Evolution of Audio Recording in Field Surveys - RTI International

Evolution of Audio Recording in Field SurveysM. Rita Thissen, Sridevi Sattaluri, Emily McFarlane, and Paul P. BiemerRTI International, Research Triangle Park, NC 27709Abstract:The tools of field survey administration change quickly.By taking advantage of new technology and adapting itfor time-honored needs, survey managers can boost theeffectiveness, efficiency and quality of data collection.One method which has evolved rapidly is computeraudio-recorded interviewing (CARI), an approach toensuring the quality of data through unobtrusiverecording by the computer of the audio portion of inpersoninterviews, much as silent monitoring has beenused to ensure quality at call centers.Several developments in the past few years haveimproved the technical feasibility of CARI for routineand inexpensive use in field studies. Advances in filecompression and available bandwidth enable collectionof longer recordings with little strain on transmissioncapacity and no burden to the interviewer. Use of asimple external file for specifying items to be recordedin a Blaise instrument offers great flexibility in selectingportions of the interview for auditing, even permittingmodification of the recorded-item list while aninstrument is in production. A web-based monitoringapplication, for use by trained reviewers in evaluatingthe audio files, can now provide access to centrallylocated audio files by geographically distributed staff.Progress has also been made from an operationalviewpoint. Work has been done to determine theminimum amount of recording needed to achieveagreement among reviewers as to the authenticity of therecorded session, and cost modeling shows that CARIcan provide quality assurance at equal or reduced costscompared to more traditional approaches of re-interviewor telephone verification.Use of CARI on several national surveys has providedproduction experience to bolster laboratory tests. Thisarticle reviews the progress of CARI technology in theyears since it was introduced, with an emphasis onfeasibility for routine use with field surveys.Key Words: survey technology; audio recording;computer audio-recorded interviewing (CARI); soundfile; quality assurance; performance management; fieldinterview; in-person interview1. IntroductionMonitoring the performance of field staff and thequality of data collection has been challenging since theearliest in-person surveys. Traditionally, field staffhave worked largely unobserved, with occasionalshadowing by supervisory personnel or re-contactingthe respondent to confirm the interview’s authenticityand inquire about the professionalism of the interviewer.It can also be difficult to evaluate or confirm theeffectiveness of questionnaire items in field surveys,whether from a usability perspective, such as the abilityof interviewers to read the questions in a fluent andunderstandable manner, or from the perspective ofclarity for the respondent, such that the responseprovides the desired information without the need forexplanation or probing. While focus groups orcognitive interviews in advance of data collection mayoffer insight into presentation and response patterns, thepractice does not fully anticipate field conditions.Now the situation has changed. Many computers nowhave audio recording capabilities, and some have builtinmicrophones. With this technology, surveys can beset up to collect digital audio recordings in anunobtrusive manner while the interview is taking place.With computer audio-recorded interviewing (CARI),sound files can be created electronically without theneed for external equipment and can be transmittedalong with response data files and tracking information.Because the recording process is “invisible,” onceconsent has been given, it can provide a faithfulrepresentation of the reality of in-person data collection.The technology provides a potent tool for deterring anddetecting falsification, providing performance feedbackand enabling study of questionnaire item effectiveness.2. Audio Recording Technology, Past and PresentFrom the marketing of the Dictaphone in 1907 (NuanceCommunications 2005) to the availability of miniaturerecorders embedded in portable electronic devices today(Dwyer et al,1998), people have been discovering waysto take advantage of audio recording tools to capturevoices for later review. While the early acousticrecorders proved helpful for journalistic interviews, theywere not usable for large-scale research surveys; theintroduction of cassette tapes improved convenience forinterviewing (Stockdale, 2002).With the advent of digital recording, and as computersbegan offering built-in sound cards, the task ofcapturing audio records became easier and sound

quality improved. Sound files now can be recordedelectronically through sound cards and software onlaptops, handhelds and other portable devices, makingthis technology handy for use in field surveys.Figure 1. Milestones of audio recording historyAudio Storage Invention WidespreadUseWax cylinder 1885 Early 1900’s to1940’sMagnetized wire 1898 1940’sMagnetic coatingson plastic tape1928 1930’s topresentCompact cassette 1963 1960’s to 1990’sPulse code1937 1990’smodulationDigital audio 1971 1990’smicroprocessorSoundBlaster audio 1989 1990’scardFirst use of CARI 1999 PresentPortable digitalvoice recorders2003 PresentIn 1999, use of digital audio recording was firstdeveloped and deployed on a national field survey, asthe result of innovative work by RTI developers R.Suresh, A. Bethke and P. Cooley. Use of CARI hasgrown since then, as the feasibility and utility of theapproach have been confirmed. Electronic recordingrequires little attention during the interview, as there areno tapes to change, no additional equipment to set upand no distraction during the interview. Feedback fromrespondents and interviewers indicates that most peopleforget about digital recording when the microphone isinternal, once the interview gets underway.Many laptops now have built-in microphones, soundcards and adequate disk space for conducting audiorecording. Handheld digital recorders from somecompanies offer audio recording capabilities butfunction much like analog tape recorders, requiring theuser to switch them on and off manually. A fewhandhelds offer programmed recording capabilities plusan internal microphone. Most laptop and many handheldcomputers allow use of an external microphone insteadof the internal one, for improved audio fidelity.However, the visible hardware calls attention to therecording process and may be more likely to affect therespondent’s and interviewer’s behavior.Audio fidelity from any device depends on a number offactors. When using a laptop with internal microphone,these include• Placement of the microphone with respect to noiseproducinghardware (keyboard, fan and disk drive)• Placement of the microphone with respect to theinterviewer and respondent.• Microphone control settings.Some internal microphones are adequate to capturevoices within 8 feet or so of the laptop when configuredproperly, at a quality level that allows a listener todistinguish among multiple voices and discern thespoken content.3. Audio Recording for Quality AssuranceAlthough there are several advantages offered byimplementing CARI, perhaps the most compellingreason is to confirm the authenticity of data for areduced cost compared to traditional verificationmethods. CARI can act as a deterrent to curbstoning andas a tool for detecting questionable interviews.Interviewers who are aware that monitors may listen toparts of each interview may be less likely to falsify data,because the audio file acts as a “witness” to theiractions. In this way, the simple presence of CARI canreduce cheating.Speech patterns heard in audio files provide informationto the monitors about the veracity of the interview, asindicated by the timing and phrasing of questions andresponses. In a normal interchange, people pausebetween words, phrases or sentences, as they considertheir answers or express their views (Kowal et al, 1975;O’Connell and Kowal, 1983).Figure 2. Indicators of questionable authenticitySilence No voices can be heard, although roomnoises and key clicks are audibleMumbling The interviewer can be heard, but appearsto be speaking to him or herselfUnnatural The respondent answers too quickly orpatterns laughs in inappropriate placesComments The respondent or interviewer makescomments suggesting the interview isbeing falsifiedSame The same respondent’s voice is heard invoice multiple interviews or does not match thestated sex or age of the respondent.For example, when an interviewer acts alone andfalsifies data, there may be no voice at all in therecording or only one voice without the expectedpausing, inflection or clarity of voice which would beexpected in a two-way exchange. If the interviewerenlists someone to pose as the respondent, theaccomplice may display inappropriate attitudes oremotions, make unexpected remarks, respond without

pausing to understand the question or pause at unnaturalplaces while inventing an answer. CARI monitors listento the recordings, and quickly become adept atdistinguishing between recordings of normalinterviewing circumstances and suspicious ones, bylistening for characteristics such as those in Figure 2.(Thissen and Rodriguez, IBUC 2004)Using CARI, a survey organization may reduce its recontactefforts and costs. CARI monitoring may replacemost telephone verification calls or field re-interview.However, it remains important to have a second meansfor following up a small sample of the cases since somerespondents may refuse to allow audio recording, andinterviewers may attempt to use that option to preventdetection of poor interviewing habits or curbstoning.The benefits of CARI plus optional re-contact aretwofold: it tells the interviewing staff that they cannotavoid monitoring even if they discourage theirrespondents from allowing CARI, and it allowscomparison of the two approaches to confirm thevalidity of the results.4. Data Collection MethodologyAnother benefit of CARI is that it provides a method foridentifying questionnaire problems and data collectiondifficulties in interviewer-respondent interactions. Fieldstaff do not always conduct interviews in an optimalmanner, and it can be difficult to obtain reliableinformation about their performance. While personalobservation can provide a wealth of information, thepresence of an observer may bias the evaluation. CARIoffers a unique opportunity to listen to the interviewexactly as it took place, without observation effects.During the first few weeks in the field, feedback can bean important tool for reinforcing lessons learned duringtraining. A CARI monitor may be able to providefeedback, either praise or constructive criticism, aboutthe way in which the interview was conducted.Improper question administration which can be detectedthrough CARI includes• Paraphrasing• Improper probing• Suggested responses• Poor enunciation• Improper commentaryCARI can be used to identify positive behavior such as• Precise adherence to protocol• Adept handling of difficult situations• Consistency, honesty, and professional behaviorCARI can also be used to evaluate the usability ofquestionnaire items. The audio recording of aninterviewer’s presentation of an item and the subject’sresponse provides a clear indication of whether the itemsucceeds in several ways:• Readability – based on the interviewer’s fluency inpresenting the item• Clarity of content – based on the respondent’s easeof understandingSurvey items which evoke negative reactions or requirefrequent explanations are detrimental to the responserate and increase the level of burden on both interviewerand subject. Using CARI, especially during field testingof an instrument, allows the survey specialist to evaluatethe success of the questionnaire items in eliciting thedesired information.5. Privacy, Security, Consent and LegalitiesFor CARI to be used during an interview, participantsmust give express consent for the interview to berecorded. Respondents are told that their participation isvoluntary and that their information and responses areconfidential and will only be used for statisticalpurposes. In two national field studies using CARI,approximately 83% of respondents in one survey on ahighly sensitive topic agreed to allow the interview tobe recorded, and 93% of respondents on another lesssensitive survey agreed. (Wrenn-Yorker and Thissen,FedCASIC 2005). For those who do not allowrecording, traditional verification methods such astelephone verification interviews are used.All survey data, including CARI recordings, must besafeguarded. In addition to design considerations basedon user needs, careful attention must be paid to securityand privacy issues when dealing with human data. Inthe United States, laws and regulations direct themanagement of personal identification information,health records and other specific types of data.Computing professionals must be aware of federal, stateand local requirements for confidentiality in storage,transmission and release of personal information.Institutional review boards oversee all research data onhuman subjects, to ensure that the studies contribute tothe greater good without harming individuals. Tocomply with guidelines and regulations, informationsystems may need to include authenticationmechanisms, audit histories and user records. These areregulatory rather than usability requirements but areessential components of survey information systems.Given heightened consciousness of confidentiality andsecurity concerns, care is required in handling audiorecordings. Even though the survey may not deliberatelyrecord personally identifying information, it cannot beguaranteed to avoid it. For this reason, audio files arebest treated as sensitive data, much the way response

data is handled. Encryption may be desirable for digitalrecordings, and special care may need to be taken inhandling tapes, if the recording is by analog device.6. Audio File FormatsDigital audio recording can take place with variouslevels of sound quality, and the resulting files may bestored in various electronic formats. The soundrecording algorithm affects the following:• Audio file size and storage requirements• Required software for recording and playback• Quality of sound on playback• Platform requirements and CPU demands• Cost and licensing issues.Many audio file formats have been developed over theyears, and their sheer variety may seem baffling to thenew observer. Recent attention has been given to mp3(Motion Picture Group Audio Layer 3) format, butmany other file formats exist as well. A few of thecommon formats are listed in Figure 3.Microsoft Windows operating systems include SoundRecorder software which writes to the wave file format,and the Windows Media Player which can play backwave files and a number of other non-proprietaryformats. The PCM (pulse code modulation) digitalrecording algorithm is used in various encodersincluding Sound Recorder, and records uncompressedsound with no required licensing.Figure 3. Common audio file formatsNameFile UseExtensionWave .wav WindowsuncompressedMP3 .mp3 Compressed audioRealMedia .rm Compressed audioRealAudio .ra Compressed, forstreaming audioAIFF .aiff Macintosh defaultuncompressedCD Audio .cda Music CD tracksActive StreamingFormat.asf Streaming audioWave files are not especially efficient at storage, but therecording process places little demand on the computer.The size of a particular wave file depends on therecording parameters selected in its creation. For eachavailable audio file format, there is a choice of samplingrate, bandwidth, number of channels and otherparameters. For RTI’s current CARI system, thestandard configuration is 16 bit bandwidth, 11.25 KHzsampling rate and a single channel. Recording twochannels (stereo) would require twice the storage spaceand provides no extra quality since a single microphoneis generally used. Audio quality also is affected bysampling rate, compression and audio file format, andthe settings given above are minimal for useful files.To reduce the space for audio files, a compressiontechnique may be used. Coder-decoder algorithms(CODECS) offer ways to store recordings in less space.They eliminate silence and mathematically map thesampled analog sound frequencies instead of preservingthe actual data points. CODECS may be employed as apost-processing step after creation of the sound file or asa real-time action at the time of recording.CODECS (compression–decompression techniques)were developed for use with audio recording, to reducethe size of sound files. It is possible for audio recordingto combine the digitizing process and compression atonce. For use in surveys, the system designer canchoose among simple recording with no compression,simultaneous recording and compression or recordingfollowed by compression. Section 11 discusses theseapproaches in a comparison of post-recordingcompression with simultaneous recording andcompression7. Integrating Audio Recording with SurveySoftwareA variety of technologies have been in use to implementsurvey instruments, such as Blaise (StatisticsNetherlands), CASES (University of California,Berkeley) and web-based technologies like ASP.NET(Microsoft). Audio recording components have beensuccessfully incorporated in all these environments.One of the challenges of incorporating audio recordingin a survey instrument is to make the processunnoticeable to the interviewer. The recording processmust not slow the system or provide any visual oraudible clue as to when it starts and stops.Audio recording can be added to Blaise instruments byusing either of two programming approaches. Oneapproach uses a Blaise procedure which in turn invokesan external application to start and stop the recorder.Using this approach requires complex programmingwithin Blaise in every place the recording applicationneeds to be invoked, to keep track of whether recordingis already in progress or needs to be started or stopped(Thissen and Rodriguez, IBUC 2004).The second approach uses the Blaise alien router.Starting with version 4.6, Blaise introduced the alienrouter as part of the Blaise component pack. The alien

outer technology allows the invocation of an externalcomponent before and after every survey item. Use ofthe alien router externalizes the complexities of trackingthe recorder state. It also opens up the possibility ofmaintaining a text list of items to be recorded, externalto the instrument. This reduces the complexity ofinstrument programming and allows easy modificationof the list of items to be recorded, without any need tomodify the data model or recompile the instrument(Thissen and Sattaluri, 2006b).For CASES instruments the recording can be integratedby spawning a separate application to start and stop anexternal recorder (Wrenn-Yorker and Thissen, 2005).When a survey is offered in multiple modes by using aweb-based instrument, field interviewing may takeplace through a website running on the laptop withoutcontinuous connection to the internet. In that case, theaudio recording component can be achieved byinstalling a client side Java applet and Java scripting,similar to the way in which CARI can be implementedfor internet-based surveys (Suresh, 2005)Once a survey instrument has been enabled with CARItechnology, survey information systems (Thissen, 2004)must also be expanded to handle the audio data files.From a case management and data security perspective,CARI files are no more than response data stored in adifferent format. Issues and concerns are the same forfiles containing audio response data as they are for filesof textual responses. File protection on the laptop,transmission to a central site, central storage, access byauthorized researchers and eventual deletion must all beplanned with the same security and confidentiality usedfor traditional response files.9. TransmissionThere are several options for transferring audio filesfrom the field laptop to a central management system.The files can be sent using dialup transmission,broadband, or removable media like flash drivesshipped by secure delivery methods. For small surveys,it may be practical to leave audio files on the laptopsuntil the end of data collection. With the pervasivenessof broadband access at homes through cable modem orDSL (digital subscriber line telephone service), thecapacity for transmitting large files has greatlyincreased. Still, researchers must plan for transmissionwhen using CARI, since audio files can be large.The choice of transmission option may depend on thesize of files being transmitted. It is found thatuncompressed audio recording consumes about onemegabyte of disk space for each minute of recordeddialog. (See Section 11 below for a comparison ofrecording parameters and file sizes.) Assuming aninstrument were programmed to collect three oneminuterecordings which were compressed to 100KBeach, the case management system would have 300KBto transmit for every case. If the interviewer transmitsone case each day, these files can be sent using a dialupconnection. The use of broadband allows transmitting alarger number of files or larger size files at a faster rate.The third option, using removable external media andshipment, can be used when entire interviews or lengthysections are recorded. However, security concerns, theeffort of handling external media and the possibility ofloss make this approach less desirable than automatictransmission via dialup or broadband. Still, it mayprove useful when recording interviews in their entiretyor when other forms of file transfer are not available.Audio recordings may contain personal identifyinginformation, whether by intention or by accident, and soit is important to protect these files by using encryptiontools while they reside in any location accessible tounauthorized individuals. In addition, if files aretransferred over the internet, secure socket layer (SSL)certification can be used, which provides a way toencrypt the data stream during transmission.10. CARI MonitoringAfter audio files are received at a central location, themonitoring process may be as simple as opening up thefiles using a free player tool like Windows Media Playeror Real Player. However, since manual casemanagement is impractical for all but the smallest ofsurveys, it is best to build a system that provides aninterface for reviewing the files and a database forrecording evaluations.The monitoring system might be a client-serverapplication or a browser-based application located on aninternal or external network. Client-server applicationsrestrict access to an organization’s internal network andlocally-located users, due to poor performance ofdatabase connections over long distance. A web-basedapproach has advantage of being available from anyworkstation which has access to the network, supportingorganizations with review staff distributed nationally oreven internationally (Thissen and Sattaluri, 2006a).Regardless of the implementation, it should providerole-based access to protect the security of theinformation stored in the audio files. For example,three levels of access might be designed into the system:• CARI monitoring staff, who listen to andevaluate audio files

• Supervisory staff, who designate monitors,manage caseloads and track review-completionstatus• System administrators who configure newsurveys and create new logins and passwords.For large surveys, the system may also include analgorithm to select a specified percentage of files to bereviewed per interviewer. Ideally, it would offer theflexibility to adjust review rates for any fieldinterviewer for any active survey, so that qualityassurance personnel can increase monitoring of anyinterviewer who has been suspected of improper datacollection practices. (Hartman et al, 2006)11. Audio and Operational ResultsIn this section, we present some results of RTI’sexperience with CARI technology. The data givenbelow were obtained by lab test, field test andproduction survey use of CARI processes.A comparison of recording alternatives is shown inFigure 4, with an indication of the resulting playbacksound quality. The column labeled “MB Per Min” liststhe number of megabytes of storage required for oneminute of sound when using the uncompressed wavefile format. Similar patterns of relative file size can befound for other file formats.Figure 4. Recording parametersBandwidthSampling ChannelsSoundQualityMB PerMin8 bit 11.25 KHz 1 Low 0.6616 bit 11.25 KHz 1 Medium 1.318 bit 22.5 KHz 1 Medium 1.7916 bit 22.5 KHz 1 High 1.1916 bit 44.1 KHz 1 Very 5.25High16 bit 44.1 KHz 2 VeryHigh12.3We have looked at alternative processes forcompressing existing audio files. A wave file wascompressed as a separate step after recording, using aspecific CODEC and selected recording parameters. Interms of a CARI system, this process might beperformed by the case management system after theinterview was completed but prior to transmission.Using this approach, compression ratios ranged from afactor of 2 to 75. In general, if the recording was ofvery high fidelity stereo, the original file would be verylarge and compress greatly. Lowering the recordingquality produces a smaller file originally butproportionally less compression.At RTI, files are recorded with the Windows nativeSound Recorder software called from Blaise or CASES,resulting in file sizes of about one MB/minuteuncompressed. Use of the LAME (The LAME Project)open source compression algorithm and appropriateparameters yields an average compression ratio ofapproximately 11:1 without loss of audio quality,resulting in about 100KB files for one minute of audio.Figure 5. File sizes obtained by concurrent recordingand compressionCODECInputSoundNumberOf FilesAverageMB/MinQuality TestedMPEGRec Low 4 0.98MPEGRec Mod 3 1.68MPEGRec V.High, 2 0.96MonoMPEGRec V.High, 1 1.80StereoRealMedia Low 24 0.34RealMedia Mod 3 0.51RealMedia V.High, 2 0.34MonoRealMedia V.High,Stereo1 0.47In another experiment, we recorded sound directly to acompressed format, without intervening storage as awave file. In a CARI system, this requires theinstrument to call a specific recording application andCODEC, such as MPEGRec (mp3), producing acompact file that is ready to encrypt and transmit. Thesimplicity of this approach was attractive becausecompression was immediate and effective, as shown inFigure 5. On the down side, simultaneous compressionand recording tax the computer’s processing power.This reduces system performance, produces lag andvisible indication of recording processes, and limits itsusefulness.Figure 6. Loudness Effect on File SizeFileFormatSoundLevelAveragedOver # ofMB PerMinuteFilesWave Silent 6 1.30Wave Quiet voice 9 1.31Wave Voice 6 1.32MP3 Silent 6 0.97MP3 Quiet voice 8 0.96MP3 Voice 6 0.97RM Silent 6 0.34RM Quiet voice 8 0.34RM Voice 6 0.34

We tested whether loudness had any effect on the sizeof the recorded output file by looking at the level ofsound in audio files compared to file size, for CARIfiles which where all recorded under identicalconfiguration settings on the same laptop. Figure 6shows the results of the comparison, demonstrating thatthere was no apparent effect of loudness on audio filesize.The quality of the sound files from the field is ofinterest, as an indicator of the feasibility of gatheringinformation for large numbers of interviews. Figure 7shows results from reviewing a sample of 11% of thefirst 1500 completed interviews from a survey. Theasterisk (*) indicates that the default rating was chosen,as opposed to an explicitly-defined score. Rating thefile quality rating was optional through the monitoringinterface if the quality was acceptable for review(Hartman et al, 2006).Figure 7. CARI sound file quality distributionSound Quality Number of Interviews1 – Poor 42 – Passable 53 – Adequate 21* – Acceptable 484 – Good 495 – Excellent 37Problems noted with audio files included backgroundnoise, static, faintness of voices, key tapping, hum andother recording problems which interfered withdetection of vocal content. Audio files were consideredadequate if voices could be plainly heard andunderstood, regardless of other noises. This definitionof quality differs from any commonly used to rate thequality of audio recording for other purposes, such asmusical entertainment, but it is appropriate for surveyevaluation purposes.Figure 8. Field performance problems detectedthrough CARICount % of Problem DefinitionCases13 0.2 Authenticity Questionable217 3.9 Reading - Minor Deviation72 1.3 Reading - Major Deviation73 1.3 Recording Errors44 0.8 Unprofessional Behavior86 1.5 Inappropriate Probing79 1.4 Feedback not Neutral1 0.01 Incorrect Incentive ProvidedWe have also gathered operational information on fieldstaff performance from production use of CARI. Figure8 shows the distribution of field performance problemsfound in one study after review of approximately 5600interviews. A single case might be assigned multipleproblem codes, and so the problem count total is greaterthan the number of affected cases (Wrenn-Yorker andThissen, FedCASIC, 2005).In general, field interviewers and respondents have beenaccepting of the technology. In a feedback study, 82%of interviewers felt neutral or positive about use ofCARI and a post-interview survey of 283 respondentsfound that 70% of the respondents reported they had noreaction one way or the other, 15% reported liking theidea, while 13% disliked the idea (Herget et al, 2005).As noted above, assent to CARI by respondents rangedfrom around 83% in one survey to 93% in another. Thisassent was independent of consent to conduct theinterview (Wrenn-Yorker and Thissen, 2005).A small experiment was conducted to determine theminimum number of CARI audio files required formaking consistent monitoring evaluations, that is, howmany audio files were required before reaching a pointwhere listening to additional audio files for an interviewhad no effect on the determinations. This worksuggested that three audio files each of 30-secondduration may be adequate for verification purposes.After review of three files, CARI monitors reached 97%agreement with the ratings found by review of five files,indicating that three files provide sufficient informationfor evaluation purposes.It is difficult to compare costs precisely between CARIoperations and more traditional re-interview orverification processes, because the traditional systemstend to be well established while CARI systems are stillevolving. A theoretical cost-analysis model was createdto compare the expected costs of operating both systemsat the same “steady state” in which all systems had beenimplemented. Analysis of that model suggests that thesteady-state cost of verification is less with CARI thanfor the traditional approach, but actual data were notavailable for that comparison.12. Visions of the FutureLooking forward, we see expanded use of CARI in fieldsurveys, for monitoring survey quality and also as anintegral part of data collection. Advances in digitalsignal processing may support automation of activitiesnow being done by CARI monitors or coders.With regard to data quality monitoring, it may bepossible one day to screen a large portion of the audiofiles automatically for evidence of falsification. Forexample, software may be able to distinguish between

audio files with and without voices and to identify thenumber of differing voices within a single recording.This technology could be employed for a populationcensus or large survey that requires many interviews tobe screened very quickly for falsification. Audioprocessing software may be able to determinerespondent qualities such as whether a voice is male orfemale, or to match spoken interviewer words with thepredefined question text, for evaluation of how well theinterviewer followed protocol.CARI can also be used as a data collection tool. Anumber of surveys tape record respondent responsesthat are subsequently coded, and CARI offers aconvenient, unobtrusive alternative for collecting theserecordings. Matching audio responses to a dictionary ofexpected words might allow automated coding of openendeditems or of an “other-specify” option of multiplechoiceitems.Farther in the future, recordings may be transcribedautomatically to text with can be parsed and analyzed.Current commercial software often requires “training”the package to recognize the user’s voice, which limitsusefulness in the field. However, research is underwayon speech-to-text conversion tools in uncontrolled or“noisy” surroundings (Ming, et al, 2006), which maybroaden its applicability to include home environments.AcknowledgementsThe authors would like to acknowledge the work ofAlbert Bethke, Phil Cooley and R. Suresh in theinvention of CARI, the contributions of Frank Mierzwain cost modeling and of Pauline Robinson in filecompression studies. Finally we would like torecognize the contributions of the U.S. Census Bureauto the field of audio-recorded interviewing.ReferencesBiemer, P.P., Hergert, D., Morton, J. and Willis, W.(2000), “The Feasibility of Monitoring FieldInterview Performance Using Computer AudioRecorded Interviewing (CARI)”, Proceedings of theAmerican Statistical Association’s Section on SurveyResearch Methods, pp. 1068-1073Dwyer; J.J, Godin, D.K., Colon, R.S., Sr., Rothschild, S.Pawlowski, J.J., and Vaughan, J.C. (1998), “Voicefile management in portable digital audio recorder”,United States Patent 6671567Hartman, P., Wrenn-Yorker, C., Sattaluri, S. andThissen, M.R. (2006), “Research and Developmentin Audio-Recorded Interviewing”, Presented atFederal Computer Assisted Survey InformationCollection (FedCASIC) ConferenceHerget, D., Biemer, P.P., Morton, J. and Sand, K.(2005), “Computer Audio Recorded Interviewing(CARI): Additional Feasibility Efforts ofMonitoring Field Interview Performance”,Presented at Federal Conference on StatisticalMethods.Kowal, S., O'Connell, D.C. and Sabin, E.J. (1975)“Development of Temporal Patterning and VocalHesitation in Spontaneous Narratives”, Journal ofPsycholinguistic Research, Vol. 4, p. 195-207.The LAME Project, LAME Compression Software,http://lame.sourceforge.net/index.phpMing, J., Hazen, T.J. and Glass, J.R. (2006), "SpeakerVerification Over Handheld Devices with RealisticNoisy Speech Data," Proceedings of theInternational Conference on Acoustics, Speech, andSignal Processing, pp I-637 to I-640.Nuance Communications, Inc. (2005), “AboutDictaphone”,http://www.dictaphone.com/aboutus/history.aspO’Connell, D.C. and Kowal, S. (1983), “Pausology”. InComputers in Language Research, Sedelow, W. A.Jr. and Sedelow, S.Y. (eds), Berlin-New York:Walter de Gruyter & Co., pp. 221-301.Statistics Netherlands, Statistical InformaticsDepartment, P.O. Box 4000, 2270 JM Voorburg,The Netherlands.Stockdale, A. (2002), “Tools for digital audio recordingin qualitative research”, Social Research Update,pp 1-4Suresh, R. (2005). “Web-Based Computer AudioRecorded Interview (Web-CARI).” Presented at theInternational Field Directors and TechnologyConference 2005, Atlanta, GAThissen, M. R., and Rodriguez, G. (2004), “RecordingInterview Sound Bites Through BlaiseInstruments”, Proceedings of the InternationalBlaise Users’ Conference, pp. 411-423.Thissen, M.R., and Sattaluri, S. (2006a) “ComputerAudio-Recorded Interviewing (CARI)”, Presentedat The International Field Directors andTechnologies Conference, MontrealThissen, M.R, and Sattaluri, S. (2006b), “Research andDevelopment in Audio-Recorded Interviewing, PartII”, Presented at The International Field Directorsand Technologies Conference, Montreal, CanadaUniversity of California, Berkeley, Software SupportServices, “Computer-Assisted Survey ExecutionSystem (CASES),” CSM Program, 358 BarrowsHall #3820, Berkeley, CA 94720.Wrenn-Yorker, C. and Thissen, M.R.(2005), “ComputerAudio Recorded Interviewing (CARI)Technology”, Presented at the Federal Computer-Assisted Survey Information Collection(FedCASIC) Conference.