type of coverage and of persons uninsured for the previous calendar year since the late 1970s (Levit et al., 1992). During 2001, there were an estimated 41 million civilian, noninstitutionalized uninsured persons in the United States, approximately 16.5 percent of the population under age 65 (Mills, 2002). There is considerable geographic variation in the distribution of uninsured persons regionally and from state to state; see Figures B.1 and B.2 and Tables C.1, C.2, and C.3 in Appendix C.
Using CPS-derived estimates of uninsured rates to compare geographically defined communities involves acknowledging some important methodological limitations related to the sample sizes of the CPS survey and the period of time over which coverage status is measured. For example, while the Committee’s focus would ideally be on communities no larger than a city or rural county, and perhaps an even smaller area, and on comparisons among areas on a yearly basis, the relatively small sample sizes of the CPS survey mean that the smallest type of community for which uninsured rates are estimated is the MSA of more than 500,000 population. To attain sufficient statistical reliability, these MSA-level estimates are based on a three-year average.
In addition, uninsured rates differ according to the length of time (denominator) over which the number of uninsured persons (numerator) is measured because of transitions into and out of coverage. For example, the 2001 California Health Interview Survey includes three sets of estimates for uninsured rates in the state,