Recruiting samples that are more representative of illicit drug users is an on-going challenge in substance abuse research. Respondent-driven sampling (RDS), a new form of chain-referral sampling, is designed to eliminate the bias caused by the non-random selection of the initial recruits and reduce other sources of bias (e.g. bias due to volunteerism and masking) that are usually associated with regular chain-referral sampling. This study provides a methodological assessment of the application of RDS among young adult MDMA/ecstasy users in Ohio. The results show that the sample compositions converged to equilibrium within a limited number of recruitment waves, independent of the characteristics of the initial recruits (i.e. seeds). The sample compositions approximated the theoretical equilibrium compositions, and were not significantly different from the estimated population compositions—with the exception that White respondents were over-sampled and Black respondents were under-sampled. The effect of volunteerism and masking on the sampling process was found not to be significant. Though identifying productive seeds and improving the referral rate are significant challenges when implementing RDS, the findings demonstrate that RDS is a flexible and robust sampling method. RDS has the potential to be widely employed in studies of illicit drug-using populations.
Available at: http://works.bepress.com/robert_carlson/159/