Abstract
Synthetic methods using age, race, sex and Hispanic origin (ARSH) arrays specific to regions have been used in combination with survey data to produce inexpensive estimates of health insurance coverage for different geographic regions. Because health insurance status is typically derived from survey data, a synthetic approach is unreliable when a region has few persons in sample with certain ARSH characteristics. For instance, a region with few persons of Hispanic origin and sparse survey coverage will produce less reliable estimates for that population subgroup.
This article focuses on a hybrid model of synthetic estimation and hierarchical modeling. Hierarchical modeling can minimize the unexplained differences between regions making it possible to increase the reliability of the health insurance status estimates. Information on health insurance status and age, race, sex and Hispanic origin comes from the Survey of Income and Program Participation. State-level variables that influence the State's average health insurance rate were gathered from multiple sources.