Respondent-driven sampling bias induced by community structure and response rates in social networks

摘要

Sampling hidden populations is particularly challenging using standardsampling methods mainly because of the lack of a sampling frame.Respondent-driven sampling (RDS) is an alternative methodology that exploitsthe social contacts between peers to reach and weight individuals in thesehard-to-reach populations. It is a snowball sampling procedure where the weightof the respondents is adjusted for the likelihood of being sampled due todifferences in the number of contacts. In RDS, the structure of the socialcontacts thus defines the sampling process and affects its coverage, forinstance by constraining the sampling within a sub-region of the network. Inthis paper we study the bias induced by network structures such as socialtriangles, community structure, and heterogeneities in the number of contacts,in the recruitment trees and in the RDS estimator. We simulate differentscenarios of network structures and response-rates to study the potentialbiases one may expect in real settings. We find that the prevalence of theestimated variable is associated with the size of the network community towhich the individual belongs. Furthermore, we observe that low-degree nodes maybe under-sampled in certain situations if the sample and the network are ofsimilar size. Finally, we also show that low response-rates lead to reasonablyaccurate average estimates of the prevalence but generate relatively largebiases.