Interpreting the estimates from four national surveys of the number of people without health insurance.

Four national surveys conducted between 1977 and 1980 seem to yield four different estimates of the number of people under 65 years old who lack health insurance. In this paper four explanations for the different estimates are assessed. The sample framework designs and methods for ex post weighting of the surveys' respondents are very similar. Nonresponses to the health insurance questions on the surveys were handled differently, and this remains a possible explanation of some of the differences in the estimates. But the most likely explanation of the differences is the following. Three of the four surveys ask people about their health insurance at the time of the interview. The fourth, the Current Population Survey, asks people about their health insurance during the previous year. But the Current Population Survey estimate of the uninsured population is remarkably close to the point-in-time estimates from the other three surveys. The same holds true for the surveys' estimates of the Medicaid population and people covered by private health insurance. Hence, if we assume that the Current Population Survey respondents are answering the health insurance questions with respect to when they are interviewed, the four surveys' estimates of the uninsured population at a point in time are very similar.