Highway engineers frequently look at collision data independent of exposure rates or injury severity when considering what changes to make to reduce collisions. For instance, a fairly low speed urban street between a college campus on one side and bars and restaurants on the other my have one of the highest pedestrian collision counts in the state despite having an extremely low collision rate normalized per million pedestrian crossing events. A suburban arterial may have a low number of pedestrian collisions despite having a the highest collision rate in the state normalized based on the number of pedestrian trips there.

The highway engineers may elect to modify the first road (at the college campus) and do nothing at the second (the suburban arterial), because the first location, having the largest number of collisions, can result in the largest reduction if the engineers' actions are successful. The second location, despite having a higher fatality rate and more severe injuries to the pedestrians, may not get any action. I base these examples on two real-world locations and resulting engineering decisions that have been made in central North Carolina. (This doesn't mean that I endose those actions - I believe the suburban location should have received some minimal improvements to bring it in line with MUTCD, at least.)

In Cary, we haven't had any cyclist fatalities in a very long time, and if we broaden the region for analysis to include surrounding cities, the fatalities have been evenly distributed among intersection errors and mid-block overtaking scenarios (including where motorists swerved off the roadway onto shoulders). Among non-fatal collisions, intersection and crossing movements are clearly the primary problem, and wrong-way sidewalk cyclists are greatly overrepresented compared to roadway cyclists and same-direction sidewalk cyclists. Overtaking collisions on roadways were rare, and rarer still were overtaking collisions in wide lanes. Therefore, it became clear that the city's prior plans to designate more sidewalks as bikeways could not reduce the number of car-bike collisions, since it would most likely increase the number of cyclists riding against traffic on sidewalks, and this was the single most common existing cause of collisions. The engineers instead decided to increase wide lane or on-roadway bike lane treatments since such designs were not associated with as many collisions.

Transportation consultants have attempted to gather exposure rates in Cary, such as by polling cyclists about their behavior. I would like to see more of this done. Exposure rates allow us to measure relative safety about different locations and actions, so that we can make better decisions about best practices on a per-user basis. However, there are other practical safety-related planning and engineering decisions that can be made without exposure data.

I personally promote my Road 1 course for the purpose of increasing confidence and consequently encouraging cycling. I believe it increases safety per mile traveled, but I cannot guarantee this, and since I know that the increased confidence that cyclists express from having taken the course has increased the miles they travel and increased the level of challenge they are willing to undertake, it is possible that the net safety effect per student comes out as a wash (although probably not the safety per urban cycling mile or per intersection negotiated). Regardless, the students like the class, and consistently report that they are glad they took it. This includes the transportation engineers and planners who have taken it. Until we can get better exposure data, that's good enough for me.