Lacking a """"""""gold standard"""""""" for the measurement of sexual risk behavior, AIDS prevention researchers typically rely on retrospective, self-reported outcome measures for evaluating the efficacy of HIV risk-reduction interventions. However, a number of studies have pointed to various respondent-based characteristics (for example, problems of comprehension, memory failure, and self-presentation bias) and instrument-related factors (for example, problematic question wording or response formats, overly-intrusive modes of collection, etc.) that may contribute to high rates of unexplained error in survey-based measures of sexual behavior. Several new methods have been developed to improve the accuracy of self-report data, including the use of memory cues in formatting sexual behavior survey items and the adoption of computerized (CASI) assessment modes. However, there has been little research that systematically examines whether self-report data elicited by these new techniques is any more valid (accurate) than data collected by more traditional survey methods. The proposed research will implement a strategy for provisionally validating sexual behavior self-report data collected from samples of three populations at risk for HIV -- men who have sex with men, heterosexual men, and heterosexual women. It will consist of a randomized experiment to examine the impact of both survey question format and mode of assessment on the accuracy of self-report data. Following formative sociolinguistic studies and pilot tests, aimed at developing appropriately worded and formatted instruments for the main experiment, we will recruit 504 participants, drawn equally from each target population. For the next 3 months, these participants will report their sexual behavior prospectively, using structured diary forms, which they will return by mail on a daily basis. Upon completion of the diary period, they will be assessed on their behavior recalled for those same 3 months. The respondents will be assigned to six assessment conditions that vary both on question format (i.e., the presence or absence of partner-specific memory cues) and on data collection mode (i.e., paper surveys -- SAQs, text-only CASI, and audio-assisted CASI). The main analysis will evaluate the response agreement between diary reports and the survey-based measures of sexual behavior. We hypothesize that results will show higher agreement levels between the diary-based and survey-based frequencies of sexual activities and condom use for respondents to partner-specific question formats and CASI assessment instruments in retrospective self-reports.
Pinkerton, Steven D; Galletly, Carol L; McAuliffe, Timothy L et al. (2010) Aggregate versus individual-level sexual behavior assessment: how much detail is needed to accurately estimate HIV/STI risk? Eval Rev 34:19-34 |
McAuliffe, Timothy L; DiFranceisco, Wayne; Reed, Barbara R (2010) Low numeracy predicts reduced accuracy of retrospective reports of frequency of sexual behavior. AIDS Behav 14:1320-9 |
McAuliffe, Timothy L; DiFranceisco, Wayne; Reed, Barbara R (2007) Effects of question format and collection mode on the accuracy of retrospective surveys of health risk behavior: a comparison with daily sexual activity diaries. Health Psychol 26:60-7 |