The relationship between community public health, behavioral health service accessibility, and mass incarceration

Table 3 Beta regression model for predicting jail population per capita

	Estimate	standard error	Z-score	p-value
(Intercept)	-5.263	0.016	-332.413	< .001
Income inequality	0.009	0.015	0.554	0.580
High school graduation rate	-0.054	0.013	-4.046	< .001**
County size Medium vs Small	-0.156	0.059	-2.659	0.008**
County size Large vs Small	-0.335	0.123	-2.712	0.007**
Poor physically unhealthy days	0.144	0.017	8.616	< .001**
Primary care physician rate	0.007	0.018	0.405	0.686
Health care costs	0.089	0.016	5.701	< .001**
Percent of drug treatment paid by Medicaid	-0.049	0.013	-3.664	< .001**
Psychiatrists per capita	-0.039	0.018	-2.159	0.031*
Community MH centers per capita	0.009	0.014	0.604	0.546
Violent crime rate	-0.011	0.019	-0.59	0.555
Police per capita	0.150	0.017	8.736	< .001**

** represents significance at the 0.01 level in a two-tailed test; * represents significance at the 0.05 level in a two-tailed test
Note 1: Considering that beta regression models use a log transformation when modeling the response variable, estimates/coefficients need to be exponentiated before reporting/interpreting the strength to which they contribute to explaining the response variation
Note 2: The z-scores indicate the strength of the relationship between each predictor and the outcome variable while holding everything else constant, and the p-value column is used to evaluate whether each variable plays a statistically significant role in predicting the outcome. The direction of the relationship is designated by whether the z-score is negative or positive

ISSN: 1472-6963