Crash reduction factors are used to identify and prioritise the most effective safety improvement measures, and prioritise and allocate available resources optimally for a highway safety improvement project. Simple before and after analysis does account for the regression-to-the-mean bias. This research employs an Empirical Bayes (EB) methodology that overcomes the regression-to-the-mean property that is encountered in traditional before and after analysis. Traffic, geometric and crash data for both the treatment and comparison sites were collected from Ohio in developing the crash reduction factors. Using data collected from Ohio, the EB methodology was applied in developing crash reduction factors for the following improvement categories:add a two-way left turn lane, install a median barrier, flatten slope and remove guardrail, remove or relocate a fixed object, flatten vertical curve, providing highway lighting and close median opening. (Author/publisher)
Abstract