Fascinating, but not unexpected. A new analysis by University of Washington masters’ student Jill Schulte reveals that in urban King County, lower-income folks, as well as people of color, are more likely to live close to busy roads. And presumably, they’re more likely to suffer from poor air quality and other environmental harms as a result.
The findings about income and traffic density were exactly what I’d expect. Given a choice, most folks would rather not live along a major thoroughfare. But folks of limited means may not have the money to make a different choice—so I’m not surprised at all to find that low-income folks live in high-traffic places.
Yet what I found most troubling was that, even after controlling for incomes, race all by itself served as an independent predictor of living in dense traffic. That is, if you have two census tracts with the same income profiles, and with the same level of English proficiency, the one with a higher share of people of color will typically have heavier traffic.
Find this article interesting? Support more research like this with a gift!
I emailed with Jill Schulte, the report’s author, and she shared unpublished results showing that, even after holding English proficiency and income constant, a 10 percentage point increase in the percentage of residents of color was associated with a 2.8 point increase in the “weighted road density” score that the author used to measure proximity to heavy traffic.
And that means that neighborhoods with lots of people of color tend to have lots of traffic—even if the neighborhoods themselves are fairly well-off. That, in turn, suggests disparate health impacts on people of color; consider, for example, the new evidence that there’s a link between autism and exposure to traffic-related pollution in early life.
Yet there’s good news here too: Schulte’s analysis found that the relationships among race, income, and traffic density have grown weaker over time. In 2000, lower income folks were substantially more likely to live in high-traffic neighborhoods than in 2010; and the connection between race and traffic density has attenuated as well.
To me, this study highlights the importance of two closely interrelated tasks. First, as a society, we should be searching for ways to reduce the disparities in traffic impacts among different racial and ethnic groups—so you don’t face elevated risks just because you’re poor or a person of color. And second—and perhaps more importantly—we would be wise to find ways to reduce the health impacts of traffic for everyone. After all, it’ll count as little progress if the risks remain the same, but they’re just shifted to a different group of people.
Many thanks to King County for sharing the results of this study!
It’s a very interesting project on the area of environmental justice, a field I think deserves more attention. I do have a couple of thoughts about the results that might be addressed in the full paper. I imagine there is significant correlation between income and race which makes it difficult to completely control for income. Also, if my eyes haven’t failed me, the map is based on census tracts which may lump neighborhoods together that have different race & income profiles (Leschi/Madrona/Central District comes to mind) – it might be worth looking at a finer resolution like census block groups or blocks (something you can get at the new Census RDC).
I also think the idea of sorting is really interesting in this context. If on average minorities have lower income they move to more affordable areas with higher traffic densities. Then, higher-income off people from the same ethnic group may want to live in those neighborhoods if they value being near similar people. There are other political economy issues that may come into effect but I believe that thinking about disparities in environmental quality is important and should be implemented in policy (maybe it is). Great stuff Clark and Jill!
Thanks, Danny! Very astute comments. On the correlation between race and income: Jill did a multiple linear regression, which statisticians use to hold one factor (say, income) constant while looking at the effect of another factor (say, race). I’m not enough of a statistician to understand the potential pitfalls — but this is probably the best that can be done to untangle the two, given the available data.
And on the census tracts — you’re exactly right! Jill found a weaker relation at the census block level than at the tract level — suggesting, perhaps, that wealthier people of color within a census tract tend to live in quieter, lower-traffic parts of the tract.
Absolutely correct about sorting. I figure that’s at least part of the explanation for the link between race and traffic.
I’m actually in grad school for economics (Clark – you helped me get set up on a SR-167 project that is coming along nicely – thanks!), and in the interest of not getting too technical I didn’t go into the concept of multicollinearity. This is a problem when to predictor variabels are highly correlated and can bias regression estimates. There are statistical tests for multicollinearity, but that said economists use both ethnicity and income in regressions for many different outcomes.
One quick and easy non-technical test is to plot census tracts (or blocks) on both income and race. If there is a clear trend (probably a downward sloping line if income is on the y-axis and race is on the x-axis) it may be tough to disentangle the two effects. Alternatively, if the points are more diffuse (a cloud as opposed to a line) then there may be sufficient variation in race after controlling for income to isolate the causal effect of race.
If Jill is reading this and wants to get in touch feel free to contact me.
Glad to hear that the project is coming along, Danny! I’d love to hear about any results you’re getting.
And thanks for the offer — I’ve passed it on to Jill.
One complication is that, at the tract level, household income is based on sample data, rather than a complete count. So there’s likely some significant error in the income numbers — both sampling error and respondents’ inaccuracies — that may make the statistical relationships even harder to tease out. (E.g., the _actual_ distribution could be colinear, but the error term could obscure the colinearity.
Thanks for your feedback, Danny. I agree that collinearity is a concern, but only between the English proficiency and race/ethnicity variables, which are highly correlated (~0.86) for obvious reasons. The correlation between race/ethnicity and median household income is considerably weaker (~0.52). From my research, a correlation coefficient this low is not generally considered to be in violation of the assumptions of multiple linear regression. In addition, the variance inflation factor (VIF) of the model with only race/ethnicity and median household income is 1.38. VIF is an estimate of the collinearity between multiple variables; a VIF of 1 indicates no collinearity, while a VIF of 5 is a fairly well-accepted minimum threshold for collinearity. This is far from a perfect model, but I don’t believe that the correlation between the two variables invalidates the findings that Clark so eloquently describes above.
Glad to hear you dotted all your i’s and crossed your t’s. It’s tough to comment on summaries of research when you are interested in some of the gritty details because you need to leave out a lot in these types of settings. It is quite interesting research and I think equal access to environmental quality will be a hot topic.
Clark: We’re still collecting data from WSDOT (we needed to submit a Public Disclosure Request) and TRAC, but we should probably get some results soon. I’ll keep you in the loop.
Following up on the findings, but not having the complete work by Jill Schulte to read (salute to Danny above, making good points), I can’t off hand grasp why certain ethnic groups seemingly want to live closer to expressway on and off ramps! Something cultural? Maybe because on average people in some groups need/want more mobility and have embraced the dominant automobile mode, accepting, or ignorant of, the trade off with increased health risk from living close to expressways? Clark, what’s your theory?
Because of the massive consumer embrace and market share of the car driving mode, it’s very important for our society to take action to make cars more environmentally benign — less pollution, fewer accidents, more energy efficient, less noise, smaller footprint, higher vehicle occupancy. The Obama administration is certainly working in this direction, and based on Chapter 2 of Jay Inslee’s book, Apollo’s Fire, the new Washington State administration will be working along these lines also.
The process of technology development and applications are making all of these improved attributes of cars possible, and there is progress month by month. http://www.aboutcates.org Many people are working on making cars better for the environment around the world, in the USA and Canada, and in our Cascadia region.
My working hypothesis is that the improvements in the environmental footprint of cars are good for the region no matter how the attempts to modify land use pattern and density work out, and no matter how much rail transit gets built and used. The computer-generated forecast from the PSRC regional planning agency is that cars will still have a 95% share of motorized travel in 2040 even after large increases in transit infrastructure and with a policy focus on increasing densification near transit stations.
The people of 2040 who live near expressways will not be so damaged when the roar of internal combustion engines is replaced by the buzz of electric motors.
Good points as always, John. Reducing the impact of cars — health, noise, etc. — is an important way to help alleviate these sorts of disparities.
My sense is that others on the comment string have some good ideas about the links between race and traffic density. Historic patterns of settlement (driven in large part by red-lining) came to my mind, as did self-sorting. Another big factor may be the difference between wealth and income. Income gets measured by the census, but it’s temporary; wealth is often a better indicator of actual financial circumstances over the long haul. Finally, data aggregation at the tract level could be obscuring what’s happening in individual households.
The analysis is interesting but I’m not sure you could apply this in Boise, Idaho. Some of our busiest residential streets also happen to have some of the wealthiest people and the most expensive real estate.
There are arterials,that are also very congested and folks of low income, mostly white, live along them. In a community dominated by white people, not people of color or other immigrants it is the same throughout the city.
While I don’t have firm data to back this up in my town I think it’s fair to say that it is coincidental that in Seattle it is race that factored in. People of a specific race, enthnicity or culture often congregate in similar areas because there is comfort in numbers. If you don’t speak the “preferred” language, if you don’t eat the same foods, if your religion is different than the norm, if your customs are considered “unusual” you will tend to congregate with others who are similar to you. And you will choose the place that fits your economics. And living by a road or a busy intersection is typically one of the cheapest places. Unfortunately, living in areas of crime (a chicken and eggs scenario indeed) makes them less expensive to live in. I think the researchers need to look at their independent variables more closely to determine what the real controlling factors are.
Interesting perspective on Boise. I suppose Seattle has its correlates: the most expensive real estate is downtown, where there’s lots of traffic. And in downtown, you tend to find either very high-end condos or rent-subsidized apartments — both ends of the income scale.
For sure, more work needs to be done in this area — I think that Jill’s work is just one piece of a much larger body of research on housing and transportation choices.
Dont forget that some of the results are explained by the fact that the Sprawl Lobby prefers to target minority communities for highway/carburban infrastructure that will speed wealthy suburbanites through the “felony flats” sectors, like the proposed Salem River Crossing, the Mini-Me to the Columbia River Crossing, which reflects the prior iteration of the same idea up in Portland.
Interesting! That’s certainly true in many parts of the NW & the US: transportation departments would rather run a new road through cheaper real estate than through expensive real estate — which by default means that they’re targeting lower-income neighborhoods.
I’m African-American,and obviously can’t speak for “all people of color”, but will suggest that many factors go into deciding where to live (and it’s not, “gee, I’d like to live near a freeway on-ramp”!). Cost is certainly a biggie. The desire to live in a neighborhood where there are other people of color is another. If one has school-age children one might choose a neighborhood for its schools. Housing discrimination still exists which would, of course, impact neighborhood “choice”. People who don’t own cars tend to select housing near a useful transit line. Having come to the PNW from NYC where average lay people have been savvy about the relationship between Housing and Transportation when determining cost-of-living for decades, access to a public transportation hub was a huge factor in determining where I wanted to rent initially and later buy a home. Hope this was helpful!
Definitely helpful! Thanks!!!