PERSONAL NETWORK SAMPLING, OUTDEGREE ANALYSIS AND MULTILEVEL ANALYSIS: INTRODUCING THE NETWORK CONCEPT IN STUDIES OF HIDDEN POPULATIONS

Abstract

Populations, such as heroin and cocaine users, the homeless and the like (hidden populations), are among the most difficult populations to which to apply classic random sampling procedures. A frequently used data collection method for these hidden populations is the snowball procedure. The prerequisite for using this procedure is the existence of a network among the members of the population. Most studies of hidden populations treat the existence of networks as implicit, and subsequently the analysis remains at a qualitative level. In this article a practical approach to these populations is described that can be used simultaneously not only as a tool for locating a reasonable number of members of a hidden population but also as a tool for analysing aspects of network structure. By taking into account the personal network in the sampling design (personal network sampling) as well as in the analyses (outdegree analysis and multilevel analysis), meaningful quantitative information about aspects of network structure will be obtained. This approach is illustrated by data from a cocaine users project in the Dutch city of Rotterdam.