Abstract
By using historical crash data, several regression-type models were developed with which to estimate the effects of the installation of cable median barrier on crash rates for several crash types while taking into account several other factors associated with variation in crash occurrences or crash rates. The data for these analyses consisted of counts of crashes of the various types for each section of North Carolina Interstate highway, along with the associated roadway characteristics, for 1990 through 1997. Certain roadway characteristics (in particular, traffic volume) changed from year to year as, of course, did the crash counts. The entire North Carolina Interstate system was taken as a reference population; a negative binomial error structure was assumed for many of the models, and the models contained specific effects for each crash year to account for various unmeasured effects that may vary from year to year (e.g., weather conditions). These models used the reference population to predict numbers of crashes on the cable sections. After cable barrier installation, none of these crashes occurred. In summary, these analyses indicate that several types of crashes (e.g., ran-off-road-left, hit-fixed-object) increased on the sections where cable median barrier was installed. However, these sections showed improved overall safety through fewer serious and fatal crashes as well as fewer head-on crashes. Overall severity index values were greatly reduced after cable barrier installation.
Get full access to this article
View all access options for this article.