Evaluation of Small-Area Estimation Method Used in AskCHIS Neighborhood Edition

• Activity Number: 78
• Type: Contributed
• Date/Time: Sunday, August 9, 2015 : 4:00 PM to 5:50 PM
• Sponsor: Survey Research Methods Section
• Author(s): Yueyan Wang and Pan Wang* and Jean D. Opsomer and Ninez A. Ponce and Hongjian Yu
• Companies: UCLA and UCLA and Colorado State University and UCLA and UCLA
• Keywords: small area estimation ; semi-parametric model ; prediction mean-squared error ; Jackknife replication ; California Health Interview Survey

Abstract

AskCHIS Neighborhood Edition disseminates small area health estimates based on California Health Interview Survey for zip codes, cities, legislative districts and counties in California. Census tract-level auxiliary data and population data are used to provide information for granular areas such as zip codes where survey samples are scarce or even non-existent. A semi-parametric model with penalized splines over auxiliary data and random effects at sampling strata-level is used. Due to the common form of minimization criteria, fitting of penalized spline can be converted to solving for coefficients of mixed models. A Jackknife method is used for estimating variance and prediction mean-squared error of the modeled estimates. We conduct an empirical comparison to examine the validity of the proposed approach.