New approach for including traffic volumes in crash rate analysis and forecasting

A critical part of any risk assessment is identifying the appropriate measure of exposure to the risk in question. Incorporating vehicle miles traveled per year into highway crash analysis was an important step forward for identifying truly hazardous locations, as opposed to locations with high volumes but a low number of crashes per vehicle. This exposure measure is now ubiquitous in highway crash reporting, although it has not been proved to be the most accurate representation of highway crash risk. Research results demonstrate that the relationship between traffic volume and crash count is more complex than this and that it relates to quantities such as the distribution of traffic through the day and the types of crashes experienced. A proposed specification for models of crash count incorporates all these considerations in a form that is accessible to most highway crash analysts and permits comparison of crash rates between various highway locations in a way that accounts for the nonlinearity of the crash rate for traffic volume.