Using geographical data and rolling statistics for diagnostics of respondent-driven sampling.

ABSTRACT

Respondent-driven sampling (RDS) is commonly used to sample from key populations without a sampling frame since traditional methods are unable to efficiently survey them. Surveying these populations is often desirable to inform service delivery, assess effectiveness of programs, and determine prevalence of diseases. However, there are concerns about how RDS works in practice due to its many assumptions. To assess some of these assumptions, we develop diagnostics using geographical data and demonstrate their utility by identifying lack of convergence and characterizing RDS reach in surveys conducted among female sex workers and men who have sex with men in Kampala, Uganda.