A random parameter ordered probit model to understand the mobilization time during hurricane evacuation

This paper presents a random parameters ordered probit model to capture underlying unobserved characteristics in the timing behavior of the evacuees that elapses in between their evacuation decision and actual evacuation i.e. the mobilization time for an evacuee. The ordered probit model has been developed by using Hurricane Ivan data and the estimation findings suggest that the mobilization time involves a complex interaction of variables related to household location, evacuation characteristics, and socio‐economic characteristics among others. In the model, six variables‐ source and time of evacuation notice received, work constraint, previous hurricane experience, race and income‐ were found to be random and the random parameters (all normally distributed) suggest that their effect varies across the observations. In addition, the model introduces some new factors that impact the mobilization time (for example, the mobilization time for evacuees evacuating to public shelters is significantly lower) which have not been found in the earlier literature to the best of our knowledge. The findings of this study are useful to determine different fractions of people evacuating early or delaying for some time once they actually decide to evacuate, for a given socio‐demographic profile. These fractions can be used in the future to develop more accurate dynamic travel demands for use in traffic simulation models.