Multiple seed structure and disconnected networks in respondent-driven sampling

TL;DR

This paper introduces a new estimator for respondent-driven sampling that accounts for multiple seeds and disconnected networks, improving accuracy over previous methods especially with many seeds.

Contribution

The paper presents a novel estimator based on a random walk with teleportation that handles multiple seeds and disconnected social networks in RDS.

Findings

Estimator outperforms previous methods in simulations.

Effective for large seed proportions and disconnected networks.

Recommended for use in complex RDS studies.

Abstract

Respondent-driven sampling (RDS) is a link-tracing sampling method that is especially suitable for sampling hidden populations. RDS combines an efficient snowball-type sampling scheme with inferential procedures that yield unbiased population estimates under some assumptions about the sampling procedure and population structure. Several seed individuals are typically used to initiate RDS recruitment. However, standard RDS estimation theory assume that all sampled individuals originate from only one seed. We present an estimator, based on a random walk with teleportation, which accounts for the multiple seed structure of RDS. The new estimator can also be used on populations with disconnected social networks. We numerically evaluate our estimator by simulations on artificial and real networks. Our estimator outperforms previous estimators, especially when the proportion of seeds in the sample is large. We recommend our new estimator to be used in RDS studies, in particular when the number of seeds is large or the social network of the population is disconnected.