Random Parameter Model of Postdisaster Household Relocation

Abstract

Household recovery and relocation after a disaster is an exigent and complex decision. This complexity is mainly due to the variety of factors that can influence such a decision. Therefore, it is the goal of this study to develop random parameter models incorporating the unobserved heterogeneity that can result in predicting these recovery decisions and their societal economic impact. To develop these models, samples comprising 192 blocks (5,338 households) in Staten Island, New York, and 58 blocks (2,393 households) in Moore City, Oklahoma, were selected. Block-level data were extracted from the 2010 United States Census. The analyses indicate that a random parameter approach offers a better statistical fit than fixed parameter models. It was also found that the proportion of family households and proportion of directly impacted households are random parameters, while other variables such as race, average family size, and gender show fixed effects. This study is unique and innovative because it highlights the significance of random parameter models in identifying unobserved effects within the context of postdisaster recovery modeling.