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.