Introduction

For studies of hidden and hard-to-reach human populations, often the most effective way of obtaining a sample is to use link-tracing methods. Conventional designs, such as a random stratified household sample, tend to produce a very small yield of very rare subpopulations and an even smaller yield of subpopulations with stigmatized or socially marginalized behaviors, such as illegal drug use or commercial sex-related activities. The usual frames used in surveys, such as landline and cell-phone numbers and household addresses, lead to underrepresentation of subgroups, such as persons who are homeless, traveling, or in institutionalized settings. Undocumented workers may be hard to survey because of geographic mobility and uncertain legal status. Social network connections can in some cases provide access not easily obtained by other means.

For some studies, it is important to understand the network structure of a population as well as the individual characteristics of the people in it. This is especially true in the case of epidemics of contagious diseases. The epidemic of the human immunodeficiency virus (HIV) has compelled societies throughout the world to try to understand sexual and drug-using behaviors and reach at-risk, hidden populations in an effort to understand and alleviate the spread of the disease. More broadly, individual behaviors and social connections are related, and understanding of social network structure is necessary for the understanding of each. Link-tracing sampling designs provide a natural means for studying and understanding socially structured human populations.