Respondent-Driven Sampling with LGBT Older Adults

Background and Purpose: Relying on peer-to-peer recruitment chains, respondent-driven sampling (RDS) aims to make population-level inferences from social network data. While RDS has been successful in efficiently recruiting large samples from many hard-to-reach populations, it has not been successful with all populations, and some RDS samples have been shown to be biased. Critics have questioned whether RDS assumptions are reasonable in all of the contexts in which it has been applied. RDS assumes that the population studied comprises an interconnected network and that respondents can provide a reliable count of the people in their personal networks from the target population. RDS also assumes that participants recruit randomly from their networks and reflect reciprocal relationships. This exploratory study examined whether lesbian, gay, bisexual and transgender (LGBT) older adults would participate in peer-recruitment and whether their behaviors were consistent with RDS assumptions. Methods: A gay man, a lesbian, a bisexual man and a transgender woman, at least 50 years old, were selected randomly from lists of eligible prospects known to contacts of the research team. Respondents completed a brief demographic and health survey and were asked to recruit two participants from among their peers. They completed a follow-up survey approximately three weeks later. Descriptive analysis of recruitment patterns was conducted. Reports of the size of personal networks were compared at both time points. Qualitative data from the telephone survey were examined for evidence regarding random recruitment, reciprocity, and overall amenability to peer-recruitment procedures. Results: Sixteen LGBT adults 50 years of age and older were recruited, including 3 lesbian/gay women, 4 bisexual women, 8 gay men, and 1 bisexual man. No respondents recruited transgender peers. Participants recruited primarily within their own race, age-cohort, and socio-economic status but across gender and sexual orientation sub-groups. Participants reported knowing relatively few bisexual and transgender peers, and some reported confusion regarding judging their peers' age and bisexuality. They identified non-random recruitment based on relational proximity and anticipation of openness to participate. Evidence of violations of reciprocity was found with regard to perception of sexual orientation and perception of age. Conclusions & Implications: While all participants stated openness to participate in peer-recruitment research, recruitment chains failed to generate the desired sample size, and almost half of participants failed to recruit anyone. Whereas gay men and lesbians/gay women appear highly interconnected, bisexual and transgender individuals appear infrequently networked with gay men and lesbians 50 years of age and older. Findings raise concerns as to whether LGBT older adults are networked across racial and socioeconomic subgroups and between younger and older age cohorts. Bisexuality and age appeared difficult to reliably perceive for some participants, challenging the assumption of reciprocal recruitment relationships. Future efforts to utilize RDS to study LGBT older adults may perform better restricted only to gay men and lesbians, but conflicting understandings of bisexuality and the unreliable perception of age may continue to challenge inferential reliability of RDS data with this population.