The Impact of Computer-assisted Personal Interviewing (capi) on Interviewer Performance: the Cps Experience

1. Introduction For the past several years, the Census Bureau and the Bureau of Labor Statistics (BLS) have been working on a major redesign of the Current Population Survey (CPS) with the aim of producing both improvements in data quality and savings in time and money (see Butz and Plewes, 1989). Key elements of the redesign include extensive changes to the basic CPS questionnaire, the collection of a subset of interviews from centralized telephone facilities (using CATI) and the use of personal laptop computers for household interviewing (using CAPI). In order to test the impact of such design changes on key estimates of employment and unemployment, the Census Bureau and BLS developed the CATI-CAPI Overlap (CCO) study. The CCO began in October 1991 and will conclude in December 1993, with the design changes being implemented on the full CPS in January 1994. The CCO study is based on an independently selected national sample of about 15,000 households per month designed to test the new instrument (known as the Monthly Labor Survey or MLS) and the use of computer-assisted interviewing (both CATI and CAPI). This paper focusses on the CAPI portion of the CCO Study. One issue surrounding the CPS design changes is the potential impact of the use of CAPI on the Bureau's workforce of field interviewers. Bateson and Hunter (1991: 24) note that the role of the interviewer is by far the most changed under CAPI. To date, however, much of the research effort on CAPI has been focussed on improvements in hardware and software, while little attention has been given to interviewer performance. In an effort to examine the impact of CAPI on the CPS interviewer, a variety of data collection efforts were launched at the Census Bureau in conjunction with the CCO Study. Key among these was a series of mail surveys to CCO interviewers, designed to elicit their attitudes and expectations toward, their experiences with, their reactions to, and their performance on, CAPI. Two rounds of the questionnaire were administered, one prior to training and the other after two full MLS assignments had been completed. Two key questions can be raised with regard to the implementation of CAPI and its effect on survey interviewers: 1. How do interviewers react to the introduction of CAPI? 2. What effect does CAPI have on interviewer performance? We addressed the first of the questions in an earlier paper (Couper and …