Skip to main content
. 2023 Oct 3;2(10):pgad295. doi: 10.1093/pnasnexus/pgad295

Table 4.

Spatial regression model results: alternative specifications.

Dependent variable:
log(outage duration)
Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8
log(peak outages) 0.495a 0.391a 0.407a 0.358a 0.349a 0.350a 0.348a
(0.039) (0.036) (0.037) (0.035) (0.035) (0.035) (0.035)
log(population) 0.499a 0.393a 0.396a 0.355a 0.389a 0.387a 0.376a
(0.057) (0.050) (0.050) (0.044) (0.042) (0.042) (0.042)
log(county size) 0.186 0.185a 0.154a 0.147 0.177a 0.172a 0.159a
(0.099) (0.084) (0.075) (0.077) (0.077) (0.077) (0.077)
Socioeconomic vul. 0.684a 0.540a 0.504a 0.386a
(0.267) (0.231) (0.220) (0.157)
Household characteristics vul. 0.294 0.193 0.087 0.037
(0.152) (0.129) (0.120) (0.107)
Racial and ethnic minority status vul. 0.064 0.126 0.473a 0.010
(0.211) (0.180) (0.178) (0.161)
Housing type and transportation vul. 0.138 0.028 0.031 0.109
(0.161) (0.139) (0.130) (0.104)
ρ 0.400a 0.406a 0.488a 0.385a 0.385a 0.386a 0.383a
(0.032) (0.034) (0.039) (0.036) (0.036) (0.036) (0.036)
Average direct socioeconomic impact 0.787 0.596 0.549 0.384
Minimum-length outage (min) 5 30 60 60 60 60 60
Minimum sustained wind swath (knots) 0 0 34 0 0 0 0
Observations 1,104 943 691 862 862 862 862
Log likelihood 1,633.431 1,173.041 718.029 981.319 980.965 981.164 979.025

Note: Other controls: IOU customer ratio, flood risk index, wind speed, state, and storm dummy variables (see online supplementary material note 1 and Table S1 for complete results).

aP<0.05 .