摘要：We test sensitivity of model specification on the choice of contractual physician arrangements by hospitals through use of spatially defined covariates, rural/urban subset models, and geographically weighted regression (GWR) models. Results show that hospitals are significantly influenced by neighboring hospitals' physician arrangement choices whether they are market-oriented or integrated arrangements. Further, some spatial covariates were shown to be nonstationary across space, including variables that were insignificant in the global parameter models. Results suggest health care scholars consider varying parameter models in order to avoid global statements concerning health care relationships and evaluating the performance of health policy.
