Deriving State-Level Estimates from Three National Surveys: A Statistical Assessment and State Tabulations. B. Combining Data from Multiple Rounds of the Same Survey


A relatively inexpensive method for improving the accuracy of state estimates is to combine data from multiple years of the same survey. The precision of estimates isn't quite doubled when two years of data are combined, since the sample will typically make use of the same PSUs. This additional clustering will somewhat limit the gains in precision that would otherwise be expected.

With SIPP it is also important to assure that different panels are used in different years, rather than asking the same respondent the same question at two time periods. For questions that appear in the CPS core questionnaire, it is possible to carefully combine samples across months as well as years, but one must take into account the complex pattern CPS uses for re-interviewing respondents. The NHIS uses the same questionnaire throughout the year, so an entire year's interviews can be used.

Another issue in combining data across years is that the question of interest must be asked in each year. Many interesting questions appear on NHIS supplements or SIPP topical modules that are not asked every year. Also, combining multiple years limits how quickly the estimates can detect changes over time. Thus, this procedure may not be ideal if one is interested in measuring the affect of policy changes such as welfare reform.

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®