Considerations for the Use of Small Area Analysis in Survey Analysis for Health Policy
Considerations for the Use of Small Area Analysis in Survey Analysis for Health Policy: Example from the 2015 Ohio Medicaid Assessment Survey
Marcus E. Berzofsky
RTI International
Bo Lu
Division of Biostatistics, Ohio State University
Daniel Weston
Ohio Colleges of Medicine Government Resoruce Center
G. Lance Couzens
RTI International
Timothy Sahr
Ohio Colleges of Medicine Government Resource Center
State-based surveys are often required to produce reliable estimates at multiple levels of geography (e.g., state, sub-state region, and county). Often the optimal design varies depending on the level of geography used in analysis. Small area estimation (SAE) can assist in improving precision in sub-state areas that do not have enough interviews to produce reliable estimates directly. The 2015 Ohio Medicaid Assessment Survey (OMAS) allows for the measurement of health status, access to the healthcare system, and health determinant characteristics for Ohio's Medicaid, Medicaid eligible, and non-Medicaid populations. Understanding the variation in these outcomes at different levels of geography is important to both practitioners and legislators. This paper describes the approach taken to design the 2015 OMAS to optimize the number of direct estimates that can be produced while ensuring that a minimal number of interviews are available at all levels of geography to support SAE estimates. Our paper demonstrates how this process leveraged data from the previous iteration of OMAS to minimize the design effects in as many areas as possible.