Spatial autocorrelation between human responses and Winter storm Grayson

This study explores the relationship between Winter storm Grayson and human responses in the US by considering a multitude of periods, regions, socio-demographic variables and spatial autocorrelation for the number of tweets. This study suggests that the proportion of tweets is highly concentrated in the north-east region, which is the most damaged region in the winter storm week. Second, there is significant spatial autocorrelation for the number of tweets related to the winter storm. The lag coefficient values of the spatial lag model and the spatial error model are 0.362 and 0.437 at a 0.01 significance level, respectively. Third, younger people and people who have less than a high school degree show a positive coefficient value, whereas males, people with a high school degree, and those who have high income show a negative coefficient value for the number of tweets. This study shows that spatial regression models would be better models to understand human responses to natural disasters than the ordinary least squares model.