Balanced Repeated Replication Variance Estimates from the Medical Expenditure Panel Survey 1996--2003 Household Component

• Activity Number: 389
• Type: Contributed
• Date/Time: Wednesday, August 1, 2007 : 8:30 AM to 10:20 AM
• Sponsor: Section on Survey Research Methods
• Author(s): Dan Liao*+
• Companies: University of Maryland
• Keywords: single PSU stratum ; balanced repeated replication variance estimation ; pooled stratification

Abstract

In this report, we investigate Balanced Repeated Replication (BRR) variance estimates in the Medical Expenditures Panel Survey (MEPS), Household Component (HC), 1996--2003, for several crucial medical expenditure variables by comparing the results from the data of individual years to the ones from design-based alternatives and the corresponding Taylor linearization variance estimates. For longitudinal analyses of MEPS-HC data, a pooled stratification was created for multiple years to increase the sample size and statistical power. However, in this pooled stratification, some single primary single unit (PSU) strata exist in some individual years which make the design improper for BBR. Three different ways of solution are proposed to adjust this pooled stratification design and the BRR variance estimates of the adjusted designs are tested and evaluated at the end.