The CPS estimates are based on ERIU tabulations of data collected by the U.S. Census Bureau through the 2004 Current Population Survey (CPS) Annual Social and Economic Supplement (formerly called the Annual Demographic Supplement, or the March Supplement). CPS estimates of the uninsured are intended to represent persons who are uninsured throughout the calendar year. A person is considered insured if they were covered at any time during that calendar year; individuals insured for only one month are counted as insured. CPS estimates of the uninsured should be interpreted with caution. Respondents are asked to report on a time period that begins roughly 15 months prior to the time of the interview. Some respondents are likely to have trouble remembering back this far, while others may be confused by the time frame and respond instead about current coverage status. Studies that have examined and compared CPS estimates with data from other surveys, such as the Survey of Income and Program Participation (SIPP) and the Medical Expenditure Panel Survey (MEPS), suggest that CPS estimates of the uninsured are much closer to “point in time” than true full-year estimates.
The MEPS estimates are based on ERIU tabulations of 2003 Medical Expenditure Panel Survey (MEPS) data collected by the Agency for Healthcare Research and Quality (AHRQ). The MEPS employs an overlapping panel design and collects data on two calendar years for each respondent. The tabulations reported here are based on responses to interview Rounds 3, 4, and 5 of Panel 6 and Rounds 1, 2, and 3 of Panel 7, which cover calendar year 2003 for both panels. Coverage status is obtained at the initial interview. At each subsequent interview, respondents are asked about whether their coverage has changed since the last interview and, if so, how and when. Respondents report whether they were covered by any public or private source of health insurance (Medicare, Medicaid, SCHIP, TRICARE/CHAMPUS, other public coverage, employment-based private, other private health insurance). Respondents are considered uninsured if they respond “no” to all of the coverage options; they are not asked directly if they are uninsured. A person may gain or lose coverage during a year and these data permit estimates of the number lacking coverage at a particular point in time, at some time during the year, and throughout the entire year. We label those who are without health insurance for the entire year as “all-year uninsured,” and those without health insurance for at least one month and up to twelve months as “ever uninsured.” The “point-in-time uninsured” estimates include those without coverage in the first round of the calendar year (Round 3 for those who entered the survey in 2001, and Round 1 for those who entered the survey in 2002).
For further discussion of issues related to counting the uninsured see "Counting and Characterizing the Uninsured," (Adobe PDF) by Pamela Farley Short, and ERIU Research Highlight 1: A Revolving Door: How Individuals Move In and Out of Health Insurance Coverage (Adobe PDF)
Notes on CPS Data Source and Variables (PDF)
Notes on MEPS Data Source and Variables (PDF)