696 – Analysis of Social Networks
The Relative Statistical and Operational Plausibility of Multiple Frame Sampling for Rare Population Subgroups
William D. Kalsbeek
The University of North Carolina at Chapel Hill
Bruce D. Spencer
Northwestern University
Carol C. House
National Academy of Sciences
For studies where the objective is to estimate the prevalence rate of members of a sampled population who fall in a rare subgroup, this paper examines the relative statistical precision of prevalence estimates from a multiple-frame samples compared to single-frame household samples with the same data collection budget. We first examine relative cost-efficiency for simple un-clustered samples and then consider the effect of cluster sampling. Findings are illustrated for the case where the subgroup consists of victims of rape and sexual assault (RSA) in a civilian non-institutionalized population of persons 12 years and older. Two sample designs are considered: (i) dual-frame sampling from a conventional household frame plus a frame constructed from police reports of RSA, versus (ii) single frame sampling from the household frame. We conclude for this illustration that a dual-frame design will be more cost-effective to the extent that RSA prevalence among police reports exceeds the RSA prevalence in the population as a whole. However, gains in the dual-frame design are diminished in direct relationship to the size of intra-class correlation when cluster sampling is considered.