Citation

Andreassen DC, Hoque MM. Traffic Eng. Control 1986; 27(10): 514-517.

Affiliation

Australian Road Research Board, Aust

Copyright

(Copyright © 1986, Hemming Group)

DOI

unavailable

PMID

unavailable

Abstract

Estimates of bias for the effecfs of countermeasures require a model for the distribution of accident frequencies. The distribution of 'collisions between vehicles from adjacent approaches' accidents per intersection was examined for a period of six years for Metropolitan Melbourne. The frequency percentage distribution appears to be constant over time. The negative binomial distribution did not suit the data; however, the logarithmic series distribution was found adequately to describe the data. A new test was used to judge the adequacy of the model to the observed data. The data were subdivided by the functional classes of the intersecting roads and the logarithmic series distribution was still found to describe well the distributions within all the road classes. Formulae for estimating bias for blackspot work should therefore be based on the logarithmic series distribution rather than the negative binomial.