Download data and study materials from OSF
Robert C. Sinclair
Sean E. Moore
University of Alberta
Sample size: 1641
Field period: 10/20/2005-02/12/2006
On national random sample surveys of sexual behaviour, heterosexual men report 2 to 4 times the number of lifetime sexual partners than do heterosexual women. This, of course, cannot be true because each new sexual partner for a man is also a new sexual partner for a woman; thus, the mean number of sexual partners for men and women must be the same. This finding is problematic for researchers who collect survey data addressing sexual behaviour and has important implications for researchers (e.g., epidemiologists) who track sexually communicable diseases. Some scholars have suggested that this finding reflects intentional erroneous reporting on the part of respondents; that is, men intentionally overreport and women intentionally underreport in order to convey responses that they believe are consistent with social expectations regarding sexual behaviour – this is a bad faith explanation for the finding (see, e.g., Brown & Sinclair, 1999).
Brown and Sinclair (1999) conducted a convenience sample survey study and asked respondents to report their number of lifetime sexual partners (defined as someone with who they had engaged in vaginal or anal intercourse) and then asked respondents to report the strategy that they used to generate their report (i.e., retrieval strategy). They replicated the male-female sex difference and showed that men and women use different strategies to generate their estimates. Men, on average, tend to use rough approximations (e.g., “25 + a few partners”) or rate-based strategies (about 5 a year for 5 years = 25) whereas women, on average, tend to enumerate (e.g., “I thought back over time and counted all of my partners”). Based on literature from cognitive psychology, Brown and Sinclair (1999) argue that it has been shown that rough approximations and rate-based strategies lead to overestimations because they have no lower bound and enumeration leads to underestimation because people forget. When Brown and Sinclair (1999) controlled for retrieval strategy, the sex difference between men and women disappeared; this suggests that respondents are not intentionally biasing their reports but that the strategies used to retrieve the information affect the estimate generated by the respondent. In a random sample telephone survey study of 1200 respondents, Sinclair and Brown (2001) replicated the sex difference and showed that the difference could be eliminated when they controlled for retrieval strategy and sexual attitudes (specifically, perceptions that sex should occur in the context of a committed relationship). Although the studies described above suggest that retrieval strategy is an important contributor to the biased retrieval of information regarding number of lifetime sexual partners, no studies to date have experimentally manipulated retrieval strategy to demonstrate conclusively that the cause of biased reporting is retrieval strategy.
It is predicted that rough approximation activation will lead to the highest reported number of partners, that enumeration activation will lead to the lowest reported number of partners, and that the sex difference between men and women will only emerge in the control condition. As in Sinclair and Brown (2001), controlling for retrieval strategy and sexual attitudes should eliminate the sex difference in the control condition.
The present study is designed to extend the research described above, to experimentally manipulate retrieval strategy through a strategy activation technique, and to attempt to understand methods through which the biased reporting of lifetime sexual partners can be attenuated. In the context of a national random sample telephone survey, 900 male and 900 female respondents who are 18 years old or older will be randomly assigned to one of three retrieval strategy activation conditions. One third will be asked “Off the top of your head, please provide a rough estimate of your number of lifetime sexual partners,” one third will be asked “Please think back over your lifetime starting with your first sexual partner and count all of your sexual partners up to and including your most recent partner,” and one third will serve as a control group who will be asked “Please report your number of lifetime sexual partners.” As a manipulation check and for the purpose of an analysis to control for retrieval strategy in the control group, respondents will be asked to select the actual strategy that they used to retrieve the information from a menu provided to them.
Thus, the design of the study is a 2 (sex: male, female) x 3 (strategy activation: rough approximation, enumeration, control) between groups factorial.
respondents will be asked to report their attitudes toward sex in the context of a committed relationship on two 7-point Likert scale items (see the attached survey instrument), their age, and their sexual orientation.
Please see the "proposal" document in the "data and study materials" file.