Analyzing the Effectiveness of a Non-Verbal Response Card: Evidence from Ethiopia

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Survey researchers have long known that in interviewer-administered surveys, respondents often intentionally misreport their behavior or attitudes in order to create a more favorable image of themselves in the eyes of the interviewer, or to avoid creating an awkward interaction. Reporting bias is therefore a particularly critical issue as survey research focuses increased attention on sensitive topics, such as sexual risk taking, illicit substance use, domestic violence, and attitudes regarding sexual behavior, racial preferences, and discrimination. In the proposed study we will examine the effectiveness of a new response method for soliciting responses to sensitive questions. This method was recently piloted by the investigators in the Gilgel Gibe Social and Sexual Relationship History Survey, in southwestern Ethiopia. The survey included a total of 1,268 randomly sampled female and male youth ages 13-24, and collected information on health seeking behavior and sexual experience, knowledge, and attitudes among other topics. The "non-verbal response card method" involves the use a simple two-sided card that allows the respondent to answer questions non-verbally. The card increases the level of privacy and confidentiality in an interviewer- administered survey, and it makes minimal cognitive demands on the respondent. The non-verbal response cards were tested in the Gilgel Gibe survey in a randomized control trial design in which half of the sample used the response cards and the other half of the sample provided verbal responses to interviewer-administered questions. The purpose of the proposed study is to examine the effectiveness of the non-verbal response card method for reducing reporting bias, and to estimate the nature and extent of reporting bias for sensitive questions about sexual behavior that are asked in many standard demographic surveys. Specifically, the proposed study will: (1) assess the capacity of respondents to correctly use the non-verbal response card method; (2) measure the direction and magnitude of reporting bias for different types of questions and for different population subgroups; (3) assess the potential impact of reporting bias on coefficients in multivariate regression models that use responses to questions on sexual attitudes and experiences to model sexual behavior; and (4) use multivariate models to identify population subgroups for whom reporting bias is the most severe. PUBLIC HEALTH RELEVANCE: The proposed study provides estimates of the direction and magnitude of bias in self-reports of sexual behavior in a study of youth in southwestern Ethiopia. It also presents a new methodology for recording responses to sensitive questions in face-to-face interviewer-administered surveys that increases privacy and confidentiality. To the extent that this methodology is proven useful for improving the accuracy of reports of sensitive behaviors in the Ethiopian context, it has potential for widespread use in other developing and low-literacy settings.