Counting Uninsurance and Means-Tested Coverage in the American Community Survey

Historically, the Current Population Survey (CPS), the survey of record for distributing public program funds, has provided health services researchers and policy-makers with annual state-level coverage estimates. Beginning in 2008, the U.S. Census Bureau began asking about health insurance as part of its American Community Survey (ACS).

The two surveys have methodological differences. ACS asks about health insurance at the time of the survey, compared to CPS which asks respondents to recall coverage in the previous year. ACS sample size is 30 times that of CPS and it does a better job of assigning values to missing data items. ACS, however, does not distinguish among means-tested coverage or allow state-specific names to identify Child Health Insurance Programs.

Both surveys produced similar estimates of health insurance coverage, although there were wide variations between the ACS and CPS surveys across states. Overall rates of uninsurance for nonelderly were 17 percent in ACS and 18.3 percent in CPS. The largest differences were found in rates of direct purchase (ACS 1.5 times more) and state-level means-tested coverage for children (ACS less).

While the ACS is well-poised to become a useful tool to health services researchers and policy analysts, further study is needed to quantify its bias.