Methods: Using a community-based participatory research approach, THIS used regional focus groups to develop constructs for a quantitative survey. Virginia residents 18 years and older self-identified if they considered themselves transgender according to study eligibility criteria. A review of previous transgender health assessments was conducted to initiate study design. Presentations were made to a wide variety of community and professional audiences throughout the state, and a careful plan was created to distribute paper surveys and information about accessing the questionnaire online. Distribution points included transgender community groups, house balls and Pride events, offices of supportive providers and counselors, and public health programs. TTF members were key to establishing effective distribution. The Field Coordinator, a well-respected member of the target community, was essential to the study's success.
Results: 350 respondents were included in the substudy - 61% responded online, 39% completed paper forms. Differences in sociodemographics, geography, and public health indicators by sampling modality (internet vs. paper) were explored, using SAS statistical software, with significance set at 0.05. Internet responders were significantly older (39.1 vs. 34.0 years), had higher incomes, were 4 times as likely to finish high school, and 2-5 times more likely to have health insurance. The internet sample was primarily Caucasian (85%); the paper sample primarily racial/ethnic minorities (71%). In 2 of 4 geographic regions, internet respondents significantly outnumbered those who completed paper forms. In the Central region, internet response was greater, and in Eastern, paper responses far outnumbered those online. Response mode was associated with community characteristics – among internet responders, 48% lived in suburban, 34% in urban, and 19% in rural communities. Most paper responders lived in urban areas (54%); 27% in suburban and 16% in rural areas. Public health indicators (substance use, mental health) did not vary by response mode, but higher percentages of paper responders had experienced sexual violence (31%/22%).
Conclusions and Implications: In this survey of a hidden population, dual response modes resulted in a more diverse sample than either mode would have generated alone. Subsample differences indicated that provider training and outreach efforts to transgender communities will be more effective if tailored for geographic and subpopulation demographic characteristics.