Critical issues and further questions about respondent-driven sampling: comment on Ramirez-Valles, et al. (2005)

Abstract

The application of respondent-driven sampling (RDS) is a promising new approach to understanding hidden populations, including those at high risk for HIV infection. The method has significant advantages over other sampling methods, including the possibility that representativeness samples can be accrued. However, the requirements for a respondent-driven sample to achieve representative and to demonstrate its superiority over other methods are quite strict. This report focuses on whether the accompanying paper by Ramirez-Valles et al. fulfills these requirements in terms of the theorems underlying RDS regarding sample recruitment rates, referral patterns, and network sizes. Further, it investigates the assertion that the accrued sample has proven that RDS is superior to more traditional time-location or venue-based sampling methods. Unconvinced that either is the case, the author suggests a method to test if RDS is indeed the gold-standard recruitment strategy for obtaining inclusive and representative samples of hidden populations.