Computer-assisted Personal Interviewing: An Experimental Evaluation of Data Quality and Cost
Reginald P. Baker, Norman M. Bradburn, and Robert A. Johnson
This article uses experimental data from the National Longitudinal Survey (NLS) to compare the quality and costs of data collected using computer-assisted personal interviewing (CAPI) and traditional paper-and-pencil personal interviewing (PAPI). The experiment was carried out as part of the NLS Youth Cohort (NLS/Y), a continuing longitudinal face-to-face survey of United States residents who were aged 14–21 in 1979. Interviewers were randomly assigned to conduct personal interviews using the same NLS/Y questionnaire as adapted for administration by either PAPI or CAPI. The results suggest that, compared with PAPI, CAPI yields lower item nonresponse rates and greater respondent willingness to disclose sensitive information. Few distributions of responses obtained using CAPI differ substantially from those obtained by administering the same questionnaire using PAPI. The respondent burden, as measured by the length of an interview, is about 20% less with CAPI. CAPI may initially be somewhat more expensive than PAPI, but the cost difference is likely to narrow as organizations and interviewers gain experience in using CAPI, especially if the costs of portable computers continue to decline.
Technical feasibility; portable computers; stratified sampling; random assignment; matching; field management; completion rate; skip pattern; missing data; mode effects; response effects.