Cori Mar Publishes Article on Variance Estimators for Respondent-Driven Sampling
Posted: 8/29/2017 (CSDE Research)
Cori Mar, CSDE Methods and Data Director, recently co-authored an article titled “Evaluating Variance Estimators for Respondent-Driven Sampling” that was published in the Journal of Survey Statistics and Methodology. The authors evaluated the performance of RDS variance estimators via a simulation study based on 40 respondent-driven sampling (RDS) surveys of injection drug users in the U.S. This analysis was the first to study the performance of the multiple RDS variance estimators. The study showed that confidence intervals and design effects of RDS variance estimators are often acceptable, although imperfect. There was strong evidence that simple random sample estimators and related confidence intervals significantly underestimate variance, and should therefore not be the choice for analyzing RDS data.