Comparison of Daily and Retrospective Reports of Vaginal Sex in Heterosexual Men and Women
Gillmore, Mary Rogers, Leigh, Barbara C., Hoppe, Marilyn J., Morrison, Diane M., The Journal of Sex Research
Questions about sexual behaviors are critical to the study of reproduction, contraceptive use, and sexually transmitted infections (STIs). The answers to such questions rely largely on self-reports in which people recall their behavior over some specified period of time. This complex task requires that people understand the question, retrieve occurrences of the behavior from memory, determine whether the behaviors happened during the specified time frame, infer an answer based on the instances of the behavior, and accurately report the answer (Schwarz, 2007; Tourangeau, Rips, & Rasinski, 2000).
Inaccuracies are introduced when study participants forget exactly how often or how many times a behavior occurred. Recall over longer periods of time, or for behaviors that occur at very high or very low rates, may be especially subject to forgetfulness or over-reliance on heuristics. Retrospective reports are also more influenced by recent events than events that occurred in the more distant past, and people may use a different heuristic for recalling recent as opposed to longer-term behaviors, both of which can produce bias in recall (Bogart et al., 2007; Burton & Blair, 1991; Jaccard & Wan, 1995). Retrospective reports are also subject to "telescoping"--that is, including events that occurred before or after the specified time for recalling behaviors (Graham, Catania, Brand, Duong, & Canchola, 2003). Moreover, people may intentionally under- or over-report a behavior, and this may be more likely to occur when reporting sensitive behaviors such as sexual activities. Misreporting may differ by gender or other characteristics, as Wiederman (1997) pointed out when discussing research demonstrating that men report more sexual partners than women. Such differences may be due to errors of recall or estimation, with men more likely to "round up" their estimated number of lifetime partners. Other research suggests that women, older persons, and those with less education tend to under-report sexual experiences (Wiederman, 1993).
In the absence of a valid criterion for measuring sexual behavior, daily reports have been assumed to produce more accurate reports of sexual behaviors than retrospective reports because many of the problems of recall are reduced (Graham et al., 2003; Leigh, Gillmore, & Morrison, 1998; Minnis & Padian, 2001), and events are recorded in their natural settings (Bolger, Davis, & Rafaeli, 2003). A recent review (Schroder, Carey, & Vanable, 2003b) concluded that existing studies show no clear trend toward under- or over-reporting of sexual behavior on questionnaires, relative to diaries, and that variations in item content and the samples studied (adolescents, sex workers, college students, men who have sex with men) might affect general tendencies to over-report or under-report. Some studies asked for counts of behaviors in retrospective reports, while others used categorical measures of frequency (e.g., every day, twice per week, etc.; see Schroder, Carey, & Vanable, 2003a, for a comparison of these methods of assessing sexual behaviors). Recent studies have highlighted the role of question format, data collection mode (computer-assisted self-interviewing, self-administered questionnaires), and time of retrospection (McAuliffe, DiFranceisco, & Reed, 2007; Morrison-Beedy, Carey, & Tu, 2006; Schroder, Johnson, & Wiebe, 2007).
In this study, participants reported on their sexual and other behaviors daily for eight weeks and then completed a retrospective questionnaire about their activities over the reporting period. The sample included both male and female adolescents and adults. We included both counts of behaviors and categorical frequency measures in the retrospective questionnaire, and assessed both agreement and bias in our comparisons of daily versus retrospective measures. In addition, we examined whether gender, method of daily reporting (self-administered vs. …