Knowledge CentreTechnical Resources SearchConference PapersIdentifying High Collision Locations for Roadway Segments without Traffic Volume Information

Identifying High Collision Locations for Roadway Segments without Traffic Volume Information


Safety network screening is used to select road locations, such as intersections and segments, which are identified based on an abnormally high number of expected collisions or an unusually high proportion of a certain type of collisions. The most commonly used network screening method relies on rigorous safety performance functions (SPFs), which require traffic volume as an input. Many cities in Canada, including Saskatoon, currently do not collect traffic volume for every single segment within the city limits. The lack of traffic volume for a study network severely restricts the applicability of an SPF-based network screening method. On the other hand, the binomial and beta-binomial tests, which may be viewed as formal collision diagnosis tests, can be utilized to screen a roadway network, even for the roadway segments where traffic volume is unavailable. Unfortunately, previous studies that applied these two tests did not explicitly define the circumstances under which we can apply one (or both) of the tests. This study introduced a statistical test known as the ‘C(α) test’ to determine when we can apply, for instance, the beta-binomial test instead of the binomial test to screen roadway networks. This study used three years’ (2007-09) collision data from the City of Saskatoon to demonstrate the potential benefits of applying the collision diagnosis tests as a network screening tool. We also developed collision maps using ArcGIS to visualize the screening results, and to aid governing agencies’ decision-making processes in the selection of appropriate safety countermeasures for the screened hotspots.

Conference Paper Details

Session title:
Rajib K. Sahaji
Peter Y. Park
George Eguakun
Road safety