Type of Collision and Crash Data Evaluation at Signalized Intersections
THIS FEATURE ADDRESSES THE DIFFERENT FACTORS THAT AFFECT CRASHES (BY TYPE OF COLLISION) AT SIGNALIZED INTERSECTIONS AND THE EFFECT THAT INCOMPLETE DATA HAVE ON THE FINAL RESULTS. DIFFERENT FACTORS WERE FOUND TO CAUSE DIFFERENT COLLISION TYPES AND INCOMPLETE DATA CAUSED CHANGES IN FACTOR IMPORTANCE. THE RESULTS ARE PROMISING FOR ENHANCING UNDERSTANDING OF THE SAFETY OF SIGNALIZED INTERSECTIONS.
Traffic crashes affect everyone. According to Cafiso, Lamm and La Cava, in the millions of crashes occurring worldwide each year, more than 500,000 people are killed and more than 15 million are injured.1 This corresponds to one crash-related death every minute. Although crashes can be attributed mostly to human error, it is suggested that the design of a roadway also can be responsible for causing crashes.
For example, 243,409 crashes were recorded in tJie Florida crash database during 1999. Of these, 98,756 crashes occurred at or were influenced by a signalized intersection. To describe the seriousness of these numbers, 98,756 crashes corresponds to one crash every 5.5 minutes.
Likewise, Bhesania found that, out of several thousand crashes in Kansas City, MO, USA, signalized intersections experience the largest number of incidents.2 More specifically, 9.6 crashes occur at signalized intersections per year compared to two per year where STOP or YIELD signs control traffic. This further validates the point that roadway inter sections are a common place for crashes, which may be due to several conflicting movements as well as a myriad of different intersection design characteristics.
The factors affecting crashes are not well defined. This lack of knowledge may be the source of additional crashes. There is a need to classify intersections and to quantify the effects that certain geometric aspects have on the number of crashes at a specific intersection.
Furthermore, when a crash occurs and the local police department is notified, the responding officer will determine whether to fill out a long-form or short-form crash report. For instance, if a crash involves an injury or a felony (such as a hit and run), the crash must be filed on a long form. If a crash involves only property damage (a minor crash), it usually is up to the officer to report it on a long or a short form.
Crash forms then are forwarded to their respective counties, which choose whether to file crashes reported on short forms. Only the crashes reported on long forms are forwarded to the Florida Department of Transportation (FDOT) and the Department of Highway Safety and Motor Vehicles (DHSMV), which maintain records based only on crashes reported on long forms. Many states have similar practices, which are based mostly on the towaway threshold.3
Because most crashes involve only property damage and not a serious injury or a felony, it can be argued that the FDOT and DHSMV crash databases under-represent minor crashes as well as certain types of crashes that frequently involve property damage only. Moreover, by keeping track of long forms only, these agencies exaggerate the fraction of crashes that involve a serious injury, which makes roadways appear less safe.
Therefore, by excluding minor crashes, any models developed will under-represent the true number of crashes that occurred at a location and may cause a difference in the significance of the crashrelated variables.
The rationale behind conducting this research is that vehicle crashes are common occurrences at signalized intersections. This study explores the hypothesis that different types of collisions are affected by different independent factors. Furthermore, the authors investigate the significant differences in the important crash-related factors between models based solely on crashes reported on long forms and models based on crashes reported on both long and short forms (models based on restricted and complete datasets).
Additionally, several databases were cross-checked to ensure completeness of data. The chief intention was to create statistically significant analyses for two datasets-one for the restricted dataset and one for the complete dataset-and to compare the results. The authors anticipate that these results will contribute significantly to the area of safety at signalized intersections and will consider the possible consequences of analyzing restricted datasets.
DATA COLLECTION
The data collection for this effort was extensive in that it required information on all crashes, including minor crashes. Only jurisdictions that maintained records for crashes reported on both long and short forms were contacted for cooperation. The ones used in this study were Brevard County, Seminole County, the City of Orlando and Hillsborough County. These provinces compose a significant portion of central Florida.
Each district provided hundreds of intersection drawings, which were sorted one by one. Intersection geometry was recorded into a database. The next step was to retrieve crashes from the intersections with geometric information available. The first source of crash records was the jurisdiction itself. The next two sources were the FDOT and DHSMV databases. Each source was cross-checked against the other two to ensure that all crashes were accounted for from the individual jurisdiction.