Skip to content

Spuriouser and Spuriouser

May 20, 2014

Weight LossFat HealthFat ScienceExerciseDiet Talk

A few days ago, I came across a Tyler Vigen’s Spurious Correlations. It’s really amazing the things that have absolutely NOTHING to do with each other, and yet there are some amazing correlations.

Interesting, isn't it?

Interesting, isn’t it?

I really like the next one:

Wow!

Wow!

I mean, who would have thought that eating cheese would cause a person to die by becoming tangled in their bedsheets?!

Then there’s this one, a nice downward trend:

Less divorce, less margarine?

Less divorce, less margarine?

That’s what we need to do to save marriages! Just get people to eat less margarine!

Here’s another one with death by bedsheet:

People really need to stop skiing!

People really need to stop skiing!

Now, of course, all of these correlations are spurious.  Nobody in their right mind will ever say “we have to stop skiing so the number of people who die by becoming tangled in their bedsheets lowers.”  I mean, does anyone really believe that people eating less margarine in Maine really had anything to do with the divorce rate?

And yet, that’s exactly what they believe about all sorts of medical things. There’s a correlation between obesity and high blood pressure, so what’s the first thing everybody says? “You should really lose weight and your blood pressure will drop.” There’s a small correlation between obesity and type 2 diabetes, since you have to have the genetic predisposition for it to manifest, and yet, what do people tell fat people? “You should really stop eating sugar if you don’t want to get diabetes!”

Why is it so easy to be able to tell that correlation does not equal causation when it’s the US crude oil imports from Norway that’s correlating with drivers killed in collisions with trains? Why is it so difficult to understand when it’s “if you lose weight all of your symptoms from Polycystic Ovarian Syndrome (PCOS) will go away”? True story, today I was told that it is “commonly known” that if I were to just lose weight, all the symptoms (including lack of menses, male pattern baldness, and excess beard and mustache) would all go away. Yeah, I’m betting the people who “commonly know” this don’t really understand what “genetic” means. And oh yeah, one of the symptoms of PCOS is insulin resistance (which is also known as “pre-diabetes”). And adopting the lifestyle changes may mitigate the risk of IR (and in some cases, may not) without a significant amount of weight loss.

But, you know, just lose weight, and all that will go away! Like I’ve never heard that before.

The rollover at the original site reads: Correlation doesn’t imply causation, but it does waggle it’s eyebrows suggestively and gesture furtively while mouthing “look over there.”

So just remember, just like eating too much sugar always causes diabetes, US spending on science, space, and technology causes suicides due to hanging, strangulation and suffocation. d’Oh! No, I meant just remember, correlation does not (always) equal causation.

Advertisements
20 Comments leave one →
  1. Twistie permalink
    May 20, 2014 9:21 am

    Damn. And here I thought with my butter-eating ways all I needed to do was move to Maine to guarantee that Mr. Twistie and I will never be divorced!

    Then again, if I don’t stop eating so much cheese and if skiing resorts start paying less taxes, it’s virtually guaranteed that I’ll die twisted up in my bedsheets, so it may be a moot point, after all.

  2. Kala permalink
    May 20, 2014 10:54 pm

    This kind of argument is so facile.

    It’s very easy to have correlated data with respect to time, so many processes that have nothing to do with each other move up or down consistently through the years. It’s not the same when you are comparing two variables not as they chug along through time, but as statuses within individuals. Do you use the same mathematical operation on both to get a value for correlation? Yes. Is it the same thing? No.

    This argument also ignores that for a lot of health research, findings in the lab in basic sciences and medical research support conclusions based on data analysis from observational studies that only show associations between variables.

    Causation is nearly impossible to prove with anything, even the obvious, it took decades of research to show that smoking causes lung cancer yet that isn’t a conclusion you would turn your nose up at for being spurious. In epidemiological research there is a threshold of evidence from various sources that is sufficient enough to conclude causation, the idea that being heavily overweight results in negative health consequences is well supported.

    You’ve come to a conclusion for yourself, for personal reasons and not scientific ones, and I doubt there’s any realistic study that could be run that would have you change that conclusion.

    • Bronwen permalink
      May 21, 2014 5:20 pm

      “You’ve come to a conclusion for yourself, for personal reasons and not scientific ones, and I doubt there’s any realistic study that could be run that would have you change that conclusion.”

      A realistic study, that includes the realities of women like me, will never be performed. To say that, for ALL women with PCOS, that “just loosing weight” will stop all the symptoms of it (including the hirsuteness, the ammenorrhea/dysmenorrhea, the pelvic pain, among other things) is to deny the reality of tens of thousands of women. For doctors to continue to tell women “you must not want to have any children because you refuse to lose weight”, even after being told by the patient how much they’ve exercised and how little they eat is, in my opinion, torture.

      I wanted to be able to have children. I have had ammenorrhea since I was 14. At the time the symptoms of my PCOS started (again, 14), I was walking between 2-4 miles a day (round trip to school), active in school (this was when schools had daily PE, and I was taking the “aerobics/weight lifting” elective), plus doing another hour of aerobics at home. But you know, diet and exercise *always* clears up those pesky symptoms, don’t they? Oh yeah, and while doing this, restricting my calories to 1000 per day or less. Yet still managing to gain weight all the time as well. If you think I wouldn’t do anything to get rid of the beard, mustache, and other male pattern hair growth on my body, and stop the receding hairline, you would be wrong. But even when I was on the metformin (given because it helps some of the symptoms for some women), even when I was losing weight on the metformin (because that is a known side effect of the medicine), it still did absolutely nothing for the other symptoms. The beard, the mustache, the balding, all were strong. And I’d lost 60 pounds before my body finally got used to it and it stopped having the only effect on me.

      “the idea that being heavily overweight results in negative health consequences is well supported.”

      And also, the idea that being overweight has negligible health consequences is well supported. The study done showing that an overweight person is more likely to survive a cardiac incident, for instance. Also, many of the negative health consequences have a lot of other causes/correlations as well (including genetics), but that’s not something that’s said very often. When PCOS was actually getting some research in the 90’s, being obese was a *symptom of* PCOS. You didn’t “get” PCOS because you were fat, but you might have become fat because you had PCOS. Now, the discussion is different. But then “You must not want to have babies because you have done nothing to change your genetic makeup” just doesn’t blame the individual as much as “you must not want to have babies because you haven’t lost weight” does, don’t you think?

      • Kala permalink
        May 21, 2014 6:19 pm

        In some research, being overweight has shown a protective or non-negative effect on some outcomes, but let’s remember that none of those results pertain to morbid obesity. How many writers on this website or in the fat acceptance in general are merely overweight or at a lower level of obesity?

        And what PCOS research are you even talking about, Bronwen? You feel that the state of PCOS research has done you very wrong, but what studies are you even talking about? You don’t like that doctors tell you to lose weight to manage your PCOS, and that’s about all I gather, and I’m having a hard time seeing where spurious correlations play into this.

        And beyond this talk of PCOS, it’s clear enough that the point of this article is to make the *spurious* argument that because one can take two random and unrelated processes and show a mathematical correlation between the two, this gives us open season to ignore whatever observational studies we want that have results that we don’t like.

        • Bronwen permalink
          May 21, 2014 6:56 pm

          “How many writers on this website or in the fat acceptance in general are merely overweight or at a lower level of obesity?”

          Why does it matter? Why does the number of “OMG!!DEATHFATZ!!” people who are either writers here or in the fat acceptance movement make any difference? Even *if* every single negative outcome were 100% proven to be caused by obesity instead of mostly correlations or co-morbidities with obesity, so what? Nobody is under any moral — or other — obligation to be healthy, or as healthy as possible.

          “You don’t like that doctors tell you to lose weight to manage your PCOS…” No, let’s get it straight. I think that doctors who tell women who want to be able to give birth more than life itself that the main or only reason they cannot is because they obviously don’t want to, or else they would have lost weight already, are sadists. I think that telling women that their infertility is all their fault, especially when the women say, “But I exercise and I only eat x amount of calories per day” is calling the woman a liar.

          Because, in many women’s experience (of people who have PCOS), the experience isn’t that it’s “difficult” to lose weight and keep it off, but that it’s impossible to lose any kind of weight and keep it off. Besides the fact that for *any* person, long term weight loss is 95% ineffective (that is, only 5% +/- of the people who lose weight actually maintain the weight loss for 5 years or more). I do know some people (two to be exact) who are successful in their weight loss. And both say they realize they are the exceptions.

          So you know, to say, “Just lose weight to manage your symptoms”, when obesity is a symptom of PCOS? Yeah, it doesn’t actually make sense, does it?

          And if your last sentence is all you got out of my article, then you haven’t been paying attention all this time. I don’t need an excuse to ignore the studies that are paid for by the weight loss industry. I don’t need an excuse to ignore the headlines that take the real studies (who actually show something other than what the $60 billion weight loss industry wants to show) and twist them around to say the exact opposite. I can ignore all that easily.

          Correlation does not necessarily equal causation. The fact that high blood pressure and obesity are sometimes co-morbidities does not mean that high blood pressure is caused by obesity. There are a lot of other factors involved, including medication, genetics, and outside stresses.

          Oh, and if you want to really know about PCOS, I suggest Google.

          • Kala permalink
            May 21, 2014 7:47 pm

            And to add one other thing:

            “Nobody is under any moral — or other — obligation to be healthy, or as healthy as possible.”

            I reject this outright. People are eager to clamor about their rights to health care access, among other things. And I agree that we as a society have an obligation to provide health care access. This is a collectivist type of ideal. But how can we expect that access, expect it as an inherent right, without the obligation for individuals to uphold their end of the social contract by doing their best to live in good health, as the other half of that collectivist coin. Instead you argue that no one has any obligation to even attempt healthy habits. But I’m sure you’d think it outright criminal to deny access to health care for those that choose not to try for better health.

            And let me emphasize that I mean that the idea of trying one’s best is relative. My mother is severely mentally ill, and has a number of chronic disease issues, obesity being one of many. My mother, with her very few spoons laying about, could not be expected to take care of herself to the same degree that I could be expected to do so. I don’t expect the overworked and over-stressed to put many hours into cooking healthy meals or spend significant chunks of time exercising. I don’t expect the poor to come up with resources that they don’t have. But I do reject the idea that we are under no moral obligation at all, that’s absurd.

        • Kala permalink
          May 21, 2014 7:23 pm

          ‘Why does it matter? Why does the number of “OMG!!DEATHFATZ!!” people who are either writers here or in the fat acceptance movement make any difference?”

          It only matters because you brought it up as a retort to something I said. I said that there is substantial evidence that being heavily overweight results in poorer health, and you replied that being overweight (but truly only a little overweight) is associated with lower mortality in some situations, as if this was some kind of proof against my point. I felt the need to point out that although there is quality evidence for your point, it hardly applies to the segment of the population that you and other FAs are advocating an anti-science viewpoint for.

          “Correlation does not necessarily equal causation.”
          Do you even read what I write or do you just skim over it because you think I’m full of shit? Regarding correlation and causation, I said:

          “This argument also ignores that for a lot of health research, findings in the lab in basic sciences and medical research support conclusions based on data analysis from observational studies that only show associations between variables.

          Causation is nearly impossible to prove with anything, even the obvious, it took decades of research to show that smoking causes lung cancer yet that isn’t a conclusion you would turn your nose up at for being spurious. In epidemiological research there is a threshold of evidence from various sources that is sufficient enough to conclude causation, the idea that being heavily overweight results in negative health consequences is well supported.”

          And I am already aware of PCOS. I was asking you what research you are complaining about, because after all your first statement back to me was:

          “A realistic study, that includes the realities of women like me, will never be performed.”

          You are convinced that PCOS research has not done women justice, implying that you’d have to be somewhat up to speed with what that research looks like, yet you can’t or won’t bring up any actual research. You linked to Wikipedia and told me to Google it.

          And which research in particular are you claiming is funded by the weight loss industry? Do you have any idea of what percentage of obesity research is funded by the weight loss industry?

          Everything you’ve said to me today is stuffed to the brim with spurious FA mantras on science, with no substantiation. And on top of that you keep mixing these emotional issues you have with physician providers and your PCOS problems with bold claims about research that I doubt you even read. It’s nonsensical.

          • Bronwen permalink
            May 21, 2014 8:32 pm

            Ah, because I don’t step to and do your research for you, you doubt I’ve ever read any research on PCOS. Right.

            I would suggest you read “First, Do No Harm” (http://fathealth.wordpress.com/) to see how hard it is to actually *be* healthy when many doctors refuse to do anything but diagnose “fat”. This might open your eyes. It’s not just me. It’s not just women who have PCOS. It’s many people, women and men, who have real issues that can be addressed by modern medicine, who are told to just lose weight and everything will be better. Much like the whole argument for me to just lose weight and I’d be able to have a baby.

            Once you get done with that (which, I will admit is pretty emotion ridden, but will address your comment about having an obligation to be healthy, because even when people try to be healthy, they are denied appropriate treatment just because of their weight), then I suggest you go over to Junkfood Science (http://junkfoodscience.blogspot.com/). There’s your Master class instruction on exactly which studies (at least up until she stopped updating the blog) were run by the diet industry and how to read the actual studies themselves (with links to the studies in question). Here’s a bonus link from Junkfood Science, about the study that (didn’t) say that those who are only a little overweight (but not obese) live longer than normal weight people: http://junkfoodscience.blogspot.com/2009/06/even-obesity-paradoxes-cant-excuse.html

            Here, let me cut and paste a relevant to our discussion part of this final link I posted:

            “The results of this epidemiological study were published in Obesity, the journal of the Obesity Society. This study found that none of the relative risks associated with mortality they examined were tenable [explained here], except for one. Age. At age 65, the relative risks of dying rose to 44.35 times compared to age 25; and by age 75, relative risks are 119-fold. We should stop right there, as tenable correlations are the only ones that deserve our focus. But that wouldn’t have made a news story, so what followed was splitting hairs among the rest.

            “Looking at corrected BMIs, according to the breakdowns adopted by the world’s governments, the authors found that compared to ‘normal’ BMIs (18.5 up to 25):

            “● being overweight (BMI 25 up to 30) was associated with a 25% lower risk of dying

            “● being obese (BMI 30 up to 35, which includes about 80% of all obese people) was associated with a 12% lower risk of dying.

            “● And the risks associated with the most ‘morbidly obese’ (BMIs 35+) — the uppermost 3% of this Canadian cohort— were statistically the same as those with ‘normal’ BMIs. [RR=1.09 (0.86-1.39, 95% CI) versus RR=1.0.]”

            Interesting. So the “morbidly obese” have the same statistical risk of dying as “normal” BMI. Please do go to that link especially, and see all the ways they tried to make it be deadlier to be obese than to be normal size. Further down she shows how the correlation of obesity and mortality was turned into causation, and then a cry for government action. Spurious indeed.

            Frankly, your rejection that people are under no moral or other obligation to be healthy is irrelevant to any conversation on fat acceptance. It is nothing more than condescending and derailing.

          • Kala permalink
            May 21, 2014 10:52 pm

            You are directly complaining about research and how it doesn’t serve women such as yourself well. One would think you could provide off hand something, anything that is an example of the point that you’re making. You have been mistreated by medical providers, who are not infallible and can and do go against their own established best practice in their fields. But that doesn’t make the research wrong, but according to you all the research is being done wrong.

            I’ve seen plenty of the fathealth blog, so there’s not going to be any eye-opening from there.

            I am also more than familiar with Junkfood Science, which is really rather terrible for a lot of reasons, and you do yourself a disservice by citing it. I don’t even need to sit here and trot out why I don’t take Junkfood Science seriously because Atchka already did a good job describing a lot of the issues with Sandy Szwarc and her work:

            http://fiercefatties.com/2013/09/19/broken-clock/

            I think your issue here is that you are assuming that I’m just over here existing in ignorance of all these fantastic “sources” that you are familiar with. But I’m not, I am really not. I don’t need Sandy Szwarc to tell me how to read the research because I have a degree in the field and work in research, where she herself is a writer with a nursing degree who has a history of misrepresenting information about as badly as anyone she throws accusations at. That paper from Obesity that Sandy cites is far from the only paper about the so called “obesity paradox”, and others have shown that the protective benefit doesn’t extend to the very heavy. There’s an entire other discussion on the statistics, which is beyond the scope of this argument we’re having.

            Again, you brought up the issue of a moral obligation to be healthy, not me. You brought it up, and I responded that I don’t find it to be a valid argument point because I don’t think agree with your view that there is no obligation. How is that derailing? Because you found it condescending?

            If you would have written about how you had been mistreated by doctors in your journey to try to have a child with PCOS, because you are a fat woman, I think that would have been a great article. Instead you wrote a snark-filled article about spurious correlations, implying that because one can show mathematical correlate unrelated data, that we don’t have to take associative assessments from research seriously if we don’t feel like it. And I think you are wrong, for several reasons, and that’s why I responded.

            You can call me an asshole or a troll, as I’ve been called here in the past, whatever you like, but it’s not as if I broke into your house and read your diary and gave you criticism about it. You put this out there for people to see.

            • Bronwen permalink
              May 21, 2014 11:35 pm

              “But that doesn’t make the research wrong, but according to you all the research is being done wrong.”

              Nope. I said that the research in the late 1990s showed that obesity was a symptom of PCOS. And that now, it is being stated that PCOS is not a symptom of, but one of the causes of. That is not saying “all the research is wrong.” That’s stating that they have changed the focus.

              You continue to try and make me say something I’m not saying.

              You brought up the number of obese writers on this blog or in the FA movement in general. As this is about the health consequences or non-health consequences of correlation of research, and as you were stating about the negative consequences of obesity, the idea of there being no obligation to be healthy, moral or otherwise, is relevant. You disagree. That’s nice. The derailing came, and comes, not only in the way you disagreed at first (trying to get us onto universal healthcare), but also in the way you keep bringing it up.

              Also, you don’t get to police the tone of what I write. The article was snarky. Awesome! I succeeded in my goal then, which was to write a snarky article.

              I hope you have a great night. I have things to do that don’t include going around the same points. Take care now!

            • Kala permalink
              May 21, 2014 11:44 pm

              The focus changed, and you clearly disagree. You scoff at it. Otherwise why bring it up? You think the research focus is wrong, don’t take stock in the correlations, and you directly state that you don’t think the research does women like yourself justice. You think it’s all so absurd you compare the false association between being killed by one’s bedsheets and ski resort income to the association between high blood pressure and obesity. Because how silly! Correlations can show whatever, why take them seriously? So no, while you didn’t state you think that obesity researchers have it all wrong, you certainly applied it with your sharp wit.

              And no, I suppose I don’t get to police the tone. But all the same it’s a snarky, shitty article. It’s an observation of mine, so of course you don’t have to change it. Considering you only need to preach to the chorus, I’m sure the snark is just fine.

  3. Dizzyd permalink
    May 22, 2014 6:12 pm

    Kala, I have to wonder – if you think that all of us fatties are just eating ourselves into the grave and just twisting facts to make ourselves feel better, rather than face the facts that our obesity is killing us like we’ve been told all along and are too stubborn to believe, why do you even come on here? It seems like you’d be more at home with the “obesity is killing us” crowd. I think what Bronwen was trying to say is that you shouldn’t have to partake in healthy behaviors in order to be treated with respect. Sure, we should do what we can to make our lives better, but it shouldn’t be a prerequisite to be treated with basic human dignity and respect. Does someone like Vesta not deserve to be treated with dignity because she’s unable to get out there and run marathons and eat broiled chicken and rice and salad? I know you were trying to point out the inconsistencies of correlation and causation, and maybe Bronwen did it wrong while trying to prove a point, but it just felt like you shot her down. It felt like you were so busy arguing semantics, you couldn’t see she was trying to point out the horrible experiences she had with PCOS. I’m sure you weren’t trying to, but try to see where she’s coming from, too. Could she have done a better job? Probably, but the point we all try to make on this site is that fat people are often treated poorly in society and then told it’s all our fault and WE need to change in order to become acceptable and human, oh and by the way, you’re gonna drop dead any minute now. I’m not here to get in an argument, I just sat there reading this discourse and shaking my head in sorrow that two people were not just disagreeing but arguing.

    • Kala permalink
      May 23, 2014 12:35 am

      I think there’s a pretty big gulf between “if you think that all of us fatties are just eating ourselves into the grave” and “substantial evidence that morbid obesity has a negative effect on health.” The idea that one can’t one can’t assess the sum total of health by appearances is one that I agree with, but I reject the denial that fat has quite literally no negative effects. And I made no comments about treating people with dignity. I said that it was ridiculous to say that we as individuals have no obligation whatsoever to try to live in health. I am not expecting Vesta to run, let alone run a marathon, that would be prescribing healthy behaviors (not that the marathon distance adds much of anything to health) to an arbitrary standard, and not prescribing them relative to a person.

      Bronwnen didn’t write an article about PCOS, she wrote an article that was dismissive about correlations. It became clear in her comments that truly she is really focused on how she has been treated as a woman with PCOS, but that doesn’t negate the criticism I had about the original article.

      FA is anti-science as a whole, and although not all individuals in it are anti-science, most of the big names are. I find it incongruous that you could have the very same people when faced with a health outcome paper, say that the study was poorly designed, the results misinterpreted, or be totally dismissive of correlations, yet when faced with research on weight loss effectiveness, be the first to accept the results of a study when it shows that the weight loss outcomes are poor. Why are the latter studies not designed poorly such that they don’t reflect reality, why can we not take any stock in the numbers from the former but serious stock in the numbers from the latter?

      • Leila Haddad permalink
        May 23, 2014 3:27 am

        “FA is anti-science as a whole, ….” Shannon, how did you not catch this?

  4. May 22, 2014 8:01 pm

    Reblogged this on The Cheese Whines and commented:
    I dispute the idea that eating cheese will cause people to die by becoming tangled in their bedsheets. However, it’s a well-known fact that cutting the cheese while tangled in the bedsheets with a partner may kill said partner.
    Fart joke! Ba-dum-tssss!

    • May 23, 2014 9:50 am

      Well, Windsor McKay did once devote entire comic strips to people who got tangled in their bedsheets from post-cheese-consumption nightmares. (Said comic was called “Dreams Of The Rarebit Fiend.”) So maybe that particular correlation isn’t so spurious after all.

  5. Bronwen permalink
    May 23, 2014 1:11 pm

    Kala —

    I disengaged from you because the conversation was going nowhere. You stated that you weren’t trying to derail the conversation, and demanded I show how you were. So, here’s a little lesson for you:

    You tried to get me to focus on the health care industry and universal healthcare. That may be a discussion to get into at a further time, but it has nothing to do with spurious correlations. And go ahead and say “but you brought up blah blah blah”. You started the derail, I started to fall for it, and then decided to not go there. Just because you want to continue down that rabbit hole doesn’t mean I have to follow, even if I did follow for a minute or two. That’s the grand thing about being an adult with choices — I can choose to stop falling for a derailment when I see it.

    You twisted my words around, repeatedly, to mean something I never said. You tried to mind read and say that the post was really about me and the way the doctors treated me, instead of how correlations can be made about a very many things.

    You called me a liar.

    You demanded I show you the research I’ve done over the years. You whined when I refused to do so, because I am under no obligation to anybody to do so.

    You tried to police my tone.

    After all of that, do you think anybody is going to believe that you weren’t trying to derail anything?

    At one point, you said that you’ve been called a troll many times. Well, I say, if the shoe fits …

    And as I have a life outside of the computer, this will be my final response to you on this post. So, go for it, tell me how terrible my communications skills are, how it’s all about me, how my tone is so bad. You have the opportunity for the last word with me. Not saying somebody else won’t call you on your bull shit, but I won’t be able to respond until at least Monday. And at that point, it’s really not worth my time to come back and engage.

    • Kala permalink
      May 23, 2014 1:31 pm

      You must be kidding me. You wrote an article about correlations. I responded with a comments about correlations and causation. And rather than perhaps argue about my comments on correlations and causation, you instead replied to me personally with an onslaught of information about your issues as a woman with PCOS and infertility. And from there we went back and forth. You reap what you sow.

  6. Dizzyd permalink
    May 26, 2014 11:08 am

    I agree that the examples of corellation are kind of ridiculous, but then some of the corellations brought up against fatness start to stretch it too. I think that Bronwen was making the point that anyone can corellate anything to make a point and people will go for it, no matter how big a stretch. I believe Shannon said as much in some of his earlier posts, as well as other experts – i.e. Glenn Gaesser, endocrinologist Diane Budd (quoted in ‘Fat!So?’ the book) and others. Do fat bodies have built-in problems? Yes, of course! So does every body type, even the vaunted thin body (osteoporosis). I won’t argue with you, Kala, as I’m sure you could argue me under the table regarding statistics – which is all Greek to me. 🙂 Should we all do what we can to take care of our bodies? Of course, they ARE our bodies after all! The point we make – and I’m sure you’ll agree – is that everyone, especially fat people, deserve kind and decent treatment, no matter what their health is like. (Take note, medical personnel!) And with that, I wish everyone a happy Memorial Day – or happy Monday if you’re outside the U.S.

  7. Happy Spider permalink
    May 26, 2014 1:53 pm

    I feel confused as to whether you all are talking about spurious correlations or about “”correlation does not imply causation.”

    I think some of those Vigen graphs might just be the effect of the birthday problem and so might not be correlated at all. That’s like the problem of overfitting or seeing patterns that aren’t there. If two things aren’t really correlated then you should find that the graphs look less similar as you get more data. I guess if your two things didn’t vary– like if the graphs were straight lines– then the graphs would always be correlated even though it seems like that sort of thing should be considered a false correlation, but the graphs above all wiggked up and down instead of going in straight lines so I think they should be the normal sort of false correlation that gets cleared up with more data.

    After showing the graphs, the post says that obesity used to be considered a symptom of PCOS. But if that’s the case, then isn’t obesity truly correlated with PCOS? So that’s not a spurious correlation. It’s a true correlation. The argument is not that they’re not correlated but that obesity and PCOS’ being correlated does not imply obesity causes PCOS.

    So the example of the Vigen graphs doesn’t seem relevant to talk about the relationship between PCOS and obesity. I don’t think the Vigen graphs are relevant to talk about the correlation between obesity and high blood pressure or obesity and diabetes, either.

    Besides spurious correlations and “correlation does not imply causation” there’s a third, hidden, topic that really confusing the issue. I can’t find the right words for it, but it’s sort of the idea that if someone is fat then you should ignore any of their health problems. Maybe that’s the subject “how to deal with a true correlation”?

    The three topics call for completely different responses. A spurious correlation should be ignored. A true correlation without a convincing explanation should be examined further to find out if it’s a causal relationship or if there is some third factor common to both that is influencing things. A correlation that is causal in the sense that you do believe that, say, altering your weight alters your tendency to high blood pressure, calls for investigation into how strong the correlation is and whether it’s actually possible or advisable to change the variables under discussion. This third case is where you get into discussions of how hard it is to lose weight and so how unreasonable it is to punish people for failing to do so.

    So it’s all confusing.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: