Improving Self-Reports About High Risk Sexual Behavior

University Of California, San Francisco, San Francisco CA

Investigators

Linked publications & trials

Abstract

Research findings of sex surveys provide abundant empirical evidence that the reliability and validity of selfreported sexual behavior is a major problem, particularly when assessing behavior among those with multiple partners and sexually risky behaviors. Identifying ways to improve the accuracy of responses to questions that ask respondents to report on their sexual behavior is critical for assessing risk related to HIV/AIDS. Scientific models of the survey question-answer process specify question comprehension as a necessary precursor to obtaining a response that is as error-free as possible. Most successful research efforts to improve respondent comprehension and improve data quality have focused on crafting clearer questions to be used in subsequent surveys. However, innovative methods to administer survey questions that would improve respondent comprehension of questions during a survey is a neglected area of research. We propose to conduct an experiment to test an innovative method of administering a survey interview--conversational interviewing--that has shown remarkable improvements in respondents'understanding of survey questions and in the accuracy of their responses when compared to administering it using the standardized interviewing method. The standardized method of administering survey questions stipulates that if respondents request clarification to the meaning of a question, interviewers are told to read the question again, and if confusion persists, say, "Whatever it means to you." In contrast, the conversational method argues that the basic structure of the survey interview can be maintained, while adding procedures to allow interviewers to assist respondents to understand a question that is posing difficulties. Experiments evaluating the effectiveness of conversational interviewing have been confined to surveys about employment or recent purchases. No research has explored the application of the conversational method to sexual behavior research. We will use a mixed method design (n=200): (1) to conduct an experiment;and (2) to identify the types of comprehension issues that occur during a sexual behavior survey. In our original plan we intended to conduct qualitative analysis on all 200 cases collected for the experiment. In our revised plan to ensure completion of the study in two years we will conduct the qualitative analysis to saturation, which is scientifically justifiable and likely to be significantly fewer cases.