Doctoral Dissertation Research: The Medicaid Undercount and the Role of Program Design

University Of Pennsylvania, Philadelphia PA

Investigators

Abstract

This doctoral dissertation research project will test whether state-level Medicaid program design and utilization explains variation in the Medicaid undercount. Medicaid is the largest source of public health insurance in the United States. Census surveys significantly underreport Medicaid use compared to administrative records. The underreporting of Medicaid use in census surveys is problematic, particularly because governmental bodies and researchers rely on this data to make administrative decisions and to determine the budgetary impacts of legislation. Inaccurate Medicaid measurement can lead to inefficient program implementation and incorrect budgetary projections. In previous studies, the undercount has been attributed largely to question wording and survey design. However, there is a lack of research into whether Medicaid program design generates response error in survey self-reports. This research will use linked census survey and administrate data to generate corrective predicted probabilities of Medicaid use for census surveys. These predicted probabilities have the potential to help policy makers better understand and account for the needs of the Medicaid population. Although focused on Medicaid in particular, this study may provide insights into the factors that affect rates of misreporting in other contexts. As a Doctoral Dissertation Research Improvement award, the project will enable a promising student to establish a strong, independent research career. This project will link census data to Medicaid administrative records to test the theory that the misreporting of Medicaid usage in census surveys is a function of variation in Medicaid program design. The investigators will run a series of logistic and fixed effects logistic regressions on this linked data to isolate which program features are related to the misreporting of Medicaid usage. The project also will use a difference-in-differences design to better isolate whether policy shifts play a causal role in the Medicaid undercount. Coefficients will be generated that can be used to predict the probability of Medicaid enrollment. These corrective measures will be made available to the Census Bureau, government agencies, and the general public.