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

摘要

Purpose:To study the respondent-driven sampling bias that is induced by the structure and response rates in network communities.Summary:Respondent-driven sampling is an alternative methodology to standard sampling methods that can exploit social contacts between peers to reach individuals in hard-to-reach populations. However, the bias in the respondent- driven sampling estimator is expected to be induced by the community structure and response rates in social networks. In this article, different structures and response rates are simulated to reproduce real settings to study the bias induced by network communities. Analysis is performed on how the community structure affects respondent-driven sampling bias. The respondent- driven sampling estimator is also investigated as a function of willingness to participate (i.e., response rates). Tests are also conducted on different scenarios where the variable under study is correlated with the number of contacts per individual and the size of the network community. The consequences of the biased selection of subjects and the bias that is induced by network structure in samples of real social networks are also studied in detail. (38 refs.)Results:In order to estimate the prevalence of diseases, traits or behaviors in particular social groups or even in the entire society, researchers typically rely on samples of the target populations. In order to reduce the standard error of the estimate,.