01047603

Component

This study examined relationships between crash rates and certain cross-section design, operational, and access attributes. The scope included all urban multilane roadways on the Arkansas state highway network, excluding fully controlled-access roadways (i.e., freeways). The methodology reflected a concern for the potential effects of the uneven distribution of certain attributes among the median types. Negative binomial regressions indicated that the crash rates increased when the traffic volume increased, the number of traffic signals per mile increased, the access density increased, or the commercial-and-industrial access density increased. Also, a multiple-factor model that included volume, access density, and median type was developed, although median type was not a statistically significant variable. The differences between two crash prediction models, one fit over the entire range of data and the other over a smaller but better populated subset, indicates that models that compare the crash rates of alternative cross-sections can be affected by the uneven distribution of the variables across the alternative treatments. This suggests that data distribution should be critiqued before proceeding with model development.

01042056

07-2936

English

20p

2007

Transportation Research Board 86th Annual Meeting

CD-ROM

Figures (9) ; References (6)

Crash rates; Cross sections; Freeways; Highway design; Highway operations; Medians; Multilane highways; Traffic signals; Traffic volume

Arkansas

Design; Highways; Planning and Forecasting; Safety and Human Factors; I21: Planning of Transport Infrastructure; I82: Accidents and Transport Infrastructure

Transportation Research Board Annual Meeting 2007 Paper #07-2936

TRIS, TRB

Feb 8 2007 7:39PM