Unobserved heterogeneity analysis of rail transit incident delays

Passenger rail transit systems are frequently subjected to disruptions from incidents resulting in increased passenger waiting times, and loss of revenue to operators. The objective of the present study is to investigate the impacts of rail transit-related factors on the duration of delay, if an incident takes place. In order to account for the unobserved heterogeneity for each factor and the data generation process, a random-parameters hazard-based duration approach that accounts for these heterogeneities was employed in the present study. Out of the fourteen estimated parameters found to be statistically significant at the 1% significance level, seven parameters were found to vary across observations. This indicates that if the model was restricted and only fixed parameters were estimated, the inferences from the seven random parameters will not be accurate.