242 – Nonresponse Adjustment - 3
Comparison of Traditional Weight Adjustments to Calibrated Weights in the Medical Expenditure Panel Survey for Both Non-Response and Post-Stratification
Robert M. Baskin
Agency for Healthcare Research and Quality
Lap-Ming Wun
Agency for Healthcare Research and Quality
The Medical Expenditure Panel Survey Household Component (MEPS) is an annual two year panel survey of Households sponsored by the Agency for Healthcare Research and Quality and conducted by Westat. The survey collects data on household characteristics, insurance coverage, healthcare use and expenditures. The MEPS survey is a subsample of responding households to the National Health Interview Survey (NHIS). Current non-response and post-stratification steps for MEPS are carried out in traditional sequential steps. This project investigates the possibility of using calibration through SUDAAN'S proc WTADJX to perform non-response adjustment and post-stratification on the MEPS weights at the household level. The results from this calibration approach will be compared to the current approach empirically and by simulation. The simulation will measure bias in the methods by using values on the non-respondents to MEPS created from a combination of NHIS information and modelling. Since the bias can be measured in this way, then the total Mean Squared Error for estimates under the two approaches can be compared.