Directed acyclic graph (DAG) showing the theoretical model informing analysis.

The green variable represents the exposure variable; the blue variable represents the outcome variable; grey variables represent unobserved variables; pink variables represent confounders; pink lines show biasing paths; green lines show causal paths. Additional demographics included proportion of high income earners, proportion of elderly residents, or proportion of non-Hispanic White residents depending on the model. Bivariate associations between each covariate and the CRI were assessed with Pearson correlations when continuous and Gaussian (Proportion of population ≥ 65 yrs of age, Property Crime per 100,000 in 2016, ICEincome, ICEeducation) and with a Spearman’s rank correlation when the variables were not Gaussian. Some of the covariates presented in this model were not found to be significantly associated with the CRI and were thus not included in the adjusted models.