Proper Variance Computation for Estimates from MEPS Event Files
*Sadeq R Chowdhury, AHRQ 
Machlin R Steve, AHRQ 

Keywords: Domain analysis, MEPS, Variance estimation

In addition to person-level public use files, eight event-level files are also released from Medical Expenditure Panel Survey (MEPS) every year. For each person in the MEPS sample, these event files may include no record, a single record or multiple records depending on the number of medical events the person had during the year. Therefore, an event file only includes records related to a subset of persons in the full-year person file. Since the number of persons in an event file is not known before conducting the survey, any analysis of estimates from event files should be treated as a domain analysis which requires the entire sample to take all variability into account in estimating the variance of domain estimates. That is, the analysis should ideally include all persons with and without events in the file. However in practice, since it is convenient to deal with the subset of cases with events only, users generally compute event level estimates without merging persons with no event from the full person-level file. The impact of not doing a domain analysis is usually negligible if the subset (the domain) is large. This poster presents the results from a study to assess the impact of not doing a proper domain analysis on variance estimates.