Junk Science that supports Health at Every Size is still Junk Science

A few months back, over at Big Fat Blog there was a report on this study.  The study was basically examining four ‘healthy’ habits (moderate drinking, not smoking, ‘exercising’ and eating 5 plus fruit and vegetables) and the BMI, and then doing an analysis of risk of death.  DeeLeigh from BFB summarised its findings like this:

Two things really jump out at me. First, the more healthy habits we have, the more our life expectancy matches the life expectancy of thin people with the same habits. When we’ve got all four, the gap is pretty much closed. Second, it’s only the fat people with no healthy habits who have a dramatically reduced life expectancy in comparison to thinner people.

This is a strong confirmation of what HAES advocates have been saying for years

It’s bullshit.  Of course it’s bullshit.  This study has exactly the same errors as all the other studies which people on fat acceptance blogs have picked apart and chanted “Correlation does not prove causation” at.  The most glaring of which is (as always) that it does not control for class. You cannot say anything meaningful about people’s bodies or lives if you don’t take into account the way resources are distributed in society.

I’ve always felt slightly uncomfortable about the way scientific studies are used for the cause of fat acceptance. I’ve always felt it conceded too much ground – by spending lots of energy arguing that fat isn’t necessarily unhealthy, we’re conceding the conclusion that if fat was healthy fat hatred would be justified.   But I can see that given the amount of junk-fat-hating masquerading as science there is out there debunking is useful work.  But if debunking is going to work as a persuasive factor, or (more importantly in my opinion) a way of figuring out how the world works, then people engaging with scientific studies have to be absolutely disciplined and committed to engaging with the literature as it is.  Just reposting one article that agrees with your pre-conceived views without engaging with the critical thinking that you would if it disagreed undermines that project.


That blog post was the first thing that came to my mind when I read the post and skimmed the article. But then I read the article in more detail and I became outraged on a whole new level.  Because in the article itself they provide how they’d defined exercise:

Level of physical activity was determined according to the frequency of participation in leisure-time physical activities within the previous month.

There is no justification for this definition in the article.*

I actually lose it at this point and can’t form any coherent thoughts.  You can’t measure a subset or something a pretend you’ve measured the whole thing.  You can’t claim to do one thing, when you’re actually doing something else.  You can’t just wave away the word ‘leisure-time’ as if it doesn’t exist.  Except apparently you can – in a peer-reviewed journal.

I want to know how wide-spread this is? How often in peer-reviewed articles, advice given to doctors, information passed on to us all have they told us that ‘exercise’ has a particular effect when they’ve measured ‘leisure-time exercise’.

I was vaguely aware that workplace exercise did not quite fit the chirpy model put about in videos like this one.  Partly I just knew this from studying the history of work – work that requires exercise wears bodies out – it’s nothing like the experience of exercise that people get from the gym this article has long fascinated me (warning Ben Goldacre is a fat-hating douche at the beginning).  It tells of a study of hotel cleaners, many of whom described themselves as doing no exercise (which shows how deeply the false equivalency of leisure-time exercise and exercise has worked into people’s self-definition).  In one hotel they told the workers specifically that the work cleaners were doing (which is after all hard physical labour) was exactly the sort of exercise that doctors recommend.  The group who were told that saw all sorts of health benefits over the next month.  But the question that I’ve always wondered is – why didn’t they see themselves as doing exercise?

I was recently hanging round with a five-year old who is always running around like a young spider-monkey and she was talking to herself about ‘exercise’ and describing some of the things she was doing as ‘exercise’ – and it was clear that she’d just started school and been told about the importance of exercise and she was trying to figure out what ‘exercise’ was – what part of her constantly moving around counted.  Because ‘exercise’ is not quite synonymous with physical movement – and a five year old need to figure that out – even if peer-reviewed journals only tell us so in their methodology section.

* There may some justifications in the articles that are cited, but I couldn’t access any of the cited articles in the relevant section.  I’d love to hear from people who can if there is a justification if you follow the references.

This entry posted in Class, poverty, labor, & related issues, Fat, fat and more fat. Bookmark the permalink. 

13 Responses to Junk Science that supports Health at Every Size is still Junk Science

  1. 1
    Jadey says:

    I agree that framing an entire movement on the basis of shoddy journal articles is a poor foundation indeed and that scientific literacy is essential across the board.

    That being said, I think there’s some limited value in using articles like this to demonstrate how the same unsatisfactory methods can be used to produce completely contradictory results to the other research available. If someone waves one journal article in front of me to “prove” their position, I am glad to be able to wave another one right back at them to “disprove” it along their own pathway of logic (or “logic”).

    But, as I said, you are absolutely correct that to ultimately be taken more seriously, we need better science. And, equally correctly as you said, we also need to not reinforce destructive “health” narratives, classism, and good-fatty/bad-fatty dichotomies.

    (P.S., the Ben Goldacre link is wonky. Not sure what happened there.)

  2. 2
    Nancy Lebovitz says:


    I wonder whether anything of the sort has been replicated.

  3. 3
    Nancy Lebovitz says:


    A twin study which suggests that dieting contributes to weight gain, even as early as age 25.

    It looks like reasonable research to me, but I’m not an expert.

  4. 4
    Jamie says:

    I am also having difficulty with the Ben Goldacre link. It just goes to the generic Mother Jones politics page.

  5. 5
    james says:

    I actually lose it at this point and can’t form any coherent thoughts. You can’t measure a subset or something a pretend you’ve measured the whole thing. You can’t claim to do one thing, when you’re actually doing something else.

    I think that’s just because you’ve a bee in your bonnet about physical activity and fatness. They don’t care about that. In the context of the study they’re looking at Healthy Lifestyle Habits so excluding work is natural.

    (1) From a population perspective the variety of paid manual labour available isn’t under individual control, it’s fixed and determined by the market. More people can take up jogging, but we can’t all take up cleaning jobs. There’s only so much of that type of work, and if I do it I exclude someone else.

    (2) It’s not really a reasonable request to ask people to give up their job and retrain in something more manual labor intensive, or something which is easily under many people’s control. Asking that if we voluntarily undertake leisure-time physical activities, regardless of whatever crap we have to do to earn money, will this will improve our health is more relevant to the choices within people’s reach.

    (3) Nor is taking up manual labor something public health guidance has asked, and this article is trying to evaluate established guidance. Not something that could have been asked but wasn’t.

    (4) If an independent variable is measured with error, you get an underestimate of the effect size. So theory says if they’d have gotten a better instrument they’d show a larger result.

  6. 6
    Jadey says:

    @ james, re: points 1 and 2

    But both leisure time and manual labour are confounded by class (middle-class people are more likely to work sedentary jobs but have the resources to undertake “leisure time” activities, whereas poorer people are more likely to work physical labour jobs but not have the time or money to go to the gym after work), which is why both are poor indicators on their own (although even in combination it’s hard to say that they would balance each other out). The social construction of “exercise” and associated indicators poorly operationalize the concept of “physical activity” despite assumptions. Hence the criticism.

    re: point 4

    You don’t always get an underestimate, especially if the error is systematic, not random, so it’s difficult to say exactly what a better indicator would show, actually.

    What a better indicator would *be* is hard to say – applied researchers often work with inadequate measures due to practical constraints. But leisure time activities is still a pretty poor gauge of daily physical activity and more likely to be indexing confounds associated with social class (which includes access to better healthcare, less chronic stress, etc.) if not otherwise statistically controlled for, which often isn’t the case, unfortunately (bad form), including in this study according to Maia.

  7. 7
    SharonC says:

    Thanks Maia, your point is a good one that we shouldn’t fall into the same trap as the anti-obesity rhetoric or confusing correlation with causation (or other sleights of hand), and we have to be disciplined about how we use the information from the scientific literature. I agree with Jadey too.

    I have a certain amount of trepidation about the use of scientific studies, as engaging with the anti-obesity health arguments gives the impression of legitimising their rationale for their arguments: if we say “your fat=unhealthy science is wrong” without saying “even if your fat=unhealthy was true then it is no basis for treating people badly” then it looks like we tacitly agree with their use of health status as a justification for intervention.

    However, despite misgivings, I think there are some very strong arguments in favour of engaging with the scientific evidence from research studies:

    (1) Fat haters are using health as a weapon to bludgeon fat people. Whilst one defensive approach is to point out that the weapon is irrelevant (even if fat was unhealthy that’s no justification for treating people badly), but another perfectly reasonable approach is to destroy their weapon (by showing using evidence that their assumptions are wrong). Some people will be fat haters no matter what the evidence says, but others are more compassionate and persuadable. Even destroying or weakening parts of their weapon is useful, for example, making fat haters think twice about promoting ideas without scientific backing would still be useful.

    (2) The truth is valuable. In particular, the truth about weight and health is valuable to fat people. For those fat people who do want to take care of our health, we need decent information from the research, undistorted by weight bias. Having decent information about risks (correlations and all), about preventative measures, about treatment and efficacy for fat people, this is all incredibly useful and important information. And it’s not getting out there.

    The key is using the scientific evidence in a suitable way, without appearing to support the health police or make fat people not at a pinnacle of health feel marginalised.

  8. 8
    Erik says:

    This article is one big straw man, attempting to throw down all of HAES by calling out one pro-HAES study with poor methodology. What about this study that has been widely reported in the media lately?


    This study avoids the pitfalls of both the inaccuracies of BMI by using bodyfat % as a measure of obesity, and the inaccuracies of defining exercise levels (the blog author’s supposed point) by measuring VO2 max, the scientific gold standard of measuring fitness. The results of this study were about as pro-HAES as you can get.

  9. 9
    Yessenia says:

    @ Nancy:
    A twin study which suggests that dieting contributes to weight gain, even as early as age 25.

    Correlation is not causation. Another interpretation that jumps out at me, is that twins who have a genetic and/or social propensity for weight gain would be more likely to attempt intentional weight loss. The twins who did not overeat did not have difficulty maintaining their weight without weight loss episodes, and so they did not attempt them because they were less prone to becoming overweight.

  10. 10
    closetpuritan says:

    I can’t read the study right now, but WRT the idea that some twins would be genetically predisposed to gain weight and those are the ones who would diet: that’s not the way twin studies work. You compare the twins in each set to each other; you don’t compare a set of twins who dieted to a set who didn’t.

    WRT looking at ‘social predisposition’, they could only look at that if they were raised separately. If they were raised in the same house their social predisposition should be the same.

  11. 11
    closetpuritan says:

    Alright, this is a little atypical of twin studies, and both Nancy and Yessenia are right.

    This study shows both that intentional weight loss causes a small weight increase in the dieting twin (where one identical twin diets and the other doesn’t) and also that being genetically predisposed to higher BMI makes people more likely to diet.

    Basically, the purpose of doing a twin study–looking at identical vs. fraternal twins–is so that you can determine the effect of genes vs environment in causing differences between the identical twins. They were able to find a small increase in weight for an identical twin who’d dieted compared to one who hadn’t, which could not be attributed to genetic differences because there were no genetic differences. But the difference in weight between a fraternal twin who’d dieted and one who hadn’t was even greater, leading the researchers to believe that genetic predisposition to a higher BMI leads to an increase in dieting attempts. Unusually for a twin study, they did also compare sets of twins to each other, and found that twin sets where both twins dieted had higher BMIs that sets where both did not diet.

  12. 12
    Barbara says:

    I’ve looked at this article, and to me it’s not “bullshit.” It’s very strong evidence that healthy habits matter more than BMI for mortality. The researchers controlled for “age, sex, race, education, and marital status” and although that doesn’t include class explicitly, educational level is a pretty good class indicator.

    I find these results very compelling, especially considering the number of people studied (11,000+) and the time frame of the study (14+ years).

    If your only real objection is that they equated “exercising regularly” with “leisure exercising regularly” (I can’t look at the study itself at the moment, so I’m not sure about this), you’re going to have to go a lot further to convince me it’s junk science. Sure, some people get exercise at work; but that may not correlate with health, because it may be an overall stressor, not a relaxing type of exercise. Otherwise, I see no problem with the study.

    It confirms to me the importance of adopting healthy habits, regardless of body size.

  13. 13
    closetpuritan says:

    Barbara: There may be something to the idea that work-related exercise is an overall stressor. One data point in favor of that: forcing mice to run in a lab causes their stress hormones to spike. I wrote about that a bit here.