Best Practice Application: Identifying High and Low Behavior and Performance Using

Abstract

How can we identify best-practice providers? Under the combined influence of GPRA 1, the NPR, the state and community benchmarking efforts, and GASB SEA reporting requirements, most federal, state, and local government agencies, private for-profit and nonprofit organizations delivering government programs under grants and contracts, will become involved in performance measurement. Once governments begin routinely collecting and reporting performance measurement data, policymakers and policy evaluators will be faced with the task of identifying best-practice providers. How can governments go about making comparisons among service providers using performance measurement data? Can best-practice providers actually be identified? Based on previous analysis using a Quantile Regression and SWLS model for estimation and inference, this article introduces a new approach to estimating models of extreme behavior. Quantile Regression and SWLS are investigated to lay a foundation for putting forward the new analysis technique: Segmentation Strategy. Then, some preparatory work for Monte Carlo Simulation, including determining the structure of simulated data sets, is described. Thirdly, the computational results are displayed and analyzed. Finally, some conclusions and future research directions are provided.