Abstract Three panel logit models that address correlations among repeated observations from the same Respondent, i.e., a Random Effect Logit Model (REM), a Random Coefficients Logit Model (RCM), and a logit model with both Random Effect and Random Coefficients (RECM), are developed and compared with each other. It is based on the data collected from a Stated Preference (SP) survey that explores expressway users’ diversion decisions under new expressway variable message sign information which provides travel times of an expressway and an alternate arterial road route. Results show that travel time saving and driving experience serve as positive factors, while the number of traffic lights on the arterial road, expressway use frequency, being a mid-age driver and an employer-provided car driver serve as negative factors in diversion. It is found that, all the panel models outperform the cross-sectional model; RCM fits the SP data slightly better than REM; RECM does not substantially improve the goodness-of-fit compared to RCM although the random effect does appear statistically significant; Random coefficients and random effect are basically complementary but somewhat interchangeable in a panel model. The findings of this study have implications for future efforts in driver behavior modeling and VMS systems assessment.
Investigation of driver response to the enhanced urban freeway variable message sign information
KSCE Journal of Civil Engineering ; 17 , 6 ; 1455-1461
2013-08-17
7 pages
Article (Journal)
Electronic Resource
English
Investigation of driver response to the enhanced urban freeway variable message sign information
Online Contents | 2013
|Driver response to variable message sign information in London
Online Contents | 2002
|Determinants of driver response to variable message sign information in Athens
Wiley | 2015
|Determinants of driver response to variable message sign information in Athens
IET | 2015
|