MS2 Press

The golden rule of keeping it “apples to apples” is always an essential part of making a proper comparison of multiple items.  For the most part, this mentality has been widely utilized in the traffic safety world when analyzing intersection related traffic crashes.

Traffic engineers/planners and law enforcement officials primarily identify intersection related crashes through a pre-established distance from the center of the intersection and/or an officer’s selection of the crash being “intersection related” on the actual crash report.  Utilizing these criteria, network-wide analyses are then performed to identify the intersections with the highest crashes, crash severity, crash rate, etc.

But is this really an “apples to apples” analysis or are there some oranges in there?

Roadway intersections can have very different characteristics and attributes.  Items such as alignment, geometrics, traffic volumes, and speed limits can vary greatly when looking at intersections throughout an agency’s road network.  A recent article in the Institute of Transportation Engineers (ITE) Journal(1) examines this issue and explores a methodology to more accurately identify intersection related crashes.

The article states that the default distances used by many state agencies range anywhere from 50 feet to upwards of 500 feet to determine if a crash is considered related to the intersection.  Other agencies utilize items such as vehicle movements, collision type, and the “intersection related” crash report selection as the criteria to identify intersection related crashes.

The authors go on to discuss a detailed analysis that was performed on over 13,000 crash reports in the state of Florida.  They discovered crash reports that should have been coded as intersection related (~ 15%) and examined possible relationships between intersection attributes and distances from the intersection to still consider a crash as intersection related.

I would like to say that the article concludes with some sort of cookie cutter parameters that can be applied to various intersections types.  The fact is that the variability in crash data from the initial crash report to the subsequent analysis will never be able to provide standards that can be applied in all cases.  However, the article brings to mind that it is important to recognize that there can be significant differences in intersections and, therefore, prudent to analyze similar intersections rather than intersections as a whole to actually conduct more of an “apples to apples” comparison.

(1) Abdel-Aty, M., Wang, X., and Santos, J. “Identifying Intersection-Related Traffic Crashes for Accurate Safety Representation.” ITE Journal, Dec 2009.