Data-Smart City Pod

Addressing Racial Disparities in Traffic Deaths

Episode Summary

In this episode Professor Stephen Goldsmith interviews researchers from the Harvard TH Chan School of Public Health and the Boston University School of Public Health about their new report, which reveals that racial disparities in traffic fatalities is worse than previously thought.

Episode Notes

In this episode Professor Stephen Goldsmith, Ernani Choma (postdoctoral research fellow at Harvard), and Matt Raifman (Ph.D. candidate at Boston University) discuss their new report, which revealed that the racial disparities in traffic deaths is actually worse than previously believed. They discuss why this is the case, what still needs to be studied, and how cities can collect better data to address the latent racism built into transportation systems and infrastructure. 

Music credit: Summer-Man by Ketsa

About Data-Smart City Solutions

Data-Smart City Solutions, housed at the Bloomberg Center for Cities at Harvard University, is working to catalyze the adoption of data projects on the local government level by serving as a central resource for cities interested in this emerging field. We highlight best practices, top innovators, and promising case studies while also connecting leading industry, academic, and government officials. Our research focus is the intersection of government and data, ranging from open data and predictive analytics to civic engagement technology. We seek to promote the combination of integrated, cross-agency data with community data to better discover and preemptively address civic problems. To learn more visit us online and follow us on Twitter

Episode Transcription

Betsy Gardner:

Hi, this is Betsy Gardner, senior editor at the Harvard Kennedy School and producer of The Data-Smart City Pod. Since we started this podcast, we've had great support from our listeners, and to make sure that you don't miss an episode, please find us under the new Data-Smart City Pod channel wherever you listen. Make sure to subscribe so you get each episode, and thanks for listening.

Steve Goldsmith:

Hello, Steve Goldsmith, professor of Urban Policy at the Harvard Kennedy School, with another one of our podcasts that looks at how to improve the quality of urban life through the use of data. Today, we have a really interesting session with Ernani Choma, who is a research fellow at the Chan School of Public Health at Harvard. And we have Matthew Raifman, who's a Ph.D. candidate at the Boston University School of Public Health. Welcome to our podcast. Thank you very much for being here.

Matt Raifman:

Hey, I'm Matt Raifman. I'm a Ph.D. candidate at Boston University at the School of Public Health, and I work on transportation climate research. And my background is somewhat varied. It involves performance management at the state level with the governor's office in Maryland as well as the Bloomberg What Works Cities Project. And more recently, I worked at Ford Smart Mobility, where I did autonomous vehicle pilots in Miami and Washington DC.

Ernani Choma:

My name is Ernani Choma, and I'm a research fellow in the Department of Environmental Health at the Harvard T.H. Chan School of Public Health, and my research is on health risk assessments. So basically, I assess risks associated with environmental exposures. And much of my research is focused, for example, on air pollution risks from vehicle emissions, and my primary interest in using this... assessing this risk to inform decision making, such as policy decisions. Thanks for having us.

Steve Goldsmith:

We invited you because we read your recent publication in American Journal of Preventive Medicine on Disparities in Activity and Traffic Fatalities by Race and Ethnicity. And we've been dealing with a number of cities across the country on this particular issue, which is how does a city use its tools and levers to improve safety, particularly in communities of color?

So let's start with first this concept that you are looking at disparities adjusting for differential exposure. And what does that mean? Let's just start with what does that mean and how is it different?

Ernani Choma:

So I think much of the past research has focused on estimating this disparities on a core cap on a core population basis. And so, our idea of adjusting for exposure is essentially account for how many miles each population drives by each mode of transportation. So how many miles they bike, how many miles they walk. So we get this measure of fatalities per mile travel, right.

And so, we estimate the disparities by race and ethnicity in fatalities per mile travel as opposed to just fatalities per capital. And when we do this adjustment because different populations of different race, ethnicity, travel different amounts of miles in each mode, we actually find that the disparities are much larger than what past research had found when I just simply account and normalized by the number of population.

Steve Goldsmith:

So Matt, let's stay with this for a second because this is pretty fundamental to our conversation today. Is this another way of saying that because, in less affluent communities, the modes are more dangerous than they are in affluent communities? Like if there's more walking, there's more bike riding, there's more nighttime walking that the danger levels go up, or is that a misunderstanding of your research?

Matt Raifman:

Well, I don't think that's a misunderstanding. I think that's true that we find that in the data. But the bigger picture is that regardless of the mode, we see these disparities, and I think that's really notable. So yes, it's true that we seem to find evidence that during darkness, the rate increases, and it may be that people of color who are lower income perhaps are more likely to be coming home from work at that time.

Matt Raifman:

But it's also true that regardless of the mode, and regardless of whether we're looking at all hours of the day or just darkness, we still see these disparities. So that's suggestive that there's something else more fundamental going on than just the types of trips that people are taking.

Ernani Choma:

What I wanted to add is in many cases, we actually see in the data that, for example, Black Americans, they actually cycle less than white Americans and that we do not identify what is the reason to feel safe or they don't have access to transportation that way. And those disparities in the miles traveled, they're also mode, right.

I mean, when you see that some groups cycle less than others would know the reason why, but it could be related to they don't even feel safe because they don't have the property infrastructure in place. Right.

Steve Goldsmith:

So what you put together is some possibilities for this disparity in your article road placement, underinvestment and alternative modes of transit, disproportionate traffic stops, passenger driver pairing, and ride-hail. Could you just talk about those a little bit in terms of just help our listeners understand a bit more kind of what's causing the disparities so that they can think about how to resolve it more?

Matt Raifman:

Yeah, sure. So there's a fair amount of literature that looks at disparities by race and ethnicity in urban planning. And in some of that research, we see that there's generally a systematic underinvestment, in areas, in particular in walking and biking infrastructure. There's also this twin challenge to that of gentrification. And the theory that when you place these bike lanes or improved pedestrian infrastructure, mobility infrastructure in general, it may lead to gentrification, which then begets more disparity.

So there's this complicated relationship there that is present but hard to unravel and hard to disrupt. There's also the potential that, when you're in a crash, so say there is a crash, there's also the potential that you don't go to the same hospital. So there is some evidence to suggest, particularly in motorcycle accidents. But also, with other forms of traffic incidents, that people of color are taken to a different hospital than people who are white.

And the health outcomes may be different, even if both a white person and a Black person are exposed to the same type of crash. So it's actually interesting because the sort of safe systems approach, which I think we can get to later in discussion, which is perhaps one of the solutions to implement actually shows up in the cause structure as well.

So if you look at the types of trips people are taking, the infrastructure that people are using, as well as the post-crash care that they're receiving. There is research that suggests disparities across the entire chain. So it's complicated, and their myriad potential causes

Steve Goldsmith:

In the article, you referenced the fact that fatality rates per mile traveled for Black Americans cycling is 4.5 times the rate for white Americans. So any thoughts about why and what a city might do to address that particular issue?

Matt Raifman:

Yeah, so I think, as Ernani was referencing, so one of the issues why we're seeing such large disparities between white and Black Americans when it comes to cycling is that Black Americans, in aggregate on average, are traveling fewer miles than white Americans by bike. So that's one issue that is driving the disparities. I think there are a number of things that you can do. I think, intuitively, you could think, "Well, let's place more bike lanes in non-white communities." Right. I mean, I think that's just kind of a knee-jerk reaction.

But it's a little more complicated than that because before you even start talking about implementation strategies and interventions, I think more data needs to be collected upfront on how people of color and white Americans are traveling by bike in cities. We don't really know what routes people are taking or which streets are seeing more traffic, more bike traffic than others, because we don't have good data. We don't collect good data systematically on traffic volume for walking and cycling.

So before you even jump to kind of, "Let's respond," there's this first question, which is, what the data say around traffic volume. Because you could have, say, 10 bike fatalities at one intersection and 10 bike fatalities at another intersection, but you could have way more traffic volume at the second intersection than the first intersection.

And if that's the case, then the actual risk of a traffic fatality is very different at those two intersections. And we're missing that second piece, the traffic volume piece. So I think that as a starting point, getting better data on traffic volume for these vulnerable modes is a really crucial part of understanding what the traffic fatality rates are at different intersections across the city.

Ernani Choma:

Can I just add. So something to what Matt was saying, just reminding the audience that four and a half times, it's a national average, right. Because we draw this data from a national survey in the National Household Travel Survey, and we have a relatively small number of trips that, for example, Black American state cycling.

And so we actually... I think it would be incorrect to assume that this four and a half times is equal across the country. Right. And so I think it's the... we don't even know which cities to have a bigger or smaller problem, let alone on a neighborhood level to think about what the fatality rates per mile are. So I think this number is so big that draws attention to how big the problem is because if it's four and a half times on an average, right, you can imagine how much worse it could be in certain situations.

Steve Goldsmith:

So in a minute, I want to talk a little bit about what cities could do, but one more research question. At the end of the study, you mentioned the fact that future research would benefit from access to street segment activity data. What do you mean by that? We have a number of chief data officers with whom we work. What would you ask them to collect that might enlighten disparities?

Matt Raifman:

Yeah. So this is actually picking up on something I was saying before, which is we just don't have access right now to good data on traffic volume, particularly for walking and cycling, let alone other driving. And when we talk about street segments, we might think of that as a traffic corridor. So like a road, right. Like could be Commonwealth Avenue in the Boston area where we're located, but it also could be a specific block or section of blocks.

So collecting data at that level of granularity allows us to target interventions in a much more effective way and to understand the magnitude of the problem across an urban environment. I think where we've seen cities so far do a really good job when it comes to Vision Zero and understanding traffic fatalities is essentially taking the fatality piece and just plotting all the fatalities on a map.

By doing that, you see, "Okay, there's this street, which has a large number of fatalities, and this street has fewer, so we should target resources." And that's part of the piece. But the other part of the piece is the volume of travel. And that can be useful for adjusting traffic fatalities, but it also could be useful for placing bike lanes, for example.

Like if we know that a lot of bikes are traveling on the street that doesn't have bike lanes right now, then putting a protected bike lane there could avoid traffic fatalities in the future. So that level of data, I think, is one of the missing pieces here. It's hard to... I was thinking about this because I think, how do you actually put the results into action?

And actually, it's somewhat challenging to think of what a city can do because I think, and to some extent, a lot of these data are already collected passively by companies like Google and Apple and Facebook, and other companies that have apps on our phones that are passively collecting data through GPS on where we're traveling and they already have the potential to use that data.

So that's actually what I had in mind when we were calling out this in the paper, is that there exists this treasure trove of data in a sense on traffic volume or at least cell phone volume at the street segment level, and it exists at these private companies. And so, partnerships either between academia and these private companies or between cities and private companies could unlock potential there.

Obviously, there are issues that we could get into around PII and other sort of concerns about privacy. But I think that the data are being passively collected by our phones already. One other thing I would mention is that there are existing partnerships between cities, which I'm sure your listeners are well aware of and generating actually themselves between I think of ways connected cities, that initiative or Strava actually, the bike app has a cities-focused partnership.

So it's possible maybe I could imagine cities to work with those private companies that already have existing partnerships in place to try to access some traffic volume data, particularly for walking and cycling.

Steve Goldsmith:

So we were talking to a number of cities that were interested in taking camera and sensor data from high-risk intersections. So you would know more than just reported accidents. You could actually discern near misses volume of bike and walking traffic risk. Then you could perhaps overlay that with demographics, right.

So you can identify ethnicity and certain wealth factors. So, could that be a place to start where we would encourage our chief data officers and transportation officials to gather more real-time data and then grade it by the type of demographic of the community is a place to start?

Matt Raifman:

Yeah, I think those types of counters could be very valuable, particularly to the point you made about near misses. And that's actually something we think about in the paper a little bit, which is that we don't tackle that at all. We're only looking at traffic fatalities. We're not even looking at injuries, and injuries are another category that is captured in the data, let alone the near misses. So that would actually be a very effective way to look.

I struggle a little bit with the piece about demographic information because you have to infer demographic and socioeconomic status from a visualization, which could be challenging. You could certainly overlay that with information from census data just about that area, that census track, or even that census block. And that would be based on where people live.

So it's a little tricky because you can get good demographic data on where people live, like where their house is, but you don't necessarily know where they're traveling. And that's where potentially surveys can play a really important role because you're asking somebody their travel patterns, but you're also asking how they self-identify by race and ethnicity and their income level. So you're able to connect those data at the individual level.

But certainly, as a first place to start, using those existing data sources and pairing that with census data would be a very nice way to visualize this and start unpacking it a little bit more and would certainly be better than simply pairing the fatality data with census data.

Steve Goldsmith:

Last question Ernani and Matt. So let's assume you're talking to a hundred large city transportation officers and data officers, and you want to turn your research into action. What would you recommend they do?

Ernani Choma:

I am not a policy person. I think I'm more of a modeler. So I think it's difficult for me to say, "Well, what are the specific actions they could take?" Especially because our analysis is not causal in nature, as Matt talked about, is a lot of possible reasons why these disparity exist, right. But I think policy is going to be local. And in order to understand what are the best solutions, I think before we can even recommend policy, I think we need better data, better local data, so we can understand travel patterns so we can understand what's in one place.

Maybe the problem is we don't have bike lanes, but maybe in a different place, the problem is different. So using a national-level result to try to do local policy is probably not very good, right. So I think having this data is very important. And I think, as we're just talking about here, that we only look at fatalities, right, but probably injuries are an even bigger problem. And we have no idea how the disparities in injuries are. Especially if not only you might have different kinds of accidents.

The problem with different differential quality of care is also probably exacerbates is disparities in injuries as well. So I think before I could even recommend the policy side, I think our results, what they do is call attention to how big the problem is on average for the country. And I think it should encourage cities to collect this local data. So we can think about what the policies are to reduce the disparities.

Steve Goldsmith:

Matt, what do you have to add in terms of action items?

Matt Raifman:

I think if we think about where we are right now, like the general way I'm framing this study is, it's pretty upsetting and disappointing that this is where we are, right. That the traffic fatalities of this larger problem that we see these disparities. On the other hand, where we are right now in the United States is actually fairly optimistic when it comes to improving road safety.

And you see, one thing to note is this first kind of once-in-a-lifetime opportunity for cities to apply directly to the federal government for resources to improve road safety, and that's the Safe Streets For All initiative. And actually the NOFO, the Notice of Funding, was just released recently and it's due in September. And it's specifically geared at urban county level and tribal road safety. And actually, as part of that, they call for an equity analysis, which I think would be very welcome, as well as collecting data on exposure.

So I actually think one of the things that they're detailing in the NOFO is to collect better data that we're calling for. So I would encourage cities that are interested in applying to that program. And by the way, I hope every city does because this is an opportunity for $5 billion of resources to improve road safety, particularly for vulnerable users. And when applying, it would be great to also apply for novel ways to collect exposure data, like we're talking about here, so that we're not just using the current data, but we're actually improving data sources to improve implementations in the future.

So I think using the existing funding and the funding mechanisms that are coming online to try to adjust these data problems is a huge part of it, as well as this safe systems approach to road safety. And I think if you think of a city that's getting a lot of attention right now for being successful around Vision Zero, it's Hoboken, New Jersey.

And one of the things they're implementing there is this kind of comprehensive approach to road safety. So thinking about that in the sense of the road users, but also the road themselves, as well as speed, as well as post-crash, all of those elements together in a comprehensive approach, can be really effective it seems at reducing traffic fatalities.

The part that's missing is this disparities focus, I think. And hopefully, research like ours and others that look at disparities can call attention to the fact that there are these race and ethnicity disparities in traffic fatalities. That should also be considered.

Steve Goldsmith:

Appreciate it. Thanks for your good work. We look forward to talking to you again.

Ernani Choma:

Thank you. Thank you.

Matt Raifman:

That was good. Take care.

Betsy Gardner:

If you like this podcast, please visit us at or follow us @DataSmartCities on Twitter. And remember to subscribe at the new Data-Smart City Pod channel on Spotify, Apple Podcasts, or wherever you listen. This podcast was produced by me, Betsy Gardner, and hosted by Professor Steve Goldsmith. We're proud to be the central resource for cities interested in the intersection of government, data, and innovation. Thanks for listening.