Anomalies Under Jackknife Variance Estimation Incorporating Rao-Shao Adjustment in the Medical Expenditure Panel Survey Insurance Component

• Activity Number: 122
• Type: Topic Contributed
• Date/Time: Monday, July 30, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Section on Survey Research Methods
• Author(s): Matthew Thompson*+ and Robert Baskin
• Companies: U.S. Census Bureau and Agency for Healthcare Research and Quality
• Keywords: jackknife ; variance estimation ; establishment survey ; missing data

Abstract

The Medical Expenditure Panel Survey Insurance Component is an annual survey of establishments and governments sponsored by the Agency for Healthcare Research and Quality and conducted by the U.S. Census Bureau. The survey collects data on employee characteristics and availability and options of employer provided health insurance. Primary information collected includes availability of employer sponsored insurance, types of insurance offered, uptake by employees, employee premiums, and employer contributions. Previous work on the impact of imputation on variance estimation has indicated that for a naïve jackknife variance estimate, the underestimation of the variance of means and proportions is on the order of ten percent. However, in that work, the Rao-Shao type adjustment produced surprising and sometimes unrealistic results for some estimated proportions. It was observed that the adjusted replicate value could be out of range, i.e. negative, for a proportion. The current work investigates this issue using simulations.