The statistical analysis of highway incident duration has become an increas
ingly import research topic due to the impact that highway incidents (vehic
le accidents and disablements) have on traffic congestion. In addition, the
re is a growing need to evaluate incident management programs that seek to
reduce incident duration and incident-induced traffic congestion. We apply
hazard-based duration models to statistically evaluate the time it takes de
tect/report, respond to, and clear incidents. Two-year data from Washington
State's incident response team program were used to estimate the hazard mo
dels, The model estimation results show that a wide variety of factors sign
ificantly affect incident times (i.e. detection/reporting, response, and cl
earance times), and that different distributional assumptions for the hazar
d function are appropriate for the different incident times being considere
d. It was also found that the estimated coefficients were not stable betwee
n the two years of data used in model estimation. The findings of this pape
r provide an important demonstration of method and an empirical basis to as
sess incident management programs. (C) 2000 Elsevier Science Ltd. All right
s reserved.