June 2, 2025
The Lab @ DC partnered with Vision Zero DC and the District Department of Transportation on a project to promote safe driving behavior. We built a predictive model to identify vehicles at highest risk of a crash in the District using data on tickets from DC’s Automated Traffic Enforcement camera program. Then, in a randomized evaluation, we tested whether sending a personalized text message and/or letter to owners of the highest risk vehicles could change their driving behavior. We designed the messages based on best practices from behavioral science to tell vehicle owners about their risk relative to other drivers and what they had to gain by driving more safely. Our evaluation looked at crash involvement and tickets three and 12 months after we sent the messages and found no effect on crashes or tickets. Read more about the study and what we learned.
Vision Zero DC also conducted interviews with some of the high-risk drivers we identified in our predictive model. Drivers they interviewed said that the messages were compelling—for example, some people shared that they were struck by the information on the personalized letter about past tickets they had received and how much they had been fined.
While we found no impact of the one-time messages on driving behavior during the pilot, we learned a lot through this project about identifying risky drivers and developing behaviorally informed messages. Alyssa Huberts, a Social Scientist at The Lab and co-captain on the project, recently sat down with Charlie Willson, Vision Zero DC’s Director, and Charles Turner, Associate Director of Automated Traffic Enforcement for the District Department of Transportation. They discussed some of our lessons learned and how this work fits into the broader approach the District is taking to curb risky driving. This conversation has been edited for length and clarity.
Alyssa Huberts, The Lab @ DC:
This pilot, which sent personalized messages to risky drivers, is part of the broader Vision Zero effort in DC to eliminate traffic deaths. How do you see sending these personalized messages to risky drivers as fitting into the broader goals of Vision Zero?
Charlie Willson, Vision Zero DC:
This project was building upon a camera program that we know is extremely effective in slowing people down, getting people to follow the traffic laws and drive safer. DC is a leader in using safety cameras. We have more cameras and more types of cameras than basically all other cities. We also run the program in an extremely rigorous and data-driven way.
The camera program gives us a really unique opportunity to know exactly which vehicles are risky. Then we were able to send that data to the Lab @ DC, which, with their expertise, was able to create a model to figure out exactly who those drivers were who are more likely to be involved in crashes, and we were able to direct messages to them with a really unique and behaviorally informed message. We know that a lot of folks think they're just the average driver, and what we were telling them was that they are not.
Charles Turner, DC Department of Transportation:
One thing I love about coming to work here in the District is that we'll try the next step. And I think this pilot is a pure example of, OK, what else can we do? It was very, very timely that we attempted to go out and focus on this group of drivers with multiple citations. Then we also try to get some of them in for interviews and ask them, “Why do you consistently speed or run a stop sign?” What it did was give us some indications of how we can improve our overall safety campaign.
Alyssa:
I love hearing about how this is just one piece of a data-driven ethos. It's not just a single data project; it's fitting into the broader scale of how we can use these cameras both for enforcement and as a source of data to make better decisions iteratively.
Could you tell us about what you think you gained from this project?
Charlie:
In addition to being able to do this as a randomized control trial, working with The Lab allowed us to craft the messages in a way that was behaviorally informed in a way that we thought was likely to influence people. The predictive model itself, just showing that there is a relationship between camera violations and likelihood of being involved in a crash was also valuable.
Charles:
And as we put the project together, The Lab looked at national and international scenarios in a similar format of going out to individuals and nudging them on safety. So, one great thing that came out of this: whatever step we take next, at least we have that baseline knowledge to project from.
Alyssa:
One of the things that surprised me is that there is a large literature on risky driving interventions and how we can change people's behavior, but we found that those studies were overwhelmingly done in a laboratory setting, testing with self-reported driving behavior as the outcome, rather than in the field with real drivers. So, there's not a lot of evidence behind what works in the real world.
Charlie:
In this pilot, we knew some of these drivers are at the top of the risk distribution, people who are already willing to get multiple camera tickets that have fines attached. So, this is a group that's not easy to nudge, right?
Meanwhile, the kinds of outreach and education campaigns that this might be complementing are mass media campaigns. And very few of those have been analyzed using the kind of rigor that that this one was. They're certainly done strategically and in a way that allows for continuous improvement, but it would be hard to find a communications technique that was this cost effective that has a statistically significant effect on something like driving behavior for multiple reasons.
We have reason to believe that people's behavior can be nudged, but it's going to be almost on a cultural level. Like for example, 50 years ago, it was a little bit more socially acceptable to drive drunk, right? So, you have a few drinks and still drive. That was the kind of thing you might mention around the water cooler or at your family reunion, just sort of offhand as a joke. Well, in 2025, it's not acceptable anymore within mainstream American society. It would be hard to find one communications campaign that got us there. But what we do know is that over decades, we have seen that improvement. My hope is that similarly, driving too fast, speeding, which is somewhat socially acceptable now, will have that same progression over time. And a message like this is a way to try to nudge a key part of the population in that direction.
Alyssa:
This seemed like a powerful opportunity to target, to personalize, to inform drivers about their own behavior and how it compares to others. And we had enough statistical power. Yet we have a clear indication that there was no effect on future tickets or on crashes. Where does this lead you in terms of questions it generates? What would a version 2.0 look like?
Charlie:
There's lots of different ways we could do it differently. We sent one message once, so maybe people need to be reminded. And then, while the message was well considered and, we thought, the kind of thing that would resonate with people, there are a lot of different ways to present it. One of the things that was fascinating to me was that some of our interview participants said that when the letter listed off the tickets they had previously received and the prices, that was the first time they’d ever thought about how much money they were spending on these tickets. I thought everybody would be thinking about that. It was fascinating to me to learn that actually, that's not how people always think.
The big thing that excites me about it is that this is such a low cost way to reach exactly who you want to reach.
Alyssa:
I would love to hear how you both think about reaching different populations and the range of folks who are getting these tickets. What are the tools that are available to you, and what are some of the limitations of the DC context?
Charles:
It gave us the idea to put messages on the back of a warning ticket. In the old days, the back of the warning ticket was blank. So to make folks aware of the new cameras on buses, we put a message about safety related to clear bus lanes and clear bus stops on the back of the warning notices that used to be blank.
In terms of context, crashes and injuries are typically on the perimeters of the city: on the Maryland side and on the Virginia side. These roadways were built back in the 40s, 50s, and 60s, where it was an auto-centric mentality about “get the people in town” and “get them out of town.” That mentality's changing, particularly in the District. Long term, what's going to happen is to rebuild those roads with bike lanes and roads in such a way that it doesn't beckon you to speed.
Charlie:
We're going to use all the tools in our toolbox. As Charles mentioned, it will involve roadway engineering. We will continue to invest in transit and Capital Bikeshare, so that people have options for how to get around. It's going to involve messaging. It's going to involve enforcement. We're expanding the number of vehicles that we boot and tow. Just recently, the Office of Attorney General sued some out-of-state drivers who owe a lot of money on tickets. It's going to be a multifaceted approach.
Alyssa:
Great, well, that is a positive note to end on. I hope to continue talking about the next version of the project: what's coming next for all of this, and how you learn and implement it in the next stage. In the meantime, I will continue to drive safely and stay out of your risky driving population.