Investigating the Heterogeneity of Postencroachment Time Thresholds Determined by Peak over Threshold Approach

Inconsistency in the definition of conflict is a fundamental issue of the traffic conflict technique and raises questions about the validity of this technique. A significant aspect of inconsistency is that thresholds to distinguish traffic conflicts from normal events have not been clearly determined. The study presented in this paper proposed a peak-over-threshold (POT) approach to determine postencroachment time thresholds between traffic conflicts and normal events. The determined thresholds were evaluated by testing the correlation between the defined conflicts and observed crashes. A further regression analysis was conducted to explain heterogeneity of POT-determined thresholds. The results show that traffic conflicts defined by POT-determined thresholds have a relatively strong relationship with observed crashes, and the Pearson's correlation coefficient is .66. Moreover, the threshold heterogeneity in this study mainly stems from the variety of exposure (i.e., traffic volume). This study also implies that the POT approach, which can account for possible heterogeneity in thresholds, is promising for improving the validity of the traffic conflict technique.