Using Network Sampling in Crime Victimization Surveys

Network sampling links population households in specified ways, for reporting purposes, in order to increase the probabilities of locating households with particular characteristics. We conducted a reverse record check field experiment to test whether telephone surveys using network sampling is feasible to collect crime victimization data. Three types of crime -- burglary, robbery, and assault -- were tested along with two types of networks -- relatives and co-workers/close friends. This paper examines the extent to which victims report their victimization experiences in a general crime and victimization interview and the extent to which a randomly selected relative or close friend will report the same victimization incident in an identical interview. A number of multiplicity counting rules are compared in terms of reporting errors and a mean square error analysis. 6 tables, 37 references. (Author abstract)