PISA 2009 Hadjar

dronkersj
  • No tags were found...

PISA 2009 Hadjar

Researchcentrum voor Onderwijs en Arbeidsmarkt Systems’ Tracking, Schools’ Entrance Requirements and Educa:onal Performance by Migrants’ Children. Jaap Dronkers & Roxanne Korthals


Debates about effects stra:fied systems 1. Two levels (pupils; systems) or three levels (pupils; schools; systems): a. Schools features are formed by systems; b. schools are importance variance source; c. tracks have also different curricula. 2. Effects of educaBonal systems can be biased without early ability and role of ability & SES in selecBon into tracks. 3. Necessary to disBnguish between 1 st , 2 nd & 3d SES effects: 1 st ability; 2 nd parental choices; 3d teachers’ decisions


Examples 3-­‐level analyses -­‐ A. Dunne. 2010. Dividing Lines: Examining the Rela4ve Importance of between-­‐ and within-­‐School Differen4a4on during Lower Secondary Educa4on. Ph.D. European University InsBtute (Florence). -­‐ J. Dronkers, R. van der Velden & A. Dunne. The effects of educaBonal systems, school-­‐composiBon, track-­‐level, parental background and immigrants’ origins on the achievement of 15-­‐years-­‐old naBve and immigrant students. A reanalysis of PISA 2006. ROA-­‐RM-­‐2011/6, Maastricht University -­‐ R. Korthals. SelecBon and tracking in secondary educaBon: A cross country analysis of student performance and educaBonal opportuniBes, ROA-­‐RM-­‐2012/14, Maastricht University. -­‐ T. Bol, J. Witschge, H. van de Wer[orst & J. Dronkers. 2014. Curricular Tracking and Central examinaBons: Counterbalancing the Impact of Social Background on Student Achievement in 36 Countries. Social Forces


Debates about migrants and educa:on 1. Pupil, school & system features can have different meaning for migrants and thus their performance. 2. Migrants come from different origin countries. 3. Migrants are not equally distributed across desBnaBon countries. 4. Effects of macro characterisBcs of desBnaBon countries can be biased without migrants’ origin.


Examples of analyses migrant performance & origin -­‐ M. Levels, J. Dronkers & G. Kraaykamp, 2008. “Immigrant Children’s EducaBonal Achievement in Western Countries: Origin, DesBnaBon, and Community Effects on MathemaBcal Performance.” American Sociological Review 73: 835-­‐853 -­‐ J. Dronkers, M. Levels & M. de Heus. 2013. “Migrant pupils’ scienBfic performance: the influence of educaBonal system features of origin and desBnaBon countries” Large-­‐scale Assessments in Educa4on 1: 10 -­‐ J. Dronkers, R. van der Velden & A. Dunne. 2012. “Why are Migrant Students Beier Off in Certain Types of EducaBonal Systems or Schools than in Others?”European Educa4onal Research Journal 11(1): 11-­‐44.


Aims of the new analysis of key note 1. Cross-­‐naBonal analysis to asses system effects with 3-­‐level model (pupil, school, system). 2. Including role of ability selecBon in schools and tracks ≅ 1 st SES effect 3. Taking migrants’ origin into account. 4. Differences in parameters of system-­‐, school-­‐ & pupil characterisBcs of naBve students versus 1 st & 2 nd generaBon migrants’ pupils.


Tracking SeparaBng students into disBnct educaBonal programs in secondary school based on ability As opposed to separaBng students in ability groups (within classes or for some courses) No tracking countries: US, France, Sweden, Finland, Denmark, etc. Tracking countries: Germany (4 tracks), the Netherlands (4), Slovak Rep (5), Italy (3), ect. Much debate on tracking in Europe (lately esp. Germany, Flanders), especially in relaBon to migrants (segregaBon).


Entrance requirements tracks How oken was consideraBon given to a student's record of academic performance (including placements tests) and to feeder school recommendaBons in admimng the student to the school? Never, someBmes, always QuesBon answered by school principal (PISA 2009)


Data & Method PISA 2009 31 countries: 188.138 naBve, 11.018 2 nd generaBon & 9.786 1 st generaBon students. -­‐ “Rich” countries (GDP above OECD minimum) -­‐ school data available (≠ France, Canada, UK) 15 countries with birth country informaBon of pupil & parents: 74.588 naBve, 5.180 2 nd generaBon & 7.609 1 st generaBon students. Restricted birth countries per test-­‐country & other. CombinaBon of birth county of pupils & parents. 13 origin regions, which combines 54 origin countries. Method: 3 level random effects; Missing values replaced by averages & dummies


Independent variables


Next steps 1. Three level model for naBve & migrant pupils in 31 & 15 countries 2. Effects of origin region 3. Differences in system parameters of naBve and migrant pupils 4. Differences in school features parameter of naBve and migrant pupils 5. Differences in pupil characterisBcs of naBve and migrant pupils


Systems & entrance requirements for na:ves 5 READ: na:ves 0 1 2 3 4 5 -­‐5 -­‐10 never someBmes always -­‐15 -­‐20 -­‐25 • PosiBve effect of many tracks on performance, but only if schools consider prior performance always. • No differences with 1 or 2 tracks.


Systems & entrance requirements 1 st gen. 31 coun. 10 READ: 1st genera:on migrants 5 0 -­‐5 1 2 3 4 5 never someBmes always -­‐10 -­‐15 No significant effects of tracks, requirements or interacBons


Systems & entrance requirements 2nd gen. 31 coun. 40 READ: 2nd genera:on migrants 35 30 25 20 15 10 never someBmes always 5 0 -­‐5 1 2 3 4 5 PosiBve significant effect of tracking on performance, but only if schools consider prior performance always


Tracks & Parental Background in 31 countries Tracking decreases effect of parental background for naBve & 2 nd migrant pupils if achievement entrance requirements of schools are included


Origin of migrants included for only 15 countries Origin irrelevant for naBves, because they have no other country than desBnaBon country. Only migrants have an origin ≠ desBnaBon Fixed effects for 13 origin regions in analyses for 15 countries. Origins regions are unequally distributed across desBnaBon countries: history, chain migraBon, selecBon of migrants, openness towards migrants, path dependence. Different origins in 15 than in 31 countries, but difference is unknown.


Na:ves: 31 versus 15 countries 5 0 -­‐5 -­‐10 -­‐15 -­‐20 -­‐25 READ: na:ves 1 2 3 4 5 never someBmes always 10 5 0 -­‐5 -­‐10 -­‐15 -­‐20 -­‐25 READ: na:ves (15 cnts) 1 2 3 4 5 never someBmes always


2 nd genera:on with origin fixed effect 20 READ: 2nd gen (15 cnt, with origin) 10 0 -­‐10 -­‐20 1 2 3 4 5 never someBmes always -­‐30 -­‐40 -­‐50 One track AND many tracks if schools consider prior performance ALWAYS > more than 1 tracks if schools NEVER consider prior performance


5 1 st genera:on with origin fixed effect READ: 1st gen (15 cnt, with origin) 0 -­‐5 1 2 3 4 5 -­‐10 -­‐15 -­‐20 never someBmes always -­‐25 -­‐30 -­‐35 -­‐40 • PosiBve significant effect of many tracks on performance, but only if schools consider prior performance always. • No difference for few tracks whatever school placements.


Tracks & Parental Background with origin fixed effects Many tracks decreases effect of parental background for 1 st generaBon migrant pupils if achievement entrance requirements of schools are included; neutral 2 nd generaBon


Origin effects


School features parameters for na:ve, 1 st & 2 ndgenera:on pupils with origin fixed effects


Individual characteris:cs parameters for na:ve, 1 st & 2 nd genera:on pupils with origin fixed effects


Conclusion Tracking & School Requirements • 3 level-­‐analyses migrant pupils show comparable result as naBve pupils: pupils in educaBonal systems with many tracks in schools with achievement entrance requirements have equal or higher scores than pupils in 1 track systems with or without achievement entrance requirements. • 3 level-­‐analyses of migrant pupils show comparable result as naBve pupils: pupils in educaBonal systems with many tracks in schools with achievement entrance requirements have higher scores than pupils in many track systems in schools without achievement entrance requirements. • Many tracks decrease effect of parental background for 1 st generaBon migrant pupils, if achievement entrance requirements. • Inclusion of school level variance & achievement entrance requirements of schools = controlling for primary SES effect (on ability).


Conclusions origin, school and individual features • Origin effects are substanBal even aker control for “usual suspects”, and thus needs further analyses. • Origin differences between desBnaBon countries influence comparison outcomes, but do not overthrow outcomes fully. • School and individual characterisBcs have someBmes different parameters for naBve than for migrant pupils; both stronger & weaker. No clear paiern. • Migrant pupils are not just another kind of low SES pupils


Educa:onal system research • Systems and school effects studies should focus on secondary & terBary SES effects and disBnguish these from primary SES effects. • 2 level measurement of system effects with cross-­‐secBonal data can not disBnguish between these types of SES effects. • 3 level measurement of system effects with cross-­‐secBonal data can at best only indirectly disBnguish between these SES effects. • Joseph J. Merry, “Tracing the U.S. Deficit in PISA Reading Skills to Early Childhood: Evidence from the United States and Canada” Sociology of Educa4on July 2013, vol. 86 no. 3 234-­‐252 • Primary SES effects are not produced by educaBonal system characterisBcs but by general societal macro-­‐characterisBcs (health care, child health, racial/ethnic composiBon, family structure). • A beier explanaBon of Nordic high PISA scores and recent decline?


Longitudinal data • We need urgently longitudinal cross-­‐naBonal data. • NaBonal longitudinal data can disBnguish between primary, secondary & terBary SES effects and analyze their variance in tracks/schools/systems with less or more achievement selecBon. • NL example: Parental background, early scholas4c ability, the alloca4on into secondary school tracks and language skills at the age of 15 years in a highly differen4ated system: a test of the contradic4ons between a two-­‐ or three-­‐level approach. ROA Technical Report 2014/1. • Upcoming DE examples with NEPS & BiKS data.


1. 31 & 15 countries 2. Entrance requirements school and N of tracks 3. Regions of origin Appendices


31 & 15 countries • Argen


Entrance requirements & N of Tracks Number of tracks Percentage of students in schools that consider prior performance for student acceptance never some4mes always Number of countries 1 55 33 12 10 2 27 39 34 3 3 32 31 37 9 4 21 28 51 7 5 18 23 59 2


Regions of Origins: UN geographical subregions • No central America: Mexico=South America. • Africa is only divided in North Africa and sub-­‐Saharan Africa. • Samoa (Polynesia) added to Oceania (Australia & New Zealand).

More magazines by this user
Similar magazines