Research comparing state administrative records to Census data systems to uncover the Medicaid enrollment undercount

Estimates of the uninsured vary widely, as do estimates of the numbers of people with different types of coverage. In particular, state and national household surveys often produce much lower estimates of Medicaid enrollment than states' administrative data on Medicaid enrollment. The purpose of this project is for Medicaid data experts at the University of Minnesota to work with the Census Bureau to link state administrative records on Medicaid enrollment to several Census data systems. Findings from this matching will enhance understanding of the factors driving discrepancies between administrative and survey data, and result in improved data systems for coverage policy analysis. Project deliverables include the linked data file (which for confidentiality reasons will remain at the Census Bureau); a report on the relative contributions of different factors to the discrepancies between survey and administrative data on Medicaid enrollment; recommendations for health policy analysts on using survey data on Medicaid enrollment; and presentation of the findings at a major professional meeting. The project will be considered successful if we gain a better understanding of why survey estimates and administrative records differ and the information is used to improve analyses of proposed health policy changes.

Grant Details

Amount Awarded $340,000.00

Awarded on: 11/18/2004

Time frame: 12/1/2004 - 12/31/2007

Grant Number: 52084


University of Minnesota School of Public Health

Mayo Mail Code 197
420 Delaware Street, S.E.
Minneapolis, 55455-0381


Michael Davern
Project Director