Design Effect Variation for Estimates Derived from the MEPS (1996-1998)

The sample design of the Medical Expenditure Panel Survey (MEPS) is an overlapping panel design characterized by a multistage, complex area probability design that includes disproportionate sampling of specified policy relevant population groups. Standard methods of variance estimation which assume simple random sampling generally result in an under-estimation of variance, when used with data from a complex survey design (Cohen S., 1982). The extent of this departure from simple random sampling assumptions and its impact on the variances of survey estimates may be measured by the design effect. A summary of design effect variation for 1997 MEPS estimates has been previously published (Yu, W., 2001). Based on data from the 1996, 1997, and 1998 MEPS, this paper evaluates and contrasts: (1) the design effects achieved over time for national estimates of health care utilization and expenditures; (2) design effect variation by alternative population subgroups and by different geographic regions of the nation; and (3) the level of design effect variation in related survey estimates.