Maps Show Striking Link Between Car Commuting and Obesity

Check out these two maps, the first showing obesity rates (by county) in the United States and the second showing the percentage of commuters who travel by car (via Planetizen).

Obesity rates are highest in Appalachia and the Southeast United States. Image: Planetizen
A map showing the percentage of car commuters shows a strikingly similar pattern.

Researchers Anne Price and Ariel Godwin at Planetizen caution readers not to conflate correlation and causation. However, when comparing other economic and demographic characteristics (unemployment, educational attainment, income), no other maps displayed such striking similarities.

Furthermore, when the research team created a scatterplot comparing obesity rates in U.S. counties with commuting patterns, a “strong relationship” emerged.

This scatterplot shows the link between obesity rates in U.S. counties and rates of commuting by car. Researchers found the correlation to be "strong."

Again, Price and Godwin were cautious about drawing a direct causal relationship:

Considering that the percentage of active commuters in the U.S. is quite small, it is unlikely that walking and biking make any significant contribution to reducing the obesity rate in particular counties. More likely, counties with the highest percentage of walkers and cyclists also share other common characteristics that are driving this trend. Perhaps lower rates are driven by a cumulative effect of a more affluent and educated population. It may also be that counties with higher rates of active commuting have policies and cultures that have led to higher rates of physical activity overall.

Sounds like more research is needed — in addition to more transportation choices in the southeast and Appalachia.

It would also be interesting to examine whether the availability of alternatives to single occupancy vehicle travel contributes to some of the better economic indicators in some of the nation’s less obese, wealthier areas.

  • Anonymous
  • Kurtkilted

    Looks like Illinois and Texas aren’t reported correctly.

  • jUNKIEd

    How can a state border cause such a stark difference in obesity rates?  Look at Illinois or Texas versus its neighbors.  Could this have something to do with data collection or reporting?  Look at Kansas vs. Colorado – I have a hard time believing the counties just west of that border are that much different from the counties on the east side of that border.

  • It’s also interesting that you can see the Appalachian Mountains in the obesity trend.  It must be all the calories burned climbing those hills.

  • It would be interesting to see the scatterplot with Alaska taken out, since some of the means of commuting there are probably not reported as “driving” – flying, snowmobile, etc.

  • ilike bikes1035

    It also looks like there’s some kind of break down in Alaska. High obesity rates, low commuting by car rates, according to the maps at least. Just get everyone in those high-driving, high-obesity areas a folding bike – some places the urban (or suburban) sprawl is pretty intense and it’s hard not to drive. But there’s always the option when you’ve got a bike in the back of your car, to ride if it’s a short trip or nice weather, right?

  • Ed, actually the scatterplot removed outliers, many of the in Alaska.

  • There is a great deal of evidence that links time in a car to obesity, including a study conducted in 2004 in Atlanta that took into account sociodemographics and land use. Land use mix (residential neighborhood within 1 kilometer of commercial area) turned out the most important factor in predicting reduced likelihood of obesity (perhaps because people are more likely to walk and bike if there are destinations within walking and biking distance of their home.) Second was time spent in a car. (Each additional hour spent in a car per day created a 6% increase in the likelihood of obesity.) Third was how many kilometers of walking per day. (Each additional kilometer walked per day was associated in a 4.8% reduction in likelihood of obesity.)

    Given this information, we may have to face that jobs that require people to spend most of their day sedentary behind a wheel (truckers, bus drivers, etc.) are in fact health hazards. (Probably as great a health hazard as a smoky bar is for wait-staff.)

    Also, we tend to focus on how people commute, but given that only 15% of all trips made in the US are commute trips, it might be a good idea to consider the mode of travel for the other 85% of trips people make.

    From the Bureau of Transportation Statistics National Household Travel Survey:
    45 percent of daily trips are taken for shopping and errands27 percent of daily trips are social and recreational, such as visiting a friend15 percent of daily trips are taken for commuting
    For more info on the Atlanta study:
    http://www.ncbi.nlm.nih.gov/pubmed/15261894

  • Anonymous

    The fact that the first green category for the percentage of commuters who drive covers 9% to 86%, and then 4 categories cover the remaining 14%, shows how just ridiculously car-centric our nation is. On a linear scale (each color representing a 20% chunk), there would be almost no green except a handful of cities like NYC.

  • There is a great deal of evidence that links time in a car to obesity, including a study conducted in 2004 in Atlanta that took into account sociodemographics and land use. Land use mix (residential neighborhood within 1 kilometer of commercial area) turned out the most important factor in predicting reduced likelihood of obesity (perhaps because people are more likely to walk and bike if there are destinations within walking and biking distance of their home.) Second strongest predictor of obesity was time spent in a car. (Each additional hour spent in a car per day created a 6% increase in the likelihood of obesity.) Third was how many kilometers of walking per day. (Each additional kilometer walked per day was associated in a 4.8% reduction in likelihood of obesity.)

    Given this information, we may have to face that jobs that require people to spend most of their day sedentary behind a wheel (truckers, bus drivers, etc.) are in fact health hazards. (Probably as great a health hazard as a smoky bar is for wait-staff.)

    Also, we tend to focus on how people commute, but given that only 15% of all trips made in the US are commute trips, it might be a good idea to consider the mode of travel for the other 85% of trips people make.

    From the Bureau of Transportation Statistics National Household Travel Survey:
    45 percent of daily trips are taken for shopping and errands27 percent of daily trips are social and recreational, such as visiting a friend15 percent of daily trips are taken for commuting
    For more info on the Atlanta study:
    http://www.ncbi.nlm.nih.gov/pubmed/15261894

  • Anonymous

    Ed and Angie, another issue with Alaska is that this map was done using counties.  Which means in Alaska the data is divided up into “Boroughs”, of which the only one which is centered around a city is Anchorage.

  • MFS

    As many people note, there are some serious state effects going on here that can’t be sorted out by a scatter plot.  And because of that the VA, GA, SC & NC sides of Appalachia show low obesity (as opposed to KY, TN and WV).  If you really wanted to (and I’m sure there are epidemological studies that have done this), you would do a multivariate regression at the county or place level that included fixed state effects, population density, average VMT, income/poverty rate, and demographics in addition to car commuter rate.

  • MFS

    As many people note, there are some serious state effects going on here that can’t be sorted out by a scatter plot.  And because of that the VA, GA, SC & NC sides of Appalachia show low obesity (as opposed to KY, TN and WV).  If you really wanted to (and I’m sure there are epidemological studies that have done this), you would do a multivariate regression at the county or place level that included fixed state effects, population density, average VMT, income/poverty rate, and demographics in addition to car commuter rate.

  • Anonymous

    Yea, I have a real hard time seeing any correlation in the data here. Far West Kansas and Far East Colorado look just about the same from the ground and drive at rates expected of a rural area, yet they have very different obesity rates.

    I have never been to Southern Illinois, but I am failing to see how its obesity rate would be much higher than neighboring states, even though the respective counties drive at the rates expected of a rural area.

    What is it about just living in Colorado, and to a lesser extent Illinois, that would make the obesity rate difference so startling? 

  • MFS

    I read the original planetizen article and it covers the issue in a much, much more nuanced way than the Streetsblog piece here.  It shows that income/poverty and education are related as well.

    The beginning of their conclusion:
    “The relationship between sedentary travel and health outcomes can be
    misleading when additional contributing factors are not taken into
    account. While it is not our intent to claim a direct causal link
    between transportation modes and obesity rates, it is hard to deny the
    existence of some geographic patterns.”

    Also, the map of diabetes strongly indicates some strong state effects in measurement of obesity.

  • Jkspinning

    What I want to know is how is Texas measuring obesity? That line following the northeast border of Texas is the most striking thing I see.

  • Martisco

    Honestly, I do not see any “striking” correlation between those maps.  Look at Texas.  Look at Kansas and Michigan’s upper peninsula.  I don’t see any correlation there.  Whoever wrote this story seems fixated on Appalachia and the Deep South and is failing to notice how the rest of the maps don’t exactly match up.

  • Anonymous

    Martisco —

    That’s misleading. You can’t pinpoint two or three areas where the correlation seems grey or foggy and then ignore the rest of the country, where the correlation is generally very strong. That is biased interpretation. The fact is, the map does show a generally clear trend: *most* parts of the country reflect a positive correlation between high rates of driving and high obesity rates. Two or three exceptions to the rule do not negate the general rule. This is critically important in interpreting the significance of statistics.

  • Nicéphore Jünge

    And if these maps are to be trusted, the people of the Eastern half of the Appalachian core territory (from say around Scranton, PA, to say around Rome, GA) do commute a lot, and are not very prone to obesity nevertheless…

  • KB

    Food deserts

ALSO ON STREETSBLOG

Our Waistlines Are Expanding In Sync With Our Car-Dependence

|
States with the highest adult obesity rates also tend to be where the fewest people bike or walk to work. Image: CDC Two reports released last week underscored the increasing severity of America’s obesity epidemic. And the eye-opening findings add to the mounting evidence that stopping the spread of obesity and its attendant health risks […]

New Orleans, Honolulu Rocket Up the Bike Commute Charts

|
Cycling held its own against other modes of transportation in 2009 despite decreasing gas prices, according to Census numbers crunched by the League of American Bicyclists. The league released its annual Bike Commuter Rates report for the 70 largest U.S. cities yesterday. Established leaders in the cycling movement hold many of the top spaces, including […]