A Research about Traffic Safety Character of two-lane highway in Plain Area Basing on Random-effect negative binomial model

This paper reviews the studies which have been performed in two-lane highway traffic safety research, and select random-effect negative binomial model to illustrate the traffic safety character of two-lane highway in plain area. More than 1380 kilometers(about 56 highways) two-lane highways in Beijing and Shandong province are investigated and the traffic safety data are collected, such as accident data, geometric data and roadside data. The highways are divided into 1772 segments. The result shows that the random-effect negative binomial model provides a better result than basic model. It is also illustrated that percent of motorcycle in traffic, the percent of truck in traffic, whole driveway density of the highway where the segment is involved, the percent of village length in a highway, driveway density of each sample are associated with higher total accident occurrence, and the width of road surface, the percent of bicycle in traffic tend to point to lower total accident occurrence.