Design and analysis of qualitative collaborative studies: minimum collaborative program.

Abstract

Collaborative studies involving qualitative data are usually conducted under design constraints to fulfill the requirements for quantitative studies. The data from these qualitative studies are often analyzed in a manner that ignores the fact that collaborative studies involve matching (i.e., each laboratory analyzes a portion of each test sample). This report presents some design considerations and analysis procedures for qualitative collaborative studies that take into account that the design involves matching. Suggestions are offered as to the number of laboratories and test samples to use in the minimum collaborative program, and analysis procedures for outier screening are detailed. Method performance is assessed through such indicators as sensitivity, specificity, false positive, and false negative rates. Methods for estimating the error of the performance indicator rates are explained, and procedures are given for estimating false positive and false negative rates for lot defect rates that may occur in practice.