Network Model-Assisted Inference from Respondent-Driven Sampling Data (1108.0298v1)

Abstract: Respondent-Driven Sampling is a method to sample hard-to-reach human populations by link-tracing over their social networks. Beginning with a convenience sample, each person sampled is given a small number of uniquely identified coupons to distribute to other members of the target population, making them eligible for enroLLMent in the study. This can be an effective means to collect large diverse samples from many populations. Inference from such data requires specialized techniques for two reasons. Unlike in standard sampling designs, the sampling process is both partially beyond the control of the researcher, and partially implicitly defined. Therefore, it is not generally possible to directly compute the sampling weights necessary for traditional design-based inference. Any likelihood-based inference requires the modeling of the complex sampling process often beginning with a convenience sample. We introduce a model-assisted approach, resulting in a design-based estimator leveraging a working model for the structure of the population over which sampling is conducted. We demonstrate that the new estimator has improved performance compared to existing estimators and is able to adjust for the bias induced by the selection of the initial sample. We present sensitivity analyses for unknown population sizes and the misspecification of the working network model. We develop a bootstrap procedure to compute measures of uncertainty. We apply the method to the estimation of HIV prevalence in a population of injecting drug users (IDU) in the Ukraine, and show how it can be extended to include application-specific information.