Logit Coefficient and Random Forest Usage in Health Care Access Trend

Abstract

In the study, random forests were used and aided in predicting the governor’s decision to oppose Medicaid expansion. The importance of each variable was assessed in terms of importance by determining the average of differences in accuracy after and before permuting each variable in the forest. In turn, the difference served as an estimated point for determining the importance of a variable. On the one hand, larger values that were positive and shifted away from zero depicted more important variables. On the other hand, smaller values that were negative and closer to zero were indicative of less important variables. Findings revealed stark support of this study’s findings that political-related factors dominate governor decisions regarding Medicaid expansion when compared to need-related factors.