In all analyses, for multiple comparisons, we used Bonferroni-corrected p-values to determine statistical significance. That is, when comparing the four density measures separately across gender and race/ethnicity using independent t-tests, a p-value=0.012 (α=0.05/4 comparisons) was used. The same corrected p-value=0.012 was used when comparing the four dichotomous measures separately across gender and race/ethnicity using non-parametric Chi square tests. For AUC and regression analyses comparing dichotomous and density measures, a p-value=0.006 (α=0.05/8 comparisons) was used. Similarly, for all AUC and regression models comparing dichotomous and density measures separately across gender and race/ethnicity, for each of the outcome variables, a p-value=0.003 (α=0.05/16 comparisons) was used.
