Evaluations of collision data inevitably require summarizing the number of collisions or injuries occurring on a roadway or region. Target safety goals, baseline analyses of conditions and comparative rankings are all based on counts. However, when reviewing the numbers it’s important to make sure you are comparing apples to apples and not apples to oranges — aka collisions to injuries (or fatalities). During my time at UC Berkeley’s SafeTREC, this collisions vs injuries issue frequently presented itself and caused confusion during conversations or when preparing numbers for various reports and presentations.
To dig deeper, let’s continue with the pedestrian collision example from our previous blog post and review pedestrian involved collisions. Using the San Francisco 2011 demo set on RoadSafe GIS, we can query for pedestrian involved collisions. For this sample we also want to narrow down by collision severity and select only collisions with an injury or fatality, not property damage only. The Tables & Charts summary page will show a collision and a victim summary:
For this query 825 collisions resulted in 876 injuries plus 16 fatalities for a total of 892 victims. This is because every collision that is classified as involving an injury or fatality will have at least one or more injured victims. Victims will always be higher than collisions if you are not including property damage only collisions. This point is frequently overlooked and will cause confusion if everyone is not on the same page.
A good example of summarized data appears in the nationally available Traffic Safety Fact Sheets published by the National Highway Traffic Safety Administration (NHTSA). Traffic Safety Fact Sheets are national summaries of fatal collisions and injury estimates published each year for different types of collisions. Take a look at the 2012 Pedestrian fact sheet.
If you review the fact sheet closely you will see that most of the summary refers to the number of fatalities, not the number of fatal collisions. You can easily see there were 4,743 pedestrians killed in 2012 from the tables and at first glance it appears the entire fact sheet refers to only victim counts. An additional layer of information, fatality rates, is also introduced by dividing the number of fatalities by the corresponding population. All of this data can obscure the fact that one of the tables actually refers to the number of fatal collisions, not the number of fatalities.
Can you figure out how many fatal collisions caused the fatalities? Have a look at Table 5 – Alcohol Involvement in Pedestrian Crashes. Table 5 is actually using the number of fatal collisions, not fatalities like the rest of the fact sheet. Looking at the total column you will see 4,657, which is the number of fatal collisions. That is 86 less (4,743 minus 4,657) than the number of pedestrians killed, which is to be expected as previously explained.
So be careful when you are reviewing traffic safety numbers or generating your own with local data. The addition of denominators such as population or miles traveled to create rates can also further complicate analyses. Take time to understand the data and whether it is collisions or injuries.