How Do Surveys Differ in Reporting the Quality of Reported Medicaid Enrollment Data: NHIS, CPS, and State Surveys

• Activity Number: 239
• Type: Topic Contributed
• Date/Time: Tuesday, July 31, 2007 : 8:30 AM to 10:20 AM
• Sponsor: Section on Government Statistics
• Author(s): Kathleen Call*+
• Companies: The University of Minnesota
• Keywords: measurement error ; surveys ; administrative data ; Medicaid undercount ; uninsurance

Abstract

Survey measurement error can contribute to the tendency for surveys to show lower counts of Medicaid enrollment than administrative data (i.e., Medicaid Undercount). This happens two ways: first, respondents can misreport that they are uninsured when they really have Medicaid, biasing survey estimates of the uninsured and those on Medicaid. Second, respondents can answer they have coverage other than Medicaid, biasing the Medicaid count and not the uninsured count. The first form of error is more detrimental to policy research than the second. We compare the prevalence of both error scenarios using three types of surveys in which we know the Medicaid enrollment status of respondents: four state surveys, the Current Population Survey, and the National Health Interview Survey. Preliminary analysis shows that rates of the first type of error are much higher in the CPS than other surveys.