For years, great attention has been drawn to the Dilemma Zone (DZ) problem as it has been questioned as one of the causes to the high traffic accident records at signalized junctions. This problem has far been investigated in engineering perspectives, but few attempt to investigate influences of drivers’ decision making behavior. Since Zegeer and Dean (1978), Logistic regression model has commonly been adopted to simulate drivers’ STOP/GO decision and figure out DZ boundary; Discrete choice model, for instance Logit model which has similar underlying nature as the Logistic regression model has however few been used. The un-popularity of the Logit model might be because it requires individual attributes for establishing utility functions of STOP and GO decisions; only field observed data cannot be used for this purpose. Therefore, a new measure termed generalized distance was proposed. The proposed measure is of two folds: (1) it is used to quantify and assess drivers’ decision making behavior; and (2) it can be used to establish utility functions of STOP and GO decisions and thus Logit model was adopted in this study. The results reveal that, during the observation period, drivers were more likely to GO at amber; and, interestingly, drivers who maneuvered over speed limit were more likely to STOP. The findings of this study also suggest that drivers’ decision making behavior should first be assessed as to decide appropriate road safety interventions for the reported problem.
↧