Table 1.
Summary of measures, descriptions, descriptive statistics, and variables applied in this research.
Fig 1.
Frequency of respondents that self-reported negative impacts from extreme event types in the past three years.
Categories of “Slightly / moderately impacted” and “Very / extremely impacted” are combined for display purposes.
Fig 2.
Density of average self-reported extreme weather impacts across event types (A). Highest level of extreme weather impact reported by respondents (B).
Fig 3.
Average self-reported negative impacts from extreme weather events summarized by Census Division and organized per Census Region.
Source of base map: Wikimedia Commons [75].
Fig 4.
Extreme weather impact clusters identified using k-means algorithm summarized by the means of respondent self-reported impacts per event type.
Fig 5.
Binary sociodemographic measures (measure = 1 vs. else = 0) summarized by extreme weather impact cluster.
Each bar represents the percentage of respondents within each extreme weather impact cluster compared to the excluded category. There are statistically significant differences (chi-squared test, p<0.05) across impact clusters for Below Federal poverty level, Black, Hispanic, White, Bachelor’s degree, and Female, indicated by bolded text and borders.
Fig 6.
Binary logistic multilevel regression models predicting likelihood of extreme weather impacts cluster membership with Census Division random effects.
Points represent odds ratios and lines 95% confidence intervals, with statistical significance levels indicated by p-value thresholds of *<0.05, **<0.01, and ***<0.005. Full model results, including standard errors, can be found in the Tables A, B in S1 Text.
Fig 7.
Linear multilevel regression models predicting average self-reported extreme weather impacts across all six extreme weather event types.
Points represent linear coefficient estimates and lines 95% confidence intervals with statistical significance levels indicated by p-value thresholds of *<0.05, **<0.01, and ***<0.005. Full model results, including standard errors, can be found in the Table C in S1 Text.