Uncertain travel times are characteristic of real-world urban transportation networks, and clearly travelers' risk preferences and behavioral attitudes regarding this stochasticity vary according to the trip purpose and the specific driver. Recent modeling advances allow one to determine the optimal routing policy for a broad class of these preferences, thereby enabling comparative analysis of trip characteristics made by travelers of eachpreference type. This paper undertakes numerical comparison of optimal policies resulting from four families of disutility function (linear, deviance-based, monotonic quadratic, and Box-Cox). These results indicate that the deviance-based disutility functions, and possibly certain Box-Cox functions yield policies that are substantially different from the linear ones frequently assumed in practice, whereas monotonic quadratic disutility functions never result in different policies than those resulting from lineardisutility functions.
Abstract