Estimates of Non-Heterosexual Prevalence: The Roles of Anonymity and Privacy in Survey Methodology

Estimates of Non-Heterosexual Prevalence: The Roles of Anonymity and Privacy in Survey Methodology. Ronald E. Robertson et al. Archives of Sexual Behavior, https://link.springer.com/article/10.1007/s10508-017-1044-z

Abstract: When do people feel comfortable enough to provide honest answers to sensitive questions? Focusing specifically on sexual orientation prevalence—a measure that is sensitive to the pressures of heteronormativity—the present study was conducted to examine the variability in U.S. estimates of non-heterosexual identity prevalence and to determine how comfortable people are with answering questions about their sexual orientation when asked through commonly used survey modes. We found that estimates of non-heterosexual prevalence in the U.S. increased as the privacy and anonymity of the survey increased. Utilizing an online questionnaire, we rank-ordered 16 survey modes by asking people to rate their level of comfort with each mode in the context of being asked questions about their sexual orientation. A demographically diverse sample of 652 individuals in the U.S. rated each mode on a scale from −5 (very uncomfortable) to +5 (very comfortable). Modes included anonymous (name not required) and non-anonymous (name required) versions of questions, as well as self-administered and interviewer-administered versions. Subjects reported significantly higher mean comfort levels with anonymous modes than with non-anonymous modes and significantly higher mean comfort levels with self-administered modes than with interviewer-administered modes. Subjects reported the highest mean comfort level with anonymous online surveys and the lowest with non-anonymous personal interviews that included a video recording. Compared with the estimate produced by an online survey with a nationally representative sample, surveys utilizing more intrusive methodologies may have underestimated non-heterosexual prevalence in the U.S. by between 50 and 414%. Implications for public policy are discussed.