover a larger sample ofcountries and, for some quesons, a longer period of me. The paper disnguishes between people’s concernsabout local environmental problems, global environmental problems, and their overall willingness topay for environmental protecon. The paper begins by presenng methodologically novel non-parametricsurvey results that, by themselves, show Franzen and Meyer’s central claims to be erroneous. Mullevel modelscontrolling for potenally confounding variables then provide supporng evidence...4.21.2 Survey Measurement of Energy EfficiencyI. Ograjenšek 1 , P. Domadenik 1 , D. Trobec 11 University of Ljubljana, Faculty of Economics, SloveniaThe concept of energy efficiency (use of less energy to achieve the same effects) can be dealt with at themacro and micro level; from the economic, social, polical, ecological, philosophical, health, and many otherperspecves. The programmes introduced by governments aim at achieving an efficient use of energy resourceswith the ulmate goals of reducing (1) dependence on foreign energy resources and (2) the negaveinfluence of energy consumpon on the environment. The programmes focus both on household and organizaonalconsumers (such as hospitals and schools) on one, and industrial units on the other hand. We’restudying the energy efficiency of industrial units.The research quesons we address in this paper are thereforethe following:1...

192List of ParcipantsAbdelfatah, S. (abdelfatah.sally@gmail.com), 120Abdelmageed, S. (stascsgirl@yahoo.com), 187ADLER, I. (iriad@post.tau.ac.il)Ahami, R. (rahami@essex.ac.uk), 45Ahrendt, D. (hilde.beerten@g.com)Ainsaar, M. (mare.ainsaar@ut.ee)Al-Emadi, D. (dalemadi@qu.edu.qa), 59, 151Alahmed, D. (ahmedm@scta.gov.sa)Alanya, A. (ahu_alanya@gallup-europe.be), 118, 160Alda, H. (alda@bibb.de), 26Aleksynska, M. (maleksynska@gmail.com), 85Allum, N. (nallum@essex.ac.uk), 69, 83, 84, 101, 113, 124,128,Ambrasat, J. (jens.ambrasat@fu-berlin.de), 50Amin, A. (a.amin@uvt.nl), 101An, A. (anthony.an@sas.com)Andreadis, I. (john@polsci.auth.gr), 30Andreeva, G. (galina.andreeva@ed.ac.uk), 113Anger, S. (sanger@diw.de)Anheier, H. (helmut.anheier@soziologie.uni-heidelberg.de),51Antoni, M. (manfred.antoni@iab.de), 187Arnsberger, P. (arnsburg@hawaii.edu), 35, 154Arpino, B. (bruno.arpino@unibocconi.it), 149Aschauer, W. (wolfgang.aschauer@sbg.ac.at), 70, 177Ashenfelter, K. (ktashenfelter@yahoo.com), 186Atzmüller, C. (chrisane.atzmueller@-campuswien.ac.at),188,Aumeyr, M. (marna.aumeyr@ons.gov.uk), 106Bachteler, T. (tobias.bachteler@uni-due.de), 23, 43, 143Barber, J. (jebarber@umich.edu), 110Barni, D. (daniela.barni@unica.it), 58Barr, M. (mlbarr@hotmail.com)Barre, J. (jennifer.barre@wur.nl), 169Bartolomé Peral, E. (ebartolo@soc.deusto.es), 153Bartsch, S. (sbartsch@empas.uni-bremen.de), 90Bassi, F. (bassi@stat.unipd.it), 86Bathelt, S. (bathelt@uni-landau.de), 174Bauer, G. (gerrit.bauer@mzes.uni-mannheim.de), 133, 134Bauer, J. (johannes.bauer@soziologie.uni-muenchen.de), 181Bauknecht, J. (juergen.bauknecht@sowi.uni-stugart.de),174Beaule, A. (abeaule@isr.umich.edu)Bechert, I. (insa.bechert@gesis.org), 124, 152Beck, F. (francois.beck@inpes.sante.fr), 24, 61, 174Becker, A. (anke.becker@uni-bonn.de), 172Becker, B. (birgit.becker@gesis.org), 61Becker, D. (dominik.becker@wiso.uni-koeln.de), 56, 167, 176,Beckers, T. (lo.beckers@uni-duesseldorf.de), 36, 139, 158,176Beebe, T. (beebe.mothy@mayo.edu), 33, 54, 60Behr, D. (dorothee.behr@gesis.org), 80, 92, 100, 134Beierlein, C. (constanze.beierlein@gesis.org), 36–38, 57, 70,87, 98, 182Bendahan, S. (samuel.bendahan@unil.ch), 130Benítez, I. (ibenitez@ugr.es), 82, 122Benne, J. (benne@wiso.uni-koeln.de), 127Berger, R. (berger@sozio.uni-leipzig.de), 134Bergmann, M. (michael.bergmann@uni-mannheim.de), 184Bernotsky, R. (lbernotsky@wcupa.edu), 164Berridge, D. (d.berridge@lancaster.ac.uk), 32, 33Berzelak, N. (nejc.berzelak@fdv.uni-lj.si), 64Best, H. (best@uni-mannheim.de), 133, 134Bethlehem, J. (jbtm@cbs.nl), 69Beullens, K. (koen.beullens@soc.kuleuven.be), 147Bianchi, A. (annamaria.bianchi@unibg.it), 125Bichsel, M. (marna.bichsel@fors.unil.ch), 25Biernacka, K. (klaudia-biernacka@wp.pl)Billiet, J. (jaak.billiet@soc.kuleuven.be), 86, 107, 118, 119,Bilocq, F. (france.bilocq@statcan.gc.ca), 121Biolca Rinaldi, F. (ferruccio.biolca@unimi.it), 152Birch, L. (lisa-maureen.birch.1@ulaval.ca), 77Birkelbach, K. (klaus.birkelbach@uni-due.de), 34, 106Birkelbach, R. (robert.birkelbach@gesis.org), 72Blake, M. (m.blake@natcen.ac.uk), 104BLANERO, K. (barkarin@post.tau.ac.il)Blank, T. (thomas.blank@gmail.com), 98Blanke, K. (karen.blanke@destas.de), 117Blasius, J. (jblasius@uni-bonn.de), 71, 88Blohm, M. (michael.blohm@gesis.org), 122Blom, A. (ablom@staff.mail.uni-mannheim.de), 31, 45, 101,108, 142, 162Blumensel, J. (jan_eric.blumensel@mzes.uni-mannheim.de),49, 101Bochud, M. (murielle.bochud@chuv.ch), 50Boehle, M. (mara.boehle@gesis.org), 159Boehler, C. (chrisan.boehler@brunel.ac.uk), 167Boeije, H. (h.boeije@uu.nl), 122, 123Bohr, J. (jeanee.bohr@gesis.org), 74Branch-Mueller, J. (jbranch@ualberta.ca), 105Braun, M. (michael.braun@gesis.org), 92, 182Breustedt, W. (wiebke.breustedt@uni-due.de), 167Brislinger, E. (evelyn.brislinger@gesis.org), 53, 67Brites, R. (rui.brites@iscte.pt), 34, 78Brown, M. (m.brown@ioe.ac.uk)Brulhart, M. (marius.brulhart@unil.ch), 130Brunton-Smith, I. (i.r.brunton-smith@surrey.ac.uk), 108, 128,Brzoska, P. (patrick.brzoska@uni-bielefeld.de), 145

LIST OF PARTICIPANTS 193Budnick, A. (andrea.budnick@charite.de), 48Burton, J. (jburton@essex.ac.uk), 68, 143, 175Busse, B. (bria.busse@ifs.tu-darmstadt.de), 160Calderwood, L. (l.calderwood@ioe.ac.uk), 39, 40, 49, 60,72, 73, 91, 162Campanelli, P. (dr.pamela.campanelli@thesurveycoach.com),104, 156, 157Caro, D. (daniel.caro@iea-dpc.de), 80Carpenter, H. (hannah.carpenter@tns-bmrb.co.uk)Carstensen, J. (johann.carstensen@wiso.uni-erlangen.de),184Carvalho, D. (dianadiascarvalho@gmail.com), 78Chappuis, A. (aline.chappuis@chuv.ch), 50Chen, Y. (y.chen1@lancaster.ac.uk), 32Chiodini, P. (paola.chiodini@unimib.it), 189Christensen, A. (ach@niph.dk), 42Chua, K. (chua_khim_hwee@mcys.gov.sg)Chun, Y. (ychun2@gmail.com), 109, 175Cieciuch, J. (jancieciuch@gmail.com), 70, 87Cleary, A. (andrew.cleary@ipsos.com), 91Clifford, I. (cliffori@tcd.ie), 81Comsa, M. (mircea@mmt.ro), 26, 179, 188Conolly, A. (anne.conolly@natcen.ac.uk), 49Coombe, R. (richard.coombe@kantaroperaons.com), 170Corey, J. (joanne.corey@abs.gov.au), 68Cornilleau, A. (anne.cornilleau@sciences-po.fr), 56Coromina, L. (lluis.coromina@udg.edu), 151Couper, M. (mcouper@umich.edu), 31, 91Cousteaux, A. (annesophie.cousteaux@sciences-po.fr), 25Cuesta Azofra, M. (mcuesta@cis.es), 78Czaplicki, C. (chrisn.czaplicki@drv-bund.de), 43D’Ardenne, J. (joanna.d’ardenne@natcen.ac.uk),40, 104, 105,Datler, G. (datler@soziologie.uzh.ch)Davidov, E. (davidov@soziologie.uzh.ch), 36, 57, 70, 87, 98,182Dept, S. (steve.dept@capstan.be), 100Deshmukh, Y. (yashwant@teamcvoter.com), 115Devine, P. (p.devine@qub.ac.uk), 77Diekmann, A. (andreas.diekmann@soz.gess.ethz.ch), 27, 119Diemand, C. (chdiemand@gmail.com), 39Diewald, M. (marn.diewald@uni-bielefeld.de), 56Dillman, D. (dillman@wsu.edu), 30, 42, 54, 137Diop, A. (adiop@qu.edu.qa), 59, 151Dobewall, H. (dobewall@ut.ee), 87, 88Doerner, W. (wolfdorn@gmx.de), 52Dollmann, J. (joerg.dollmann@mzes.uni-mannheim.de), 137Domanski, H. (hdomansk@ifispan.waw.pl), 100Dorau, R. (ralfdorau@freenet.de), 71Dorer, B. (brita.dorer@gesis.org), 80, 100, 101Döring, A. (anna.doering@uni-muenster.de), 70Draisma, S. (s.draisma@ggzingeest.nl), 68, 86Duelmer, H. (hduelmer@uni-koeln.de), 154Dunne, J. (john.dunne@cso.ie), 175Durrant, G. (g.durrant@southampton.ac.uk), 108, 128Duspivova, K. (duspivova@trexima.cz), 77Eckman, S. (steph.eckman@gmail.com), 63, 108, 147Edler, S. (susanne.edler@uni-bielefeld.de), 23Edlund, J. (jonas.edlund@soc.umu.se), 124, 152Ehrlich, S. (susanne_ehrlich@gmx.de)Eifler, P. (stefanie.eifler@soziologie.uni-halle.de), 172, 187Eikelenboom, M. (m.eikelenboom@vumc.nl), 28Elcheroth, G. (guy.elcheroth@unil.ch), 137, 144Elias, P. (peter.elias@warwick.ac.uk), 124Ellert, S. (sebasan.ellert@gmx.de), 48Elliot, M. (mark.elliot@manchester.ac.uk), 23Elshout, S. (s.elshout@uvt.nl), 74Erens, B. (bob.erens@lshtm.ac.uk), 121Ernst Stähli, M. (michele.ernststaehli@fors.unil.ch), 81, 93,99, 189Ersanilli, E. (evelyn.ersanilli@qeh.ox.ac.uk), 66, 177Fairbrother, M. (m.fairbrother@bristol.ac.uk), 159Falnes-Dalheim, E. (eld@ssb.no)Farago, P. (peter.farago@fors.unil.ch)Fazekas, Z. (zoltan.fazekas@univie.ac.at), 122, 183Fedáková, D. (dfedak@saske.sk), 95Fernández-Reino, M. (marinha.fernandez@upf.edu), 111Fernee, H. (h.fernee@scp.nl), 78Ferreira de Almeida, J. (ferreira.almeida@iscte.pt), 34Ferrez, E. (eliane.ferrez@fors.unil.ch), 25Fioramon, L. (lorenzo.fioramon@gmail.com), 52Fitriana, E. (fitriana_efi@yahoo.com), 95Fitzgerald, R. (r.fitzgerald@city.ac.uk), 38, 41, 80, 102Flere, S. (sergej.flere@uni-mb.si), 41, 98Font, J. (joan.font@cchs.csic.es), 77Fontaine, X. (fontaine@pse.ens.fr), 137Forn, J. (jessica.forn@gesis.org), 38, 39Fransen, S. (sonja.fransen@maastrichtuniversity.nl), 132Franzen, A. (franzen@soz.unibe.ch), 183Friedrich, A. (ane.friedrich@bibb.de), 26Friedrichs, J. (friedrichs@wiso.uni-koeln.de), 104Fuchs, M. (fuchs@ifs.tu-darmstadt.de), 27, 47, 159, 160,173,Fücker, S. (sonja.fuecker@fu-berlin.de), 62Funke, F. (email@frederikfunke.net), 156Gabler, S. (siegfried.gabler@gesis.org), 157, 158, 180, 188,Gagné, C. (camille.gagne@fsi.ulaval.ca)Ganninger, M. (mahias.ganninger@gesis.org), 158Garnier, B. (garnier@ined.fr), 102Geisen, E. (egeisen@r.org), 186Gerhards, C. (chrisan.gerhards@uni-bielefeld.de), 23Gianeoni, L. (lavinia.gianeoni@unil.ch), 136Giesselmann, M. (giesselmann@wiso.uni-koeln.de), 151Gielman, S. (steve@mktginc.com), 114Given, L. (lisa.given@scotcen.org.uk)Glasner, T. (t.glasner@uu.nl)Glasson-Cicognani, M. (melanie.glasson@unil.ch)Glatz, N. (nicolas.glatz@chuv.ch), 50Godinet, M. (meripa@hawaii.edu), 154Goebel, J. (jgoebel@diw.de), 149Goedemé, T. (m.goedeme@ua.ac.be), 113Goesswald, A. (goesswalda@rki.de), 33Goee, L. (lorenz.goee@unil.ch), 130Gonthier, F. (frederic.gonthier@iep-grenoble.fr), 140Gorinas, C. (cgorinas@asb.dk), 111

194 LIST OF PARTICIPANTSGosen, S. (gosen@staff.uni-marburg.de), 47Goschall, C. (chris_goschall@nass.usda.gov), 117Gramlich, T. (tobias.gramlich@uni-due.de), 94Granda, P. (peterg@umich.edu)Grauenhorst, A. (grauenhorst@wiso.uni-koeln.de), 34, 106Gray, M. (m.gray@natcen.ac.uk), 104, 105Gresser, A. (anne.gresser@uni-wuerzburg.de)Griffin, D. (deborah.h.griffin@census.gov), 55, 146Grigoras, V. (vlad.grigoras@sas.unibuc.ro), 165Guillaud, E. (elvire.guillaud@univ-paris1.fr), 152Güllner, G. (guellner@uni-bonn.de), 63, 64, 88Guyer, H. (hguyer@umich.edu), 49, 74, 75Häfliger, U. (haefliger@ipz.uzh.ch), 57Halbherr, V. (verena.halbherr@gesis.org), 67Han, M. (meike.han@student.uni-tuebingen.de), 55Hans-Jürgen Andreß, H. (hja@wiso.uni-koeln.de), 103, 126,151,Hansen, S. (sehansen@isr.umich.edu), 53, 66, 117Harkness, J. (jharkness2@unl.edu), 65, 66Harrison, E. (eric.harrison.1@city.ac.uk), 78, 94, 123Hartman, A. (hartmana@mail.nih.gov)Hartmann, J. (josef.hartmann@tns-infratest.com)Harzeneer, K. (karoline.harzeneer@gesis.org), 166Heizmann, B. (boris.heizmann@uni-jena.de), 97Hermanni, H. von (hagenvonhermanni@gmx.de), 27Hidalgo, M. (mdhidalg@um.es), 82, 95Hideg, G. (gergely_hideg@gallup-europe.be), 66, 145, 189Hierro, M. (mariajose.hierro@upf.edu), 78Hill, C. (chill@r.org), 120Hillmert, S. (steffen.hillmert@uni-tuebingen.de), 55Hinz, T. (thomas.hinz@uni-konstanz.de), 172, 187Hirai, Y. (yhirai@tmu.ac.jp), 47Hoelscher, M.(michael.hoelscher@soziologie.uni-heidelberg.de),51, 52Hoffmeyer-Zlotnik, J. (juergen.hoffmeyer-zlotnik@gesis.org),79, 99, 112Höglinger, M. (marc.hoeglinger@soz.gess.ethz.ch), 27Holmberg, A. (anders.holmberg@scb.se), 109Hoogendoorn, A. (a.hoogendoorn@ggzingeest.nl), 42Hope, S. (steven.hope@natcen.ac.uk), 156, 174Horn, C. Van (vanhorn@rci.rutgers.edu), 106Horváth, B. (beata.horvath@ksh.hu)Houten, G. van (gijs.vanhouten@eurofound.europa.eu), 45,67,Houwen, K. van der (hhwn@cbs.nl), 90Howell, D. (dahowell@umich.edu), 39, 59, 151Hox, J. (j.hox@uu.nl), 30, 54, 90, 161, 183Hoyle, L. (larryhoyle@ku.edu), 54Huber, M. (marna.huber@iab.de), 24Huber, S. (stephan.huber@phz.ch), 148Hughes, G. (n.hughes@surrey.ac.uk), 83, 101, 113Hughes, T. (todd.r.hughes@census.gov), 186Huismans, S. (s.huismans@kpnplanet.nl), 178Hulliger, B. (beat.hulliger@nw.ch), 29, 185Humery, M. (m.humery@mistrend.ch)Hunkler, C. (hunkler@uni-mannheim.de), 134Huschka, D. (dhuschka@ratswd.de)Illyes, A. (agnes_illyes@gallup.hu), 66, 145, 189Im, Y. (youngjo@uchicago.edu), 130, 171Jablonski, W. (wjablonski@uni.lodz.pl), 96Jäckle, A. (aejack@essex.ac.uk), 156, 157Jacobs, D. (dirk.jacobs@ulb.ac.be), 165Jacquart, P. (philippe.jacquart@gmail.com), 130Jagodzinski, W. (jagodzinski@uni-koeln.de), 139, 153, 165,166, 178Janik, M. (m.janik@uni-muenster.de), 36, 70Jann, B. (jann@soz.unibe.ch), 26, 27, 46, 172, 187Jans, M. (mahew.e.jans@census.gov), 117Jarre, C. (caroline.jarre@effortmark.co.uk), 170Jelonek, M. (magdalena.jelonek@uek.krakow.pl), 150Jento, S. (susie.jento@ssb.no)Ježek, S. (jezek@fss.muni.cz), 142Johann, D. (david.johann@univie.ac.at), 180John, D. (john@geronto.uni-erlangen.de), 37Johnson, D. (drj10@psu.edu), 126Johnson, T. (tjohnson@srl.uic.edu), 182Jong, R. de (dejongroel@gmail.com), 131, 148Jonsdor, G. (gudbjorg@hi.is)Joye, D. (dominique.joye@unil.ch), 93, 99, 189Jutz, R. (regina.jutz@gesis.org), 178KALOGERAKI, S. (s.kalogeraki@gmail.com), 50Kaminska, O. (kaminol@gmail.com), 160Kamoen, N. (n.kamoen@uu.nl), 46, 156Kampen, J. (jarl.kampen@wur.nl), 169Kampmann, J. (jara.kampmann@gesis.org), 166Kappelhof, J. (j.kappelhof@scp.nl), 30, 141Karpinski, Z. (zb.karpinski@gmail.com), 181Katsanidou, A. (alexia.katsanidou@gesis.org), 44Keler, K. (karolina.keler@uj.edu.pl), 73Kemper, C. (christoph.kemper@gesis.org), 37, 38Kessler, G. (kessler.georg@gmail.com), 99Ketende, S. (s.ketende@ioe.ac.uk), 91, 162Keulenaer, F. De (femke_de_keulenaer@gallup-europe.be),159, 160, 173Keusch, F. (florian.keusch@wu.ac.at), 179Kharchenko, N. (nkh@kiis.com.ua), 43Khavenson, T. (xtanya@gmail.com), 55, 190Kiesl, H. (hans.kiesl@hs-regensburg.de), 135, 150, 163Kim, J. (jkim@iastate.edu), 43Kim, S. (smilegong@korea.kr), 43Kirchner, A. (antje.kirchner@uni-konstanz.de), 27Kissau, K. (kathrin.kissau@fors.unil.ch)Klausch, T. (l.t.klausch@uu.nl), 184Klausing, A. (klausing@edk.ch)Kleiner, B. (brian.kleiner@fors.unil.ch), 25, 65, 80Kleinke, K. (krisan.kleinke@uni-bielefeld.de), 131, 148Klug, S. (stefan.klug@demoscope.ch), 160Kneeshaw, J. (kneejw@essex.ac.uk), 44Knies, G. (gknies@essex.ac.uk), 143, 175Knutsen, O. (oddbjorn.knutsen@stv.uio.no), 139, 178Koch, A. (achim.koch@gesis.org), 67, 121, 122, 140, 146Koch, S. (solange.koch@uni-bamberg.de), 136Koerner, T. (thomas.koerner@destas.de), 79Koller, M. (monika.koller@wu.ac.at), 155Kononykhina, O. (o.kononykhina@gmail.com), 52

LIST OF PARTICIPANTS 195Korbmacher, J. (korbmacher@mea.uni-mannheim.de), 142,143, 162,Kosyakova, Y. (yuliya.kosyakova@uni-bamberg.de), 147Köthemann, D. (dennis.koethemann@uni-bielefeld.de), 57Kovaleva, A. (anastassiya.kovaleva@gesis.org), 37, 38Kramer, S. (skramer1958@verizon.net), 164Krause, I. (ina.krause@uni-bielefeld.de), 48Krebs, D. (dagmar.krebs@sowi.uni-giessen.de), 133, 179Kreuter, F. (reuter@survey.umd.edu), 108, 161, 162Krieger, U. (ulrich.krieger@uni-mannheim.de), 31, 82, 105,125, 126, 141, 147, 160Kroh, M. (mkroh@diw.de), 76, 129, 171Krumpal, I. (krumpal@sozio.uni-leipzig.de), 26, 27, 46, 172,Kuha, J. (j.kuha@lse.ac.uk), 41, 107, 128Kulin, J. (joakim.kulin@soc.umu.se), 87Kuntz, A. (anabel.kuntz@wiso.uni-koeln.de), 36Kunz, T. (kunz@ifs.tu-darmstadt.de), 173Laaksonen, S. (seppo.laaksonen@helsinki.fi), 157, 180, 181,188Laan, J. van der (dj.vanderlaan@cbs.nl), 142Labigne, A. (labigne@transnaonalstudies.eu), 52Laflamme, F. (francois.laflamme@statcan.gc.ca), 45, 121Laganà, F. (francesco.lagana@unil.ch), 136, 137LAGO, J. VON (joseramonlago@hotmail.com)Lai, J. (js-lai@northwestern.edu), 62, 135Lalive, R. (rafael.lalive@unil.ch), 130Lamprianou, I. (iasonas@ucy.ac.cy), 51Lang, V. (v.lang@uni-tuebingen.de), 55Latcheva, R. (latcheva@soziologie.uzh.ch), 99Laub, S. (laub@ifs.tu-darmstadt.de)Laudátová, M. (100111@mail.muni.cz), 153Laux, T. (thomas.laux@soziologie.uni-heidelberg.de), 52Lavado, S. (susana.lavado@ics.ul.pt), 140Le Goff, J. (jean-marie.legoff@unil.ch), 85Leeuw, E. De (e.d.deleeuw@uu.nl), 30, 42, 54, 90, 110, 124,161,Legleye, S. (stephane.legleye@ined.fr), 24, 47, 61, 119, 174,Leissou, E. (eleissou@isr.umich.edu)Lenski, J. (jlenski@edisonresearch.com)Lenzner, T. (mo.lenzner@gesis.org), 135, 171Leonyeva, Y. (yana.leonyeva@soc.cas.cz), 57, 178Lewin-Epstein, N. (noah1@post.tau.ac.il), 38Li, F. (fenfang@hawaii.edu), 35, 154Liebau, E. (eliebau@diw.de), 145, 165Lin, J. (chung-tung.lin@fda.hhs.gov), 117Linek, L. (lukas.linek@soc.cas.cz), 176Link, M. (michael.link@nielsen.com), 96Linne, M. (monika.linne@gesis.org), 53Lippe, E. von der (vonderlippee@rki.de), 93Lipps, O. (oliver.lipps@fors.unil.ch), 31, 45, 136, 137Litzel, N. (nicole.litzel@iab.de), 35Lloyd, K. (k.lloyd@qub.ac.uk), 40, 102Loosveldt, G. (geert.loosveldt@soc.kuleuven.be), 31, 147,183,Ludwig, V. (vludwig@rumms.uni-mannheim.de), 31, 82Lugg, P. (p.lugg@uu.nl), 105, 125, 141, 147, 160, 161,183,Luijkx, R. (r.luijkx@uvt.nl), 39, 67, 139, 153, 165, 178Lynn, P. (plynn@essex.ac.uk), 42, 156, 157, 160Maia, M. (mmaia@porto.ucp.pt), 158Malmdin, J. (joakim.malmdin@scb.se)Malter, F. (malter@mea.uni-mannheim.de), 39Manabe, K. (kazufumi.manabe@niy.com), 154Manchin, R. (robert_manchin@gallup-europe.be), 145, 159,160, 173Mangold, F. (mangold@uni-hohenheim.de), 166Marinica, B. (marinicab@yahoo.com), 170Markou, E. (markou@ined.fr), 61, 102Maroco, J. (jpmaroco@gmail.com), 186Marshall, L. (leah.b.marshall@census.gov), 110Martelli, B. (bmartelli@istat.it), 189Martens, M. (m.g.j.martens@uvt.nl), 101Mar, M. (mmar@ua.es), 112Marn, P. (p.marn@city.ac.uk), 42, 102Massin, S. (sophie.massin@univ-paris1.fr), 153Matei, A. (alina.matei@unine.ch), 135, 150, 157, 163Matsuo, H. (hideko.matsuo@soc.kuleuven.be), 31, 118, 119Maurice, J. von (jua.von-maurice@uni-bamberg.de), 68,136,Mayerl, J. (jochen.mayerl@sowi.uni-stugart.de), 89McGonagle, K. (kmcgon@umich.edu), 91McManus, S. (sally.mcmanus@natcen.ac.uk), 28McMillen, R. (rcm19@msstate.edu), 103Meixner, S. (sonja.meixner@uni-bamberg.de), 68Meltzer, H. (hm74@le.ac.uk), 61Méndez, M. (mmendez@cis.es), 145, 164, 177Menold, N. (natalja.menold@gesis.org), 63, 64, 135Meuleman, B. (bart.meuleman@soc.kuleuven.be), 107, 138,154, 166Meulemann, H. (meulemann@wiso.uni-koeln.de), 103, 126Mewes, J. (jmewes@bigsss.uni-bremen.de), 138Meyermann, A. (alexia.meyermann@uni-bielefeld.de), 23Midtsæter, H. (hege.midtseter@nsd.uib.no), 45, 112Migol, E. (ekaterina.migol@gmail.com), 190Mika, T. (tatjana.mika@drv-bund.de), 43Mikucka, M. (mikucka.m@gmail.com), 115, 138Miller, J. (jondmiller@umich.edu), 190Miller, K. (ktm8@cdc.gov), 92, 93Mochmann, I. (ingvill.mochmann@gesis.org), 29Mohadjer, L. (leylamohadjer@westat.com), 135, 136, 150,163,Mohler, P. (peter.mohler@uni-mannheim.de), 95Mohorko, A. (a.mohorko@students.uu.nl), 90Moncagaa, P. (paolo.moncagaa@upf.edu), 132, 151Morales, L. (laura.morales@manchester.ac.uk), 145, 164,177,Morselli, D. (davide.morselli@unil.ch), 36Moustaki, I. (i.moustaki@lse.ac.uk), 107Mueller, G. (gerrit.mueller@iab.de), 126Müller, C. (c.mueller@mistrend.ch)Müller, G. (georg.mueller@unifr.ch), 101, 140Müller, M. (io.ibsf@usa.net)Munzert, S. (simon.munzert@uni-konstanz.de), 144, 190Nalli, M. (michela.nalli@sssup.it), 96navarro, a. (alfredo.navarro@census.gov), 146Neri, A. (andrea.neri@bancaditalia.it), 129Neuert, C. (neuert@ifs.tu-darmstadt.de), 160Neusar, A. (a.neusar@seznam.cz), 142, 148

196 LIST OF PARTICIPANTSNevala, S. (sami.nevala@fra.europa.eu)NG, L. (ng_luan_eng@moe.gov.sg)Nicolaas, G. (g.nicolaas@natcen.ac.uk), 156, 174Nicolaas, G. (gerry.nicolaas@virgin.net)Nonnenmacher, A. (a.nonnenmacher@ipw.uni-hannover.de),104Norrie, R. (norrie@wiso.uni-koeln.de), 104O’Muircheartaigh, C. (colm@uchicago.edu), 63Oberski, D. (daniel.oberski@gmail.com), 102Odella, F. (francesca.odella@soc.unitn.it), 59Ofstedal, M. (mbo@umich.edu), 49, 74, 75Ograjenšek, I. (irena.ograjensek@ef.uni-lj.si)Oliveri, A. (oliveri@unipa.it), 88, 121Orten, H. (hilde.orten@nsd.uib.no), 45, 79, 99, 112Osgood, J. (josgood@wcupa.edu), 164Osier, G. (guillaume.osier@statec.etat.lu), 158Oud, H. (j.oud@pwo.ru.nl), 86, 107, 133Oudejans, M. (m.oudejans@uvt.nl), 110Padilla, J. (jpadilla@ugr.es), 82, 92, 93, 122Pascale, J. (joanne.pascale@census.gov), 163Pei Lin, Y. (yeo_pei_lin@mcys.gov.sg)Penic, S. (sandra.penic@unil.ch), 144Penn, R. (r.penn@lancaster.ac.uk), 32Pennec, S. (pennec@ined.fr), 103Pennell, B. (bpennell@isr.umich.edu), 47, 65, 66Perek-Bialas, J. (jolanta.perek-bialas@uj.edu.pl), 89Perri, P. (pierfrancesco.perri@unical.it), 120Petclerc, A. (ap2960@columbia.edu)Peytchev, A. (apeytchev@r.org), 121, 140, 146Pfnioo, V. (tpiaskovskaya@kiis.com.ua)Pichler, F. (florian.pichler@univie.ac.at), 97Pickering, E. (emily.pickering@tns-bmrb.co.uk)Plewis, I. (ian.plewis@manchester.ac.uk), 142Poethko-Mueller, C. (poethko-muellerc@rki.de), 129Pöge, A. (andreas.poege@uni-bielefeld.de), 44, 98Pohjanpää, K. (kirs.pohjanpaa@stat.fi), 81Pohl, S. (steffi.pohl@uni-bamberg.de), 151Polek, E. (elapolek@o2.pl), 98Pollien, A. (alexandre.pollien@fors.unil.ch), 81, 93, 99, 189,Porpiglia, A. (alessandro_porpiglia@inwind.it), 97Porter, S. (srporter@iastate.edu), 163Poznyak, D. (dmitriy.poznyak@soc.kuleuven.be), 59, 155Prevodnik, K. (katja.prevodnik@fdv.uni-lj.si), 34Quatember, A. (andreas.quatember@jku.at), 120Rammstedt, B. (beatrice.rammstedt@gesis.org), 37, 38, 62,182,Ramos, A. (alice.ramos@ics.ul.pt), 70, 140Rasner, A. (arasner@diw.de), 143Rauhut, H. (rauhut@gess.ethz.ch), 172, 187Rayp, G. (glenn.rayp@ugent.be), 84Razafindratsima, N. (razafind@ined.fr), 47, 119Reinecke, J. (jost.reinecke@uni-bielefeld.de), 131, 148Reinelt, C. (chrisane.reinelt@uni-due.de), 106Reinhardt, J. (jan.reinhardt@paranet.ch), 82, 95Remr, J. (remr@centrum.cz), 105Ren, L. (isssrenly@pku.edu.cn), 113Renschler, I. (isabelle.renschler@fors.unil.ch), 65, 80Reuband, K. (reuband@phil-fak.uni-duesseldorf.de), 34, 48Revilla, M. (melanie.revilla@hotmail.fr), 64, 132Riedmann, A. (arnold_riedmann@yahoo.de)Rimac, I. (ivan.rimac@pravo.hr), 166Ristau, I. (ina.ristau@uni-bamberg.de), 68Roberts, C. (caroline.roberts@unil.ch), 69, 124, 135Robinson, C. (chloe.robinson@natcen.ac.uk), 25Robinson, G. (gm.robinson@ulster.ac.uk), 77Ródenas, C. (crodenas@ua.es), 112ROELKE, H. (roelke@dipf.de), 100Rohrbach-Schmidt, D. (rohrbach@bibb.de), 26Romano, M. (mariafrancesca.romano@sssup.it), 96Romano Bergstrom, J. (jennifer.romano@census.gov), 117Rossmann, J. (joss.rossmann@gesis.org), 49, 96Rudnev, M. (maksim.rudnev@gmail.com), 71Russo, L. (l.russo@sssup.it), 72Rust, K. (keithrust@westat.com), 135Rusu, A. (ioanaalexandra.mihai@g.unibuc.ro), 165Ruviglioni, E. (elena.ruviglioni@unifi.it), 72, 73Sakshaug, J. (joesaks@umich.edu), 142, 162Sala, E. (emanuela.sala@unimib.it), 68, 105, 125, 141, 143,147, 160, 175Salagean, I. (ioana.salagean@ceps.lu), 167Salfinger-Pilz, B. (brigie.salfinger-pilz@stask.gv.at),Salheiser, A. (axel.salheiser@uni-jena.de), 187Sapin, M. (marlene.sapin@fors.unil.ch), 93, 99, 189Saris, W. (w.saris@telefonica.net), 132, 151Sarracino, F. (f.sarracino@gmail.com), 115Sarrasin, O. (oriane.sarrasin@unil.ch), 83Sastry, N. (nsastry@umich.edu), 61Sato, Y. (yusato@nies.go.jp)Saelberger, S. (sabine.saelberger@destas.de)Sawinski, Z. (zsawins@ifispan.waw.pl)Schaan, B. (schaan@mea.uni-mannheim.de), 49, 74, 75Schaeffer, M. (merlin.schaeffer@gmail.com), 164Schenk, M. (michael.schenk@uni-hohenheim.de), 166Scherpenzeel, A. (a.c.scherpenzeel@uvt.nl), 90, 110, 124,161,Scheuren, F. (scheuren-fritz@norc.org), 109Scheve, C. von (chrisan.von.scheve@fu-berlin.de), 62Schlueter, E. (elmar.schlueter@wiso.uni-koeln.de), 138, 154,155, 166, 187Schmeets, H. (hshs@cbs.nl), 147Schmich, P. (schmichp@rki.de), 93Schmidt, A. (alexander.schmidt@wiso.uni-koeln.de), 127Schmidt, C. (christopher.schmidt@uni-bamberg.de), 103Schmidt, P. (peter.schmidt@sowi.uni-giessen.de), 36, 57,70, 87, 98Schmidt, T. (tobias.schmidt@bundesbank.de), 51, 129Schmitz, R. (schmitzr@rki.de)Schneider, S. (silke.schneider@nuffield.ox.ac.uk), 99Schnell, R. (rainer.schnell@uni-due.de), 23, 43, 94, 143Scholz, E. (evi.scholz@gesis.org), 38, 134Schönberger, B. (benno.schoenberger@uni-bamberg.de), 149Schonlau, M. (spiel67@ymail.com), 171Schori, D. (dschori@ispm.unibe.ch), 148, 161Schouten, B. (bstn@cbs.nl), 183Schröder, H. (h.schroder@fsw.vu.nl), 100

LIST OF PARTICIPANTS 197Schröder, J. (jee.schroeder@uni-mannheim.de), 82Schroeder, M. (mschroeder@diw.de), 143Schübel, T. (mail@schuebel-thomas.de), 180Schubotz, D. (d.schubotz@qub.ac.uk), 29Schupp, J. (jschupp@diw.de), 37, 50, 62, 76Schwartz, D. (schwartz@quinnipiac.edu)Seiler, C. (seiler@ifo.de), 106Selb, P. (peter.selb@uni-konstanz.de), 144, 190Sellke, P. (piet.sellke@sowi.uni-stugart.de), 89Sengewald, E. (erik.sengewald@uni-jena.de)Serban, M. (monas@iccv.ro), 165Siegel, M. (melissa.siegel@maastrichtuniversity.nl), 132Siegel, N. (nico.siegel@tns-infratest.com)Siegers, P. (siegers@wiso.uni-koeln.de), 36Simonson, J. (julia.simonson@dza.de), 163Sinibaldi, J. (belladori@hotmail.com), 121Siponen, K. (katri.siponen@uta.fi), 113Skarbek-Kozietulska, A. (skarbek@uni-mainz.de), 27, 51Skinner, C. (cjs@soton.ac.uk)Slavec, A. (ana.slavec@fdv.uni-lj.si), 65, 174Slegrova, A. (aslegrova@scio.cz), 188Smit, J. (jh.smit@vumc.nl), 28, 42, 68, 86Smith, K. (k.smith@ioe.ac.uk), 39, 40, 60, 72, 73Smith, T. (smitht@norc.uchicago.edu), 86, 158smith, p. (paen.smith@ipsos-mori.com)Spiess, M. (marn.spiess@uni-hamburg.de), 131, 148Springer, S. (sabine.springer@fra.europa.eu)Stares, S. (s.r.stares@lse.ac.uk), 41, 107, 127Stathopoulou, T. (stathopoulou2@yahoo.com), 94Stavrova, O. (stavrova@wiso.uni-koeln.de), 103Stec, K. (kasia.stec@uj.edu.pl), 73Stegmueller, D. (d.stegmueller@gmail.com), 138, 159Steiner, P. (psteiner@wisc.edu), 130, 188Steinhauer, H. (hans-walter.steinhauer@uni-bamberg.de),150,Steinmetz, S. (s.m.steinmetz@uva.nl), 184, 185Stocké, V. (volker.stocke@uni-bamberg.de), 31, 45, 60, 136,Stoneman, P. (p.stoneman@soton.ac.uk), 84, 128Stoop, I. (i.stoop@scp.nl), 78, 94, 121, 140, 146Storfinger, N. (nina.storfinger@zeu.uni-giessen.de), 63, 64,Stracca, L. (livio.stracca@ecb.int), 41Strack, M. (mstrack@uni-goengen.de), 87, 88Streeerk, M. (m.ij.c.streeerk@uvt.nl), 74Struminskaya, B. (bella.struminskaya@gesis.org), 125Stubbe, T. (stubbe@ifs.tu-dortmund.de), 149Stuck, S. (stuck@mea.uni-mannheim.de), 69Studer, R. (raphael.studer@econ.uzh.ch), 169Sturgis, P. (p.sturgis@soton.ac.uk), 84, 108, 124, 135SUGINO, I. (sugino.isamu@ocha.ac.jp)Sundgren, B. (bo.sundgren@gmail.com), 109Suparman, Y. (yusep.suparman@unpad.ac.id), 95Szczucka, A. (ania.szczucka@gmail.com), 150Szendrő, Z. (szz@mtap.hu), 180Sztabinski, F. (fsztabin@ifispan.waw.pl), 141Sztabinski, P. (psztabin@ifispan.waw.pl), 141Tam, H. (soc.tam@gmail.com), 168Tangian, A. (andranik-tangian@boeckler.de), 190Tart, I. (itart@tlu.ee), 58Taskinen, P. (per.taskinen@stat.fi), 65Tawara, K. (kimit@kenroku.kanazawa-u.ac.jp)Teney, C. (teney@wzb.eu), 144, 177Terwey, M. (michael.terwey@gesis.org), 89Thamm, M. (thammm@rki.de), 33Thomas, R. (rthomas@icfi.com), 62, 71, 88, 168Thomas, W. (wlt@umn.edu), 73Thörner, S. (stefan.thoerner@staff.uni-marburg.de), 47Thorsdor, F. (fanneyt@hi.is)Thygesen, L. (lth@dst.dk), 109Tijdens, K. (jdens@fsw.eur.nl), 123, 184, 185Tobi, H. (hilde.tobi@wur.nl), 35, 169Toharudin, T. (toni.toharudin@unpad.ac.id), 86, 107Tolonen, H. (hanna.tolonen@thl.fi), 24tommasi, b. (beatommasi@scali.it), 72Tormos, R. (rtormos.ceo@gencat.cat), 176Torres, A. (analia.torres@iscte.pt), 34, 78Tourangeau, R. (rtourang@umich.edu), 108Trappmann, M. (mark.trappmann@iab.de), 27, 142, 162Tries, S. (simone.tries@destas.de), 170Trimarchi, E. (elaine@mktginc.com), 161Trujillo Carmona, M. (mtrujillo@iesa.csic.es), 174Tsai, L. (tlt@ntcu.edu.tw), 131Tsuchiya, T. (taka@ism.ac.jp), 47Tufis, C. (ctufis@gmail.com), 139Tufis, P. (ptufis@iccv.ro), 127, 165Turner, G. (m.m.turner@soton.ac.uk), 108Turner, S. (sturner@forsmarshgroup.com), 69, 114, 118, 125Tule, A. (alfred.d.tule@census.gov), 118Uhrig, N. (scnuhrig@essex.ac.uk), 105, 125, 141, 147, 160Upsing, B. (upsing@dipf.de), 100Uzaheta, A. (auzaheta@icfes.gov.co), 59, 150Vaart, W. van der (wvdv@uvh.nl), 28, 68, 148Vakalia, F. (fvakalia@forsmarshgroup.com), 114, 125Valentova, M. (marie.valentova@ceps.lu), 85Vannieuwenhuyze, J. (jorre.vannieuwenhuyze@soc.kuleuven.be),183Vardigan, M. (vardigan@umich.edu), 53Vecchione, M. (michele.vecchione@uniroma1.it), 71, 87Veen, F. van (floris.van_veen@uni-bielefeld.de), 76, 94Vehovar, V. (vasja.vehovar@fdv.uni-lj.si), 34, 64, 65, 174Veerlein, A. (anja.veerlein@uni-jena.de)Veykher, A. (aavej@rambler.ru), 172Vicente, P. (paula.vicente@iscte.pt), 158, 173Vis, C. (c.m.vis@uvt.nl), 110Vis, R. (rachel.vis@cbs.nl), 122Vogl, S. (susanne.vogl@ku-eichstae.de), 27, 47Voicu, B. (bogdan@iccv.ro), 154Voicu, M. (malina@iccv.ro), 153Voorpostel, M. (marieke.voorpostel@fors.unil.ch)Wackerow, J. (joachim.wackerow@gesis.org), 54, 112Waheedi, M. (mohdw@hsc.edu.kw)Wallace, E. (emma.wallace@ipsos.com), 40, 73Walsoe, H. (heidi.walsoe@tns-gallup.no)Warner, U. (uwe.warner@ceps.lu), 79, 99, 112Watmuff, R. (ross.watmuff@abs.gov.au), 146Watson, N. (n.watson@unimelb.edu.au), 126Waeler, O. (oliver.waeler@gesis.org), 56

198 LIST OF PARTICIPANTSWäyrynen, L. (laura.wayrynen@capstan.be), 100Weber, A. (anne.weber@uni-due.de), 106Weber, W. (wiebke.weber@upf.edu), 133Weerdt, J. De (info@surveybe.com)Weichbold, M. (marn.weichbold@sbg.ac.at), 70Weinhardt, M. (mweinhardt@diw.de), 161Werorst, H. van de (h.g.vandewerorst@uva.nl), 138Widdop, S. (sally.widdop.1@city.ac.uk), 41, 80Wiegand, G. (gordon.wiegand@nw.ch), 185Wijnant, A. (wijnant@uvt.nl), 110Wilkes, R. (wilkesr@interchange.ubc.ca), 41, 176Willis, G. (willisg@mail.nih.gov), 122, 123Willis, R. (rjwillis@umich.edu), 90Willson, S. (swillson@cdc.gov), 92Winker, P. (peter.winker@wirtscha.uni-giessen.de), 63, 64,Wolf, C. (christof.wolf@gesis.org), 72, 158, 159, 176Wolff, N. (wolff@arcadia.edu), 164Wolter, F. (felix.wolter@uni-mainz.de), 27, 51Yang, K. (kemingyang@googlemail.com), 181Young, C. (christopher.young@unifr.ch), 163Young, R. (rly116@psu.edu), 46, 126Zandberg, I. (izandberg@aol.com)Zeglovits, E. (eva.zeglovits@univie.ac.at), 46Zelený, M. (zelenym@vse.cz), 77Zheng, M. (mlzheng@umac.mo), 76Ziegenfuss, J. (ziegenfuss.jeanee@mayo.edu), 54Zmijewska-Jedrzejczyk, T. (teresa.zmijewska@ifispan.waw.pl),94,Zuber, S. (zuber@mea.uni-mannheim.de), 69Zuell, C. (cornelia.zuell@gesis.org), 134

199CoverDesign:KevinWilliamsCoverPhoto:2011iStockphoto(vladacanon)Content:LocalOrganizers/ESRAPrintedby:GerberDruckAGOriginalCopies:700Lausanne,June2011„Due to limited space in the programme book the editors were forced to shorten the abstracts. You can findthe full version of each abstract on the ESRA conference website“

Joao Monteiro © UNILESRA would like to thank the following organisations who have sponsored this conference:Canton of VaudcApStAn Linguistic Quality ControlCity of LausanneData Documentation InitiativeEconomic Development Initiatives (EDI) LimitedGerman Data Forum (RatSWD)German Socio Economic Panel Study (SOEP)Methodology Institute at the LSEM.I.S. TrendNational Centre for Research Methods (NCRM)Routledge publishersSAGE publishersSwiss Household PanelSwiss National Science Foundation (SNSF)TNS P&SUniversity of Duisburg-EssenWestat