I lived, walked, and biked in DC for six years, until August 2021. I continue to read the local news and follow local efforts to improve safety in the city.

Since the COVID pandemic, traffic incidents are on the rise across the country. The limitations of Vision Zero plans put in place over the last decade are showing.

3 bicyclists were killed in as many weeks this summer. 2 women were killed as they sat on a restaurant patio. A child killed on their bike in a crosswalk.

With these incidents in mind, I retrieved DC's crash data from the city government's online portal. I planned to analyze the 2021-22 data to date and look at the hotspots for incidents involving bicyclists, pedestrians, and vehicles.

The city's crash data set showed 45 incidents resulting in 50 fatalities. The fatalities include 1 bicyclist, 22 pedestrians, 24 drivers, and 3 passengers.

I found 19 incidents missing from the dataset. These incidents resulted in 22 fatalities. The fatalities included 4 bicyclists, 7 pedestrians, 5 drivers, 3 passengers, and 3 uncategorized fatalities.

I researched the dataset further and found a number of significant caveats. MPD officers report to the scene of a crash and complete the crash report form. This report is completed and published to the dataset each week. However, the crash report is not updated. If a victim dies hours, days, or even months later that isn't updated in this data set. Additionally, MPD officers decide whether or not to call the Major Crash Unit based on the observable severity of injuries at the scene. Finally, not all crashes are mapped correctly or mapped at all.

To the city's credit, they were upfront with these caveats. However, on the same page where they list these caveats, they also state that the purpose of the dataset is to “provide crash location data to users interested in exploring spatial patterns and creating maps in support of safety analysis.” Since I was focused on traffic violence involving pedestrians and cyclists, I couldn't afford to use a dataset that was missing so many confirmed incidents.

Now I was confronting a new problem. More data than ever is available for us to download and use. This is a double edged sword. No dataset is perfect, so how do we decide what is good enough? The choice has implications on analysis and storytelling. Do we not tell a story because the data isn't perfect? Do we move forward knowing the conclusions we draw may leave something out?

22 of 72 total traffic deaths between January 1, 2021 and November 1, 2022 weren't included in the official public crash dataset. That's a 30% undercount. Most of the traffic deaths occurred in Wards 7 and 8. With this in mind, we should ask if Vision Zero efforts are reaching those they need to reach.

Vision Zero efforts include building bike lanes, lowering speed limits, automating traffic enforcement using red-light and speeding cameras, building self-enforcing streets, and increasing speeding citations. The map below shows several patterns.

There are very few bike lanes in Wards 7 and 8, forcing bicyclists into unsafe streets

The goal of Vision Zero is noble, but it may be time for cities to reevaluate their progress. This should involve looking beyond numbers and should focus on the equity of their effors.

Maps, Graphics, Text, Code

Aileen Clarke

Sources