An Application of Small Area Estimation Techniques to Derive State Estimates of Health Insurance Coverage from the 1987 NMES 1

Abstract

There is a growing need to obtain State-level estimates of the uninsured population to assist in the formulation of appropriate health care policies directed to this population subgroup. However, reliable estimates of these parameters at the State level are rarely directly obtainable from national sample surveys. This evaluation focuses on alternative small area estimation strategies that are applicable to data from the 1987 National Medical Expenditure Survey (NMES-2). These techniques use model based estimators that combine national survey data on variables of interest, with auxiliary data and population characteristics of subnational areas to yield estimates for these subnational areas. A comparison of the accuracy of these small area estimators is presented. Prior to making State-level estimates of the uninsured, the alternative strategies are used to estimate Medicaid recipients, and comparisons are then made with actual program statistics.