What methods do researchers use to measure recent same-sex sexual behavior in married men?
Executive summary
Researchers measure recent same-sex sexual behavior among married men using a mix of direct survey questions about recent partners, household roster approaches that infer couple sex composition, specialized online modules that time partnerships, and crosswalk methods that combine survey reports with population counts—each method trades off different biases related to stigma, recall period and question framing [1] [2] [3] [4].
1. Direct survey questions about behavior, attraction and identity
National probability surveys commonly ask respondents separate questions about sexual behavior (e.g., sex with a man in the past 12 months, five years, or ever), sexual attraction, and self-identified orientation, and researchers use those behavioral items to flag recent same-sex sex among men who report being married [1] [2]; the National Survey of Family Growth and other large studies provide tables of same-sex sexual contacts by marital status and specified recall windows [2].
2. Distinguishing behavior from identity and attraction—conceptual measurement choices
Studies repeatedly emphasize that sexual behavior, attraction, and identity are distinct domains and that choosing one over the other changes who is counted as having “same-sex behavior,” so researchers must decide whether the outcome is recent acts (behavior), tendencies (attraction), or labels (identity)—a critical conceptual point discussed in methodological reviews and the National Academies report on measuring sex, gender identity and sexual orientation [5] [6].
3. Household roster and administrative inference for married couples
Census and federal household surveys identify same-sex married or cohabiting couples by combining recorded sex of household members with relationship status on the roster; this approach captures couple-level same-sex households but cannot detect hidden same-sex acts by a married man whose household appearance is different, and historically data editing sometimes recoded apparent same-sex married responses as sex-report errors [3] [7] [8].
4. Timing modules and online tools to capture recent concurrency and partner overlap
To pin down whether same-sex acts occurred during a marriage or recently, researchers deploy partnership-timing modules and enhanced web-based questionnaires that collect first/last-sex dates for each partner and use logic to detect concurrency; these modules have shown stronger agreement with direct concurrency questions than some prior measures and can better situate same-sex encounters in relation to the marriage timeframe [4] [9].
5. Web-based, multilingual questionnaires and psychometric scales for behavior nuances
International and online comparative studies capture multiple facets of sexual behavior—satisfaction, sensation seeking, compulsivity, coping with attraction and engagement in specific acts—using validated scales (e.g., Sexual Sensation Seeking Scale, Sexual Compulsivity Scale) and tailored questionnaires that allow researchers to detect recent same-sex behavior and its correlates among men who may identify as heterosexual but report male partners [10].
6. Population estimation: combining surveys, surveillance and census denominators
When estimating how many men have recent same-sex sex, researchers often combine survey-derived percentages of men reporting male partners with census counts to arrive at population estimates; public health work on men who have sex with men (MSM) uses these survey-based proportions together with HIV surveillance and other data to generate population-level figures, acknowledging that stigma and survey nonresponse can depress self-reports [1].
7. Sources of bias, data editing and interpretive limits
All methods face limitations: stigma and confidentiality concerns suppress disclosure in surveys, recall windows (past year vs past five years) change prevalence estimates, surveying wording and mode (computer-assisted vs interviewer) affect honesty, household roster editing historically misclassified same-sex married reports, and concurrency measures can over- or under-estimate behavior tied specifically to marriage rather than pre- or post-marital periods—points methodological papers and reviews explicitly flag [11] [3] [1] [9].
8. Best-practice syntheses and trade-offs for research design
Recent recommendations converge on using multiple, clearly worded items (behavior with defined recall windows plus attraction and identity), confidential self-administered modes, partnership timing modules to place acts relative to marriage, and combining survey results with administrative denominators when estimating population sizes; however, published assessments caution that no single approach fully overcomes stigma-driven underreporting or the mismatch between couple-level household data and individual sexual behavior [6] [1] [4].