Empirical Measurements of Travelers' Value of Travel Time Reliability

Travel time and travel time reliability are two fundamental factors influencing travel behavior and demand. The concept of the value of time (VOT) has been extensively studied, and estimates of VOT have been obtained from surveys and empirical data. On the other hand, although the importance of value of reliability (VOR) is appreciated, research related to VOR is still in its early stages. VORs have been estimated using surveys but has almost never estimated using empirical data.

This research used empirical data to take an initial step toward understanding the importance of travel time reliability. Katy Freeway travelers face a daily choice between reliable tolled lanes and less reliable but untolled lanes. An extensive dataset of Katy Freeway travel was used to examine the influence of time, reliability, and toll on lane-choice behavior. Lack of clarity on how travelers? perceive travel time reliability meant different measures of reliability had to be tested to see which best represents travelers? perception. In this research, three different measures of reliability were used, namely, standard deviation of travel time, coefficient of variation of travel time and travel time standard deviation relative to the total trip time.

Lane choice was estimated using multinomial logit models. Basic models, including only travel time and toll, yielded reasonable results. Models included VOTs of $1.53/hour, $6.05/hour, and $9.05/hour for off-peak, shoulder, and peak-period travelers, respectively. However, Adding different measures of reliability like standard deviation and coefficient of variation to the models resulted in counter-intuitive results. Positive coefficients for unreliability of travel time were obtained indicating that travelers, at least on the Katy Freeway, do not value travel time reliability as has been theorized in earlier studies on the same. It was concluded that additional research on how travelers perceive the reliability and time savings on MLs is needed because modeling real-world choices of MLs using empirical data and the standard definitions of reliability and time savings did not concur with the existing theory on travel time reliability and led to counter-intuitive results.