Explaining the diffusion of Medicaid home care waiver programs using VPRS decision rules.

While public preferences and legal decisions require extended Medicaid Home and Community-Based Services (HCBS), uneven program development is a major concern of policy makers and consumers. This paper presents the first Variable Precision Rough Sets (VPRS) analysis of national data to examine inter-state variation in the Medicaid HCBS waiver program. The exposition provides a detailed discussion of the methodological options and processes, tests the generated rules using a leave-one-out cross-validation, and compares VPRS classification accuracy with regression analyses of the same dataset. The results demonstrate that VPRS offers a robust method with two distinctive features for health care research. First, for policy makers and their audiences, VPRS results are presented as "if...then..." decision rules with likelihoods stated as percentages. Because this output form can be easily understood by non-specialists, the potential impact of research is enhanced. Second, for analysts generating evidence for health policy, VPRS provides a rigorous data mining tool and acknowledges inherent analytical uncertainty in the field.