Journal of Official Statistics, Vol.13, No.4, 1997. pp. 341–365
Sources of Error in a Survey on Sexual Behavior
Roger Tourangeau, Kenneth Rasinski, Jared B. Jobe, Tom W. Smith and William F. Pratt
Abstract:This article describes the results of a methodological study on survey reports about sexual behaviors and other sensitive topics. More than 1,000 women took part in the experiment; the study also included a comparison sample of 100 men. Most of the respondents were selected from an area probability sample; in addition, some of the women were selected from the rosters of two abortion clinics. Questionnaires based on the one used in the National Survey of Family Growth were administered to the sample; the questionnaire included items on abortion, sexual behavior, and illicit drug use. The experiment investigated five experimental factors; the one with the most consistent effect on the results was the method of administering the questions. Self administration significantly increased the number of sexual partners, sexually transmitted diseases, and the level of condom use reported by the women compared to administration by an interviewer. Self administration also increased the number of sexual partners reported by men. Computer assistance occasionally interacted with the site of the interview (in the respondent's home or outside the home) to effect reporting. The other two experimental variables – the version of the questionnaire and the data collection staff – had few discernible effects.
Keywords:Computer administration; CASI; interviewer effects; mode effects; sensitive questions.
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group