A RESAMPLING ANALYSIS OF FEDERAL FAMILY ASSISTANCE PROGRAM QUALITY CONTROL DATA An Application of the Bootstrap

Federal quality control (QC) programs for the measurement of payment allocation errors in federally funded Family Assistance Programs have been at the center of a hot controversy. In 1986, Congress directed the Department of Health and Human Services to contract with the National Academy of Sciences to study these QC programs. The Academy’s Panel on Quality Control of Family Assistance Programs has recently completed its work on the study. Existing federal QC methodology employs a two-phase sampling scheme, giving rise to a regression estimator of the allotment error rate. A key issue in the QC debate is the appropriateness of the regression estimator in view of marked departure of the QC data from the standard model-based conditions for linear regression-linearity, constancy of variance, and normality of error. In this study, we employ bootstrap resampling methods to investigate the effects of these departures on the efficiency of the regresscon estimator of the error rate. The analysis is based upon actual QC data from the Oregon Food Stamp Program. The origin of quality control procedures in federal Family Assistance<A Programs (FAPs) may be traced back to the Aid to Families with Dependent Children (AFDC) program in the early 1970s. Today, three FAPs- food stamps, Medicaid, and AFDC- are all subject to quality control procedures that are descendent from the early AFDC program. Regardless of the initial intent of these quality control programs, and possibly contrary t