A Systematic Review of Characteristics of Nonprobability Community Venue Samples of Sexual Minority Individuals and Associated Methods for Assessing Selection Bias.

Abstract

The majority of research on sexual minority individuals relies on nonprobability community venue samples. These samples are prone to selection bias; however, empirical syntheses of evidence of these biases are not available. We conducted, therefore, a systematic review of published sexual minority health research to summarize methods used to identify characteristics and health outcomes found to differ in nonprobability samples. We searched five health and social science databases to identify observational studies that included a nonprobability sexual minority community sample and applied an empirical method to infer selection bias. We extracted data regarding sociodemographic characteristics, behaviors, and health outcomes and examined whether the nonprobability sample was found to differ disproportionately (over- or underrepresenting the characteristic) based on appropriate statistical tests (p < 0.05 was considered statistically significant). A total of 21 studies were included. Nineteen studies used a cross-sample comparison, one used time/location sampling weights, and one used frequency of venue attendance adjustment to assess differences. The evidence was mostly consistent, that is, associations were in the same direction in >50% of studies examined, for nine variables. Nonprobability community venue samples tended to overrepresent sexual minority individuals with the following characteristics: higher income, current employment, lesbian/gay-identified, greater number of sex partners, past-year sexually transmitted infection diagnosis, suicidal ideation, alcohol use, and substance use; nonprobability community venue samples tended to underrepresent married/partnered sexual minority individuals. This review provides a nuanced empirical picture of aggregate differences in sample characteristics presumed to threaten the validity of nonprobability sexual minority community venue studies, and highlights feasible methods that can be applied to future studies to add specificity to researchers' description of selection biases.