295 – SPEED: Big Data, Small Area Estimation, and Methodological Innovations Under Development, Part 1
A Practical Guide to Small Area Estimation, Illustrated Using the Ohio Medicaid Assessment Survey
Rachel Harter
RTI International
Jeniffer Iriondo-Perez
RTI International
Kasey Jones
RTI International
Bo Lu
The Ohio State University
Amang Sukasih
RTI International
Akhil Vaish
RTI International
Much literature has been written about the theory and statistical properties of small area estimators, but very little has been written about the practical aspects of producing small area estimates. This paper summarizes the basic steps for producing small area estimates. The steps involve identifying requirements such as dependent variables of interest and small areas or domains of interest; identifying and compiling auxiliary data and selecting significant predictors; determining an appropriate model, estimation method, and software for running the model; and producing, validating, and reporting the estimates. The steps are illustrated by the production of estimates of the proportion of adults without health insurance coverage, by county, using data from the Ohio Medicaid Assessment Survey.