A Critical Threshold for Design Effects in Respondent-Driven Sampling

• Activity Number: 239
• Type: Topic Contributed
• Date/Time: Monday, August 10, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Survey Research Methods Section
• Author(s): Karl Rohe*
• Companies: University of Wisconsin - Madison
• Keywords: Respondent driven sampling ; social network analysis ; hidden population ; marginalized population ; Volz-Heckathorn

Abstract

Respondent driven sampling (RDS) is a social network link-tracing sampling technique for studying marginalized or hard to reach populations. Previous literature models the referral process as a random walk on the social network. However, participants in RDS are incentivized to refer between three and five future participants. To account for the branching structure of the process, this talk studies the RDS process as a Markov chain *indexed by a tree*. Several results follow. An exact formula is given for the variance of the popular Volz-Heckathorn estimator (VH). There is a critical threshold that relates the referral tree to the social network and past this threshold, VH is not square root N consistent.