Do the Same Factors Affect Accident Frequencies on Highway Segments with Different Traffic Volumes and Traffic Compositions
MetadataShow full item record
Accident frequencies can be influenced by the level some motorists are alert, or by how some motorists perceive risk, under different conditions. For example, under low (overall or truck) traffic volume conditions, some motorists may be less alert while driving, which can increase their accident occurrence likelihood. At the same time, some motorists may compensate for the same low traffic volume conditions and drive faster, which also results in accident prone driving behavior. The opposite also stands; high (overall or truck) traffic volume conditions can keep some motorists more alert while driving, or can make them drive slower to compensate for the high-risk conditions, which in both cases reduces the motorists’ accident occurrence likelihood. This thesis seeks to investigate the possibility that different factors can affect single-vehicle accident frequencies, when the latter are observed on highway segments under different traffic volumes (e.g., low, medium, and high volumes, based on observed traffic data) and traffic compositions (e.g., low, medium, and high truck volumes, characterized by the amount of trucks observed over the same segment and period of time). To account for risk-taking and perception-specific heterogeneity, random parameters negative binomial models are estimated, and through the use of likelihood ratio tests, it is shown that different parameters (such as roadway geometrics, pavement condition, vehicle counts, and environmental characteristics) affect single-vehicle accident frequencies on highway segments under different traffic volumes and traffic compositions. The findings of this thesis should be viewed as an incremental step towards identifying safety countermeasures and high crash locations that are volume- and composition-specific.