Examining Heterogeneity of Driver Behavior with Temporal and Spatial Factors

Temporal and spatial characteristics of the road environment are known to influence driver behavior and, consequently, the risk of a crash that causes injury or fatality. Nonetheless, much of the understanding of the risks of injury and fatality associated with driving relies heavily on police crash records. These records capture the most serious of crashes but underreport other events. Studies that rely on these data sources typically ignore the temporal and spatial factors. Advances in technology have enabled more detailed study of driving on a day-to-day basis and, therefore, provided the opportunity to examine driver behavior for the same driver across time and space. However, this ability has its own challenges. These challenges include extensive intra- and interdriver heterogeneity, which is not apparent when traditional data collection methods are used. A framework and a methodology for isolating the influence of drivers' inherent characteristics on driver behavior are presented. This study was done by constructing temporal and spatial identifiers that controlled for the influence of the road environment. Results include analyses conducted by using empirical driving information collected from 106 vehicles in Sydney, Australia, to examine the effectiveness of this approach. The results indicated that in 80% of road environments there was less intradriver variability in speeding behavior than interdriver variability when temporal and spatial characteristics were accounted for. Clustering and regression analyses for the most frequently observed road environments are also presented. Driver personality characteristics are significant for evening trips home on residential roads, and acceleration profiles are significant in evening trips on roads with 50- and 60-km/h speed limits.