Institute of Transportation Studies at UC Berkeley

As jurisdictions update their High Injury Networks, discrepancies between the initial and updated HINs are to be expected. However, this lack of stability and consistency can negatively impact the prioritization of limited resources. In order to mitigate known issues with crash data underreporting and statistical biases, I examined strategies for utilizing data on underlying roadway characteristics to augment traditional collision analysis.

Using the City of Oakland as a case study city, I assessed the stability of the pedestrian High Injury Network across two consecutive five-year periods (2012-2016 and 2017-2021), created with the same methodology. I found that the two HINs identified similar segments, particularly along arterials, but were less consistent in identifying the segments’ start and end points due to variation in crash data. I propose a methodology for finalizing High Injury Network extents based on segment characteristics (number of lanes, posted speed limit, and functional classification), and intersection characteristics (traffic signal presence and estimated pedestrian volumes). Applied to the Oakland case study, this approach results in a High Injury Network that is more stable over time, more focused (fewer street miles), and captures a higher percentage of fatal and severe crashes. This approach has the potential to smooth over inconsistencies in crash reporting, reduce the frequency of network updates needed, and shift High Injury Networks from being reactive to more proactive.