Author: William D. Berg

Date of Publication:  1999 (?)

Sponsoring Agency:  U.S. Transportation Research Board, Committee on Railroad-Highway Grade Crossings

Abstract:

A significant portion of recent research on rail-highway grade crossings has focused on the development of effectiveness ratios for various improvement actions and the incorporation of accident history into accident prediction and resource allocation procedures. Data are presented to demonstrate that these procedures can introduce significant bias and lead to a misallocation of resources. A preferred procedure is to use a modeling approach to estimate expected accident reductions and to ignore the adjustment of predicted accident rate based on recent accident history. The latter adjustment is shown to repre- sent a modeling of the regression-to-the-mean phenomenon and as such is not appropriate for resource allocation studies.

A significant portion of recent research on rail-highway grade crossings has focused on the development of effectiveness ratios for various safety improvement actions and the incorporation of accident history into accident prediction and resource allocation procedures (1-6). The purpose of this paper is to offer some additional theoretical and practical insights into the interpretation and application of effectiveness ratios and accident history in the selection and programming of crossing improvements. Data will be presented to demonstrate that the current recommended resource allocation procedures can lead (and presumably are leading) to a misallocation of resources for grade crossings. Finally, recommendations will be offered on the direction of future research on grade crossings.
 

No. of Pages:  13