What Drives Obesity? An Economist Takedown of The Economist

(iStockphoto)

Is higher obesity due to the rise in driving? Perhaps. It’s an intriguing hypothesis. But our friends at The Economist should know better than to report nonsensical correlations. Here’s the evidence they cite (drawn from this entirely unconvincing research paper published in Transport Policy):

Looks impressive, right?  (Well, apart from putting the explanatory variable on the vertical axis.) But before concluding that there’s anything here, let’s try a different variable, instead—my age:

Unlike the authors of the original paper, I didn’t even need to fiddle with the lag structure to get such a good fit, nor test alternative definitions of the variable. In fact, my variable fits even better than vehicle miles traveled.

Okay, I’m not arguing that my aging is causing higher obesity. Rather, when you see a variable that follows a simple trend, almost any other trending variable will fit it: miles driven, my age, the Canadian population, total deaths, food prices, cumulative rainfall, whatever.

Sure, The Economist offered the usual caveat that “correlation does not equal causation.” But this is so completely unconvincing as to warrant a different warning: “Not persuasive enough that you should bother reading this article.” I’m not saying the relationship doesn’t exist, simply that it makes more sense to highlight more persuasive research on this question.

P.S.: How cool is Stata’s “Economist” scheme? It lets stat-nerds like me replicate magazine-quality art.

Leave A Comment

Comments are moderated and generally will be posted if they are on-topic and not abusive.

 

COMMENTS: 45


  1. Gary says:

    I love spurious correlations. Back in the 80′s I won myself a scholarship worth GBP 150 per year for my 3 years of Uni (a lot of money to an 80′s student), partly on the back of an essay where I employed the correlation between sardine consumption and cancer to illustrate the non-causative correlation leading to some dodgy conclusions. I love sardines too, by the way. And I don’t have cancer (yet?).
    The thesis at that time was that sardine consumption and cancer had a common causal factor, namely wealth. Just a thought, but I wonder if obesity and car mile driven might also have a somewhat positive correlation with wealth. I see a particularly high r-squared value for that one…

    Thumb up 6 Thumb down 2
    • Justin Bassett says:

      So you’re saying that when Biggie Smalls talks about how he “used to have sardines for dinner,” he’s looking nostalgically on the relative comforts of his childhood?

      Well-loved. Like or Dislike: Thumb up 22 Thumb down 2
  2. Clancy says:

    They might be totally unrelated, but causality could also run the other way. As people get fatter, walking requires more effort, so they drive more. Eventually they stop considering walking as an option at all and they have no problem moving to unwalkable suburbs and driving increases again.

    The important question is: How can we design a conclusive study?

    Thumb up 4 Thumb down 3
    • Tom says:

      Why the bizarre qualification of ‘unwalkable’ for a suburb?
      You’ll find people do walk here as much or more, as you do not need to worry about being clipped by a car or bothered by corners crowed by kids hanging out.

      When you do walk, you walk further and its more enjoyable.
      I sometimes wonder if some commentors have ever been to a suburb.

      Thumb up 2 Thumb down 3
      • Enter your name says:

        Depends on the suburb.

        Some of them have zero sidewalks but ample fences, so walking to the store means walking in the gutter, just inches away from cars driving 25 to 40 mph.

        In one I know, the legal places to cross most of the major streets appear just every half mile. So if you’re in the middle of that half-mile stretch of road, to get to a store just across the street legally requires you to walk a quarter of a mile north, cross the street at the light, and a quarter mile south. You literally have to spend 20 minutes and walk one mile to buy a loaf of bread at the store almost directly across the street from your house. Alternatively, you can legally back your car out of the driveway, change lanes, and turn left into the parking lot: 100 feet of driving, one minute.

        But more commonly, the stores simply aren’t within a reasonable walking distance. For example, my aunt lives in a suburban area. It would take 15 minutes of walking just to get out of her housing division. From there, the nearest grocery store is a mile and a half away, on the other side of a four-lane, zero-sidewalk major surface street with a posted speed limit of 40 mph. Most people refuse to spend an hour and a half on traveling to the grocery store.

        Oh, and it was 96 degrees F and 60% relative humidity at her house yesterday afternoon. I suspect that no sane person with any alternatives would have spent an hour and a half yesterday on that unsafe, ugly, and unpleasant walk. If they did, we’d see more people being treated for heatstroke.

        Well-loved. Like or Dislike: Thumb up 15 Thumb down 0
      • thehumble1 says:

        Sometimes I wonder if some commentators have ever been out of their suburb.

        sure you can go for a nice, leisurely walk. You can walk your dog around your nice sidewalked neighborhood and you can even walk to a friend’s house (as long as they are in the same housing development). But suburbs for utility walking? I’m sure you can find exceptions, but the vast majority are highly vehicle-centric and pedestrian-averse.

        Thumb up 1 Thumb down 0
  3. Manfred Esquivel says:

    I am not going to say that the analisys of “The Economist” is completely right, but I think you’re missing the fact that their dependant variable is not necesarily trended. Note that is miles per licenced driver, which takes off much of the trend that is in the variable “total miles driven”.

    Thumb up 3 Thumb down 2
    • Jamie_NYC says:

      No, you are missing a fact: if population of Canada has been increasing during the last 20 years, that time series is going to be correlated with obesity percentage time series. That is all Justin was saying: two trending variables are bound to be correlated.

      Thumb up 5 Thumb down 2
  4. KNe says:

    There was a similar article in the Economist a while back. I believe it was on the correlation between posture and leadership abilities. I recall questioning the lack of skepticism in that report, too.

    Thumb up 1 Thumb down 0
  5. Jacob says:

    There is an excellent explanation of why, and how, we get fat here: http://www.farnamstreetblog.com/why-we-get-fat

    It gets into the weeds some on LPL and insulin but it’s a great, an scientific exploration.

    Thumb up 3 Thumb down 0
    • Juba says:

      That’s a good post, Jacob. Doesn’t say much about driving, does it?

      I agree with the post: The Economist just wasted a bunch of time by doing this study. Not only did they not have enough data to determine if this correlation was of any value, but the correlation is spurious at best. Clearly biology is a first-order factor in obesity, not driving.

      Thumb up 2 Thumb down 0
    • James says:

      “There is an excellent explanation of why, and how, we get fat here:”

      No, there isn’t. There’s a blank screen and a busy cursor.

      But I suspect it’s more of the same old excuses. What this thread misses is that there likely is in fact a causal relationship between obesity and amount of driving, which is lack of exercise.

      Correlation may not prove causation, but it does give you a good place to start looking. Or would you suggest that it’s just coincidence that e.g. smokers have far higher rates of lung cancer &c?

      Thumb up 0 Thumb down 4
  6. Amy Alkon says:

    Per Jacob’s link to the blog item on why we get fat, Gary Taubes lays out a mountain of evidence that it’s carbohydrates — sugar, flour, starchy vegetables like potatoes, apple juice — that cause the insulin secretion that puts on fat.

    I’m a newspaper columnist and author and I spend my entire day seated and barely exercise, yet thanks to cutting out carbs, I am effortlessly thin. This morning, I had three strips of bacon and a cheese omelet cooked in butter. That, according to the evidence, is healthy diet food. Later in the day, I’ll have a steak with heavily buttered green beans. More diet food.

    What I don’t eat: bread, pasta, juice, etc.

    Oh, and I have the health stats (blood pressure, etc) of an elite athlete.

    Hot debate. What do you think? Thumb up 13 Thumb down 10
    • Bart Johnson says:

      Hidden due to low comment rating. Click here to see.

      Disliked! Like or Dislike: Thumb up 11 Thumb down 17
    • Joshua Northey says:

      Carbs have lots of calories. You are in fact both right. Cutting out carbs will reduce you weight because it is hard to make up those calories with non-carb food.

      At the same time exercising more, or eating less, or cutting out fat will also cause you to lose weight because they all effect your daily calorie balance.

      My weight shifts a lot depending on if I play hockey 1 2 or 3 days a week. This is because my diet is fairly constant.

      Thumb up 3 Thumb down 2
      • Bart Johnson says:

        Carbs have 4 calories per gram, as do proteins. Fats contain 9 calories per gram.

        So, it’s easy to “make up those calories” – eat half as much fatty food as the carbs you cut out, or an equal mass of proteins.

        Thumb up 6 Thumb down 10
      • Joshua Northey says:

        But people don’t eat as much protein or fat as they do carbs.

        The actual foods are you know, different. People don’t replace a meal with 20 grams of rice carbs and 20 grams of meat carbs with one that has 40 grams of meat carbs. Both because it is too expensive and because that is not the way meals are typically structured.

        Foods have different calorie densities.

        Thumb up 3 Thumb down 0
      • Al Cooper says:

        Mr. Johnson, if you will take the time to read Mr. Taubes’ book, you will discover that calories in do not equal calories out. Our bodies don’t process carbohydrates in quite the same way they do proteins or fats. Mr. Taubes merely took a look at the scientific data on diet and exercise for the past 10 years, and WWGF is a compilation of those factors into an easily accessible, scientific summary of the current state of what we know about why people gain or lose weight. Nothing he says is “double speak” or nonsense, but unfortunately your comments are.

        Well-loved. Like or Dislike: Thumb up 11 Thumb down 3
      • Bart Johnson says:

        I didn’t say that calories in = calories out. I said that weight gained is calories in minus calories out. That is an immutable, ironclad manifestation of the laws of thermodynamics.

        People gain weight because they are consuming more calories than they expend, and people lose weight when they expend more than they take in.

        That is the one and only explanation why people get fat. It is the only possible one.

        Thumb up 3 Thumb down 6
      • Steve says:

        Although calories in – calories out will provide an approximation of the effect of diet on weight, it’s incomplete partly because it neglects the thermic effect of food.

        One calorie, as used in nutrition, is simply the amount of energy required to raise the temperature of a kilogram of water by 1 degree C. For years it was assumed that a calorie is a calorie – but the body is not a beaker of water in a sealed chamber. Metabolizing a protein calorie requires significantly more energy expenditure by the body above resting metabolic rate than does processing a fat calorie. So eating “equivalent” caloric amounts of fat versus carbohydrates versus protein will result in far different energy expenditures, completely irrespective of activity. Multiply this effect by the thousands of meals eaten in a year and one calorie is absolutely not equivalent to another.

        Thumb up 4 Thumb down 0
      • thehumble1 says:

        @Bart. tell that to my poop which breaks the thermodynamics equation because I’m not a closed system. I’m also not a simple system, so how much energy I get from each kCal can be achieved by me using it up during waking activity, basal muscle use, or fat deposit.

        People who are building muscle consume more calories than they expend. They have to create that muscle from something, you don’t add mass directly from incorporating the wave energy of the bow flex.

        Thumb up 1 Thumb down 0
    • Cliff says:

      Unfortunately, high levels of protein seem to substantially increase cancer risk, among other problems. Too bad, since high levels of protein consumption are also linked to better mood.

      When you list sugar, flour, potatoes, and apple juice, those are all sky-high in the glycemic index. Of course simple carbs are probably the worst thing you can eat: nutritionally low-value, high calorie, inducing an insulin response. But complex carbs are fine and have been a staple in most cultures for thousands of years.

      Thumb up 3 Thumb down 1
      • thehumble1 says:

        I thought there were good indications that protein isn’t the cause, just a correlation. I think that most of it goes away when you control for processed proteins, so having a steak might not create nearly as much harm as having a hot dog. And things that are staples for thousands of years means that they are almost brand new to the body and we have nearly no adaptation to it. Things we’ve been eating for millions of years are much more likely to be digestible, “healthy” and easily metabolized by the body. But isn’t that just what Michael Pollen said about food? That the real problem is that we aren’t eating what we had for thousands of years. Eat food, not too much, mostly vegetables.

        Maybe I’m just making this up.

        Thumb up 0 Thumb down 0
  7. Sunfell says:

    Driving doesn’t make you fat. But going through the drive-thru, and eating while driving? That’ll make you fat, big time, especially if it becomes a regular habit. Face it- foods that you can eat with one hand are seldom of the type that lend themselves to healthy living. I have yet to see someone handing over a container of carrots and celery sticks to someone at a drive-up window.

    Thumb up 2 Thumb down 0
  8. Jeff Yablon says:

    Hmm . . . you mean STATISTICS LIE? Who’da thunk?

    Oh yeah … me:

    http://answerguy.com/2011/06/03/statistics-lie-are-you-listening-hubspot-and-shareaholic/

    Thumb up 0 Thumb down 4
  9. Researcher says:

    I also think it is important to consider indirect causal links between policies and health outcomes (such as politics and culture), not just the direct ones.

    The authors don’t get into this in their study, but it is important consider the factors of social pressures, advertising, funding priorities, redistribution of wealth, etc., and their impact on obesity, vis a vis driving’s direct relationship to those factors.

    For example, as people drive more, they may be less likely to support funding for mass transit even if the need for mass transit is dramatically rising.

    The authors of the study point out its limitations very clearly, and their point is broader than one correlation. I would suggest reading the original article more closely before tossing it aside just because The Economist didn’t do a good summary.

    Thumb up 2 Thumb down 1
  10. Amy Alkon says:

    Actually, Taubes lays out evidence as well that a calorie is not a calorie. Doctor Michael Eades also lays this out on his blog, proteinpower.com/drmike — about how the body uses carb calories very efficiently but blows off non-carb ones (he uses the term “calorie wasting”).

    Also, eating animal fat and protein is satiating. On my morning omelet and bacon, I can go till about 3 pm without eating and not feel uncomfortable.

    Taubes also wrote a piece, supported by evidence, for New York Magazine, which shows exercise does not make you thinner; it makes you hungrier. People who don’t know good scientific sources, and are simply posting scientific hearsay, always amaze me. Because you read it on CNN.com doesn’t make it so.

    I use science in my work, and have learned to read studies, and trust very few people to interpret them. There are maybe five people in the world I trust. Dr. Michael Eades and Gary Taubes are two of them. In fact, Taubes is the single most skeptical human being I have ever encountered, and the most nervous about seeing that he is putting out evidence-based science.

    Thumb up 6 Thumb down 4
    • James says:

      “Taubes also wrote a piece… which shows exercise does not make you thinner; it makes you hungrier.”

      Ignoring much contrary evidence, including decades of experience from probably millions of individuals who’ve gone through military boot camp, or the many thousands who’ve ever followed a serious body-building program, or got seriously into bike touring, or…

      Of course we have to adjust our definition of “thinner” a bit, so as to account for increased muscle mass/decreased body fat percentage, but it does work.

      Thumb up 4 Thumb down 0
      • thehumble1 says:

        I believe the conclusions are that exercise makes us hungrier initially but actually decreases overall food consumption by making us more generally sated. This is also why you can change someone’s weight just by putting them on antidepressants, which may or may not lead to changes in activity but definitely does decrease overall caloric consumption. Heavy workouts burn almost no calories compared to the additional calories needed to complete those workouts, but people who work out a lot tend to be thinner. With straight math, you find that they don’t burn enough calories during the workouts to make up for the loss of weight (10 lbs = 35,000 calories = 282 miles @124 kCal/mile). If you jogged a 5k, 3 times per week it would take you about 8 months to lose 10lbs and that’s if you ate NO additional food after all of those miles jogged. What we see is that people drop 10 lbs in 2-3 weeks after starting something like 5k 3x/week. Somebody please correct my math or logic though, because to me this seems to highlight an issue with exercise: it can’t be the calories we burn on the treadmill that is causing the change. My understanding is that it changes our metabolism so that we actually eat less, are more intensely hungry for shorter periods of time, and are more satisfied and less anxious in general which must help in some other ways.

        Thumb up 0 Thumb down 0
    • Cliff says:

      Amy,

      Aren’t you a relationship advice columnist? It’s not like you edit a scientific journal. There is copious evidence that exercise does lead to weight loss.

      Well-loved. Like or Dislike: Thumb up 8 Thumb down 2
      • Enter your name says:

        There is also plenty of research showing that if you burn 200 calories on the treadmill, and then “reward” yourself with 400 calories of treats because you were such a good boy to get on the treadmill, then you will gain weight.

        Exercise causes weight loss only if you control for diet. If you increase both exercise and calorie intake, then weight loss is not in your future.

        Thumb up 2 Thumb down 2
  11. AaronS says:

    Well, you could just as convincingly argue that obesity is linked to the rise in computer usage. Oh, and inversely related to the use of mule-drawn plows…and black and white televisions…and non-homogenized milk.

    Thumb up 2 Thumb down 0
    • thehumble1 says:

      I’d really like to think that the use of multi-colored televisions has been the culprit for weight gain, but I’m more likely to think that it’s caused by trending increases in the minimum salary (in 2011 US$) of NFL draftees which shows good regression as well.

      Thumb up 0 Thumb down 0
  12. Sigivald says:

    Or classics along these lines – Global warming vs. Pirate population.

    Thumb up 1 Thumb down 0
  13. mjay says:

    Clearly driving causes ageing.

    Can I have my economics/sociobiology/evolutionary psychology PhD now?

    Well-loved. Like or Dislike: Thumb up 7 Thumb down 0
    • Chris S. says:

      Not yet.

      Flesh it out to about 45 pages, add some cool charts, and then get your advisor to sign off. That should do it.

      Thumb up 2 Thumb down 0
  14. Daniel says:

    on the ps note: I did a presentation completely with the economist scheme, and afterwards one of the audience members asked if I was doing soft advertising for the economist by using all their graphs. It took me a bit to convince him that I was only using the color scheme and not graphs actually produced by the magazine.

    Thumb up 0 Thumb down 0
  15. Joe says:

    Stata is indeed fantastic, though the graphics take some serious learning to master.

    Thumb up 0 Thumb down 0
  16. anonymous says:

    It is far too easy for a trained applied econometrician to point out a spurious correlation reported in the media — even in a highly regarded publication such as the Economist. If we are going to criticize others’ research we ought to at least provide some recommendations for improvements. How would one go about better estimating this potential relationship, rather than just showing its spurious nature?

    Thumb up 0 Thumb down 0
  17. Mike H says:

    Clearly, as people get fatter, it causes you to age. You should encourage everyone to diet and exercise, so you can live forever.

    Thumb up 1 Thumb down 0
  18. Joe says:

    In order to ascertain if Mr. Wolfers’ existence is the root cause of obesity in the US, I propose that he perform an experiment . . .

    Thumb up 0 Thumb down 0
  19. Confounding says:

    This is sadly not the first time this has happened in The Economist. I detailed one of their other mishaps here: http://confounding.net/2009/10/21/next-on-when-graphs-attack/

    It’s gotten to the point that I start automatically rethinking their graphs the moment I see them.

    Thumb up 0 Thumb down 0
  20. Michael says:

    It is almost certainly different in different regions. Englishmen who go to live in Scotland usually put on weight because there is a much higher availability of fatty food and it’s cold, but not so cold that proper clothing is a survival matter. Put the two together and you can not feel so cold by eating the readily available fats.

    Then you spend years looking for a way to lose it. But I’m getting there.

    Thumb up 0 Thumb down 0
  21. beef says:

    The miles driven stat is a proxy for a basket of life style choices. This stats implicitly represents many degrees of freedom in the analysis. It makes a suburban lifestyle (drive everywhere) endogenous to the argument. It kind of flippant to dismiss all value in this using ‘correlation != causality’.
    The Economist is hardly known for lack of editorial oversight, you might not like some of their conclusions and they are often proven wrong in the long run, but they rarely would publish something as simplistic as you posit.

    Thumb up 0 Thumb down 0
  22. aitch555 says:

    Huff on How to lie with Statistics showed STDs and consumption of coke were highly correlated for Vietnam…additional confounding variable was presence of US troops!

    Thumb up 0 Thumb down 0