Fig. 5: Associations between smoke impacts and social vulnerability variables.

a Coefficients from unique bivariate linear regressions of natural log-transformed smoke exposure on social vulnerability variable (SVV) percentile ranks at the tract level. Colors represent the social vulnerability variables’ respective themes. b Coefficients from unique bivariate linear regressions of natural log-transformed damages per capita on social vulnerability variable percentile ranks at the tract level. White represents equal damage and exposure coefficients. Green represents a lower damage coefficient than exposure coefficient. Brown represents a higher damage coefficient than exposure coefficient. a, b Data characterize 2017. Coefficients are the percentage changes in impacts per 0.01 increase in the social vulnerability variable percentile ranks, which range from 0 to 1. Error bars show 95% confidence intervals derived using bootstrapping. Social vulnerability variables are those comprising the Social Vulnerability Index (SVI)^38,39. See Supplementary Methods 4 for details on the regression modeling and statistical analysis. See Supplementary Table 12 for coefficients, 95% confidence intervals, statistical significance indicators, and comparisons guiding the color scale in (b). Plotting uses the ggplot2 R package¹¹⁸.