It is worth noting that the potentially-aggressive driver actions shown in Figure 1 are not mutually exclusive; in 8.4 percent of the crashes, a driver was coded as having committed two or more of these actions. Arguably it is more likely that a driver’s actions were committed deliberately, as opposed to accidentally, when a driver was coded as having committed multiple potentially-aggressive driving actions.
It is also likely that some crashes in which none of these potentially-aggressive actions were coded in FARS did in fact involve a driver who was driving aggressively. A salient example would be a case in which a driver commits an aggressive driving action that causes another driver to lose control of his or her vehicle and crash. If the vehicle being driven aggressively did not contact another crash-involved vehicle, and subsequently left the scene of the crash, it is unlikely that the aggressive action on the part of the driver of the non-contact vehicle would be captured in FARS. In 2006, a data element was added to FARS to record instances in which the police reported aggressive actions on the part of a non-contact vehicle; however, this data element was only used in a total of three fatal crashes in 2006 and two in 2007, which we suspect is a substantial underestimate of the true prevalence of this scenario.
One might argue that a driver who commits a potentially-aggressive action while impaired by alcohol is not performing the action completely voluntarily, in which case potentially-aggressive actions committed by intoxicated drivers might reasonably be excluded from the analysis reported here. To investigate the impact of this, the analysis was repeated, using multiply- imputed driver BAC data in FARS (see Subramanian  for explanation of this method), and only including crashes that involved at least one driver who was coded as having committed a potentially-aggressive action and as having a blood alcohol concentration (BAC) below the legal limit of .08 mg/dL. All drivers reported to have committed potentially-aggressive actions were legally intoxicated (BAC ≥ .08) in an estimated 33,524 of the 106,727 fatal crashes involving any potentially-aggressive actions. In the remaining 73,203 fatal crashes (68.6% of the original estimate; 38.2% of all fatal crashes over the period), potentially- aggressive actions were reported for a driver with a BAC below .08, the majority of which had a BAC of zero.
Along with the contributing factors analyzed here, FARS also includes a code for Road Rage / Aggressive Driving, which was added in 2004. This code does not distinguish between aggressive driving and road rage, however, as noted previously, these are widely considered to be fundamentally different. An act of road rage, as it is typically defined, is committed with the intent of causing physical harm to another road user, whereas an act of aggressive driving is committed with disregard for safety but not necessarily with intent to cause physical harm. Thus, the interpretation of this variable in FARS is unclear. Also, the FARS driver-related contributing factors include a code for Emotional (e.g., Depression, Angry, or Disturbed). Thus, it appears that the use of this code may in some cases suggest that the driver acted with aggressive intent, for example, if the code was used to indicate that the driver was angry; however, it may also be used to indicate that the driver was experiencing some other
© 2009, AAA Foundation for Traffic Safety