Deriving State-Level Estimates from Three National Surveys: A Statistical Assessment and State Tabulations. I. INTRODUCTION


This report assesses the statistical issues involved in the production of state-level estimates related to health and welfare issues from three national surveys: the Current Population Survey (CPS), the Survey of Income and Program Participation (SIPP), and the National Health Interview Survey (NHIS). With the devolution of many welfare programs from the Federal Government to the states, there is a strong interest in being able to track the health and welfare of the population in each state. This would allow for examination of the effect of various state welfare initiatives that are to be implemented in the next few years.

Section II provides an overview of the relevant statistical issues involved in making state-level estimates from these surveys, including state stratification, nonsampling errors, and precision of the estimates (see Glossary for additional explanation of these and other highlighted statistical terms used in this report). Section III assesses the abilities of the CPS and SIPP surveys to produce four specific estimates, all expressed as percent of the total population: individuals in households with income below the poverty line, individuals receiving Aid to Families with Dependent Children (AFDC) individuals covered by employer-provided health insurance, and individuals with a disability. Each estimate is examined for the total population, and also for subpopulations of blacks, Hispanics, children, and the elderly.

The most current publicly available databases for each of the three surveys were examined. For the CPS, the most current data are from the March 1996 survey. For the NHIS the most recent data are for 1994. In 1995, the NHIS sample was completely redesigned, so examining the 1994 data would yield little information on the ability of future years to provide state-level estimates. Thus, while no NHIS data are examined, general discussions of the ability of the NHIS to provide the desired estimates are included.

For SIPP the 1993 panel data are available. Like the NHIS, the SIPP was also redesigned in 1996. The 1996 SIPP is about one-and-a-half times as large as the 1993 SIPP. However, the sample design is broadly the same, so the 1993 SIPP provides some useful indicators of the ability of the 1996 SIPP to produce state-level estimates. In addition to the 1996 SIPP, the Bureau of the Census is continuing to follow the 1992 and 1993 SIPP panels in the newly introduced Survey of Program Dynamics (SPD). The SPD will include all of the low-income households from these two SIPP panels, along with a subsample of the panels' other respondents. As shown in Section III, the sample size of each of these surveys severely limits the capability of the surveys to produce state-level estimates, particularly for subpopulations.

All three surveys are multi-stage national probability surveys of households, with questions asked about all or some members of the household. The CPS is a monthly survey of approximately 60,000 households, with a special income-related supplement asked each March. The NHIS is an annual survey of approximately 100,000 individuals in 40,000 households (sample sizes can fluctuate from year to year) with interviews spread out across the entire year. Both of these surveys are redesigned every 10 years to incorporate the latest Decennial Census information. Beginning in 1995, both surveys contain separate strata for each state, separately for metropolitan and non-metropolitan areas. The SIPP is a panel survey with households interviewed three times a year for multiple years. The 1993 SIPP panel had approximately 21,000 households while the 1996 panel has 35,000 households. Unlike the other two surveys, SIPP strata cross state boundaries.

Finally, Section IV examines alternative approaches to overcome the sample size limitations identified in Section III. These approaches include supplementary state samples, combining data from multiple years of the same survey, combining data from the three surveys, and using indirect model-dependent estimators.

View full report


"deriving.pdf" (pdf, 2.83Mb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®