U.S. Department of Commerce

Small Area Health Insurance Estimates (SAHIE)

Skip top of page navigation
Census.govPeople and Households SAHIE Main SAHIE MethodologyMethodology Review › Comment Topics

SAHIE Review: Comment Topics

Differences from the Annual Social and Economic Supplement (ASEC) of the Current Population Survey (CPS)

The SAHIE program's state estimated uninsured rate for children under age 18 was noticeably different from the Annual Social and Economic Supplement (ASEC) of the Current Population Survey (CPS) estimate for several states.

The SAHIE program's state estimate is different from the CPS ASEC state estimate because we have combined the CPS ASEC estimate with information from administrative records and demographic estimates. By forming a model of the relationship of these various pieces of information we can improve the precision of the final estimate. The SAHIE program's estimated uninsured rate for children under age 18 was noticeably different from the CPS ASEC estimate in several states due to the method we used to make the number of insured and uninsured additively consistent at the various levels of geography. This method seemed to impart an extra component of error, especially to the uninsured rate for children under age 18. We have revised the method to minimize this effect.

Differences from Census 2000 Population

Why doesn't the SAHIE program's number uninsured plus insured equal the Census 2000 population? It is substantially different in some cases.

The number insured plus the number uninsured for states and the nation equals the Annual Social and Economic Supplement (ASEC) of the Current Population Survey (CPS) population. The definitions of the CPS ASEC population and the Census 2000 population are substantially different. The CPS ASEC, for example, excludes persons living in institutions and military personnel living in barracks.

Additionally there are differences in the reference years. The SAHIE program's estimates of insured plus uninsured equal the 2001 CPS ASEC population estimates for states and the nation. The 2001 CPS ASEC asks questions about health insurance coverage in 2000.

The SAHIE program's estimates are based on a three-year average of the CPS ASEC centered on 2000. We model the averaged estimates for 1999, 2000, and 2001 obtained from the 2000, 2001, and 2002 CPS ASECs. Health insurance coverage is not currently a question on the decennial census or the American Community Survey.

Differences from an Estimate Produced for or by Individual States

The SAHIE program's estimate is substantially different from an estimate produced for or by individual states.

We have noticed that the SAHIE program's estimates are often different than health insurance coverage statistics produced in several states. One reason for those differences is that the SAHIE program's estimates are specifically for the definition of health insurance coverage used by the Annual Social and Economic Supplement (ASEC) of the Current Population Survey (CPS). The definitions used by state departments of health are often different. A particular difference may be the treatment of college students living in dormitories. The CPS ASEC counts college students living in dormitories as residing in their parental homes. Also, compared with other national surveys, the CPS ASEC estimate of uninsured more closely approximates uninsured at a specific point in time during the year than uninsured for the entire year.

Employment and Number of Firms

Why haven't you included employment or number of firms as predictors in the models?

In earlier models we included employment and number of firms from the County Business Patterns (CBP) data. These variables were not significant in the presence of other predictors. There are some limitations with the CBP data for our purposes, such as the exclusion of government employees. We are continuing to research the inclusion of employment and firm data using other data sources such as the Quarterly Census of Employment and Wages (QCEW).

Urban and Rural Counties

Why haven't you included a variable indicating the urban or rural nature of a county as a predictor in the models?

In earlier models we attempted to include the urban influence codes for 2003 from the Economic Research Service of the United States Department of Agriculture. These codes were not significant in the presence of other predictors. We are continuing to research this topic.

College and Group Quarters Population

Areas with a large college population appear to have an uninsured estimate that is too high. Why is the group quarters population excluded?

We model health insurance coverage as measured by the Annual Social and Economic Supplement (ASEC) of the Current Population Survey (CPS), where college students living in dormitories are counted as residing in their parental homes. Group quarters populations are excluded because the CPS ASEC universe consists of the civilian noninstitutional population of the United States and members of the armed forces living in the United States off post or with their families on post. It excludes all other members of the armed forces and persons residing in institutions, such as nursing homes, correctional facilities, juvenile detention facilities, and long-term mental health care facilities. For more information on the CPS sample design see Technical Paper 63RV Design and Methodology [PDF 3.3M].

American Indian Population

Areas with a large American Indian population appear to have an uninsured estimate that is too high.

We model health insurance coverage as measured by the Annual Social and Economic Supplement (ASEC) of the Current Population Survey (CPS). In the CPS ASEC people with no health insurance coverage other than access to Indian Health Service are considered uninsured. For more information see health insurance definitions.


Source: U.S. Census Bureau | Small Area Health Insurance Estimates |  Last Revised: August 29, 2013