Dept. of Public Health and Preventive Medicine
Oregon Health & Science University
Respondent-driven sampling (RDS) is useful for accessing hard-to-reach populations but requires tracking respondents in a way that may not be feasible in anonymous surveys or in some communities. Purpose This study explores modifications to RDS that preserve respondent anonymity and provides recommendations for implementing RDS. Methods Results were simulated for a hypothetical study that tracked recruitment in groups of respondents rather than in person-to-person recruitment networks. In regular RDS, recruitment network data are used to generate specific population estimates. In anonymous RDS, network data are lacking; instead, group data are used to generate ranges of possible scenarios. These possible scenarios were compared to the results that would have been obtained under regular RDS. Findings By simulating possible recruitment scenarios, it is possible to generate point and interval estimates for the distribution of characteristics in a sample that has reached equilibrium. Mean, median, and probability-weighted estimates produce intervals that vary in precision. Conclusions Modifying RDS to preserve respondent anonymity requires sacrificing some precision in analyzing the sample and generating population estimates. Anonymous methods most closely approximate regular RDS methods when the sampling scheme has been successful; however, the degree to which the sampling scheme has been successful is unknown in anonymous RDS. Further study should provide results that better reflect the overall population. Sampling methods may also be developed to provide more data about recruitment patterns without identifying individual respondents. In spite of these weaknesses, anonymous RDS is a systematic method that, in contrast to the convenience samples that are more commonly used in hard-to-reach populations, provides tools to evaluate a studyâs external validity.
School of Medicine
Ramsey, Katrina L., "Recruiting "hidden" populations methodological considerations for adapting respondent-driven sampling to preserve participant anonymity in research in native communities" (2007). Scholar Archive. 187.