Document Type

Working Paper

Publication Date



This paper assesses whether audio computer-assisted self-interviewing (audio-CASI), a technique designed in the United States to collect data on sensitive behaviors, is a feasible method of survey data collection in a developing-country setting and whether it produces more valid reporting of sexual activity and related behaviors than traditional survey methods. The analysis is based on interviews with nearly 4,400 unmarried adolescents aged 15-21 in Nyeri, a rural district of Kenya that was selected because previous research had indicated a wide discrepancy in the reporting of premarital sexual behavior between boys and girls. The study was based on a quasi-experimental design in which respondents were randomly allocated to one of three interviewing modes-faced-to-face interviews, paper and pencil self-administered interviews, and audio-CASI-with the presupposition that increased privacy would elicit more reliable responses. The interview context was found to have a substantial effect on responses to sensitive questions about sexual and other risky behaviors among young people. Results indicate substantial and significant differences in reported rates of premarital sex across interview modes. For boys, who we suspect exaggerate sexual experience in face-to-face interviews, the effect is in the expected direction, with a 23 percent drop in reported sexual activity in the audio-CASI mode. For girls, who we speculate underreport sexual activity when interviewed face-to-face, there is also a large difference by mode, but the effect is not in the hypothesized direction, with respondents reporting twice as much sexual activity in the interviewer mode as in the audio-CASI mode. While the audio-CASI technology performed well, with minimal mechanical problems, and while respondents were able to complete the survey with limited training, some members of the community reacted adversely to the survey, especially to the use of the computer. We consider whether this may have had an effect on the response patterns of adolescents.