Abstract

This article summarizes the main elements, advantages, and disadvantages of Respondent-driven Sampling (RDS). Some criticisms regarding the feasibility of the inherent assumptions, their point estimators, and the obtained variances are pointed out. This article also comments on the problems observed in the quality of reports. Surveys using RDS should be methodologically sound as they are being applied to define priorities in health programs and develop national and international policies for financing service delivery, among other uses. However, there is considerable potential for bias related to implementation and analytical errors. There is limited empirical evidence on how representative the results obtained by RDS are, and the quest to improve the methodology is still in progress. Nevertheless, to have confidence in RDS results, we must verify that the social structure of the networks conforms to the assumptions required by the theory, that the sampling assumptions are reasonably fulfilled, and that the quality of the report is optimal, particularly for methodological and analytical items.