sarahmichigan (![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
sarahmichigan) wrote2007-02-07 07:02 am
sarahmichigan) wrote2007-02-07 07:02 am
Entry tags:
Fallacy of the Day: Confusing Correlation and Causation
The classic example of this is "When ice cream sales are high, so are murder rates. So ice cream consumption must cause murders." Of course, there's no causation there; ice cream sales and murder rates just both happen to go up when the weather is hot. Other examples can be found here:
http://www.don-lindsay-archive.org/skeptic/arguments.html#correlation
There are a lot of arguments about the dangers of obesity that rely on this fallacy. People assume that because there's some (but not as much as we're lead to believe by the popular media) correlation between increasing weight and increasing incidents of certain diseases and premature death that obesity causes these diseases and premature death. But when you factor out certain lifestyle elements, such as activity, consumption of fruit and vegetables, and smoking, the picture becomes more complicated. It's more likely that lack of activity and poor eating habits both contribute to obesity and to ill health, not that obesity causes ill health in and of itself, since people who are obese but who exercise regularly have virtually the same life expectancy as thinner, active people.
This fallacy has also mislead some people to be anti-vaccination because of the supposed correlation between exposure to mercury in vaccines and the rising rates of autism. As more children have been vaccinated, the argument goes, more cases of autism have occurred. Therefore, the mercury in vaccines must be causing autism. There are no well done studies that show any link between the mercury in vaccines and autism. What is more likely going on is that over the years, screening is catching more cases of autism in children AND the criteria for diagnosing mild forms of autism have broadened, resulting in larger numbers of children being diagnosed as autistic, not necessarily that there are many, many more causes of autism occurring than in previous years.
On the other hand, you can't entirely ignore strong correlations in medical studies; this is exactly how we found out that tobacco smoking causes cancer. The strong correlative evidence between cigarette smoking an rates of cancer is something like *900 times stronger thank the link between obesity and most diseases it supposedly contributes to. And, we also understand something more about the mechanism of how tobacco smoking causes cancer, which lends additional evidence besides the correlative evidence to the smoking/cancer link.
* I remember the numbers being something close to that according to Glenn Gaesser in "Big Fat Lies" but I don't have the actual number in front of me. Post to be updated with exact number when I can track it down.
http://www.don-lindsay-archive.org/skeptic/arguments.html#correlation
There are a lot of arguments about the dangers of obesity that rely on this fallacy. People assume that because there's some (but not as much as we're lead to believe by the popular media) correlation between increasing weight and increasing incidents of certain diseases and premature death that obesity causes these diseases and premature death. But when you factor out certain lifestyle elements, such as activity, consumption of fruit and vegetables, and smoking, the picture becomes more complicated. It's more likely that lack of activity and poor eating habits both contribute to obesity and to ill health, not that obesity causes ill health in and of itself, since people who are obese but who exercise regularly have virtually the same life expectancy as thinner, active people.
This fallacy has also mislead some people to be anti-vaccination because of the supposed correlation between exposure to mercury in vaccines and the rising rates of autism. As more children have been vaccinated, the argument goes, more cases of autism have occurred. Therefore, the mercury in vaccines must be causing autism. There are no well done studies that show any link between the mercury in vaccines and autism. What is more likely going on is that over the years, screening is catching more cases of autism in children AND the criteria for diagnosing mild forms of autism have broadened, resulting in larger numbers of children being diagnosed as autistic, not necessarily that there are many, many more causes of autism occurring than in previous years.
On the other hand, you can't entirely ignore strong correlations in medical studies; this is exactly how we found out that tobacco smoking causes cancer. The strong correlative evidence between cigarette smoking an rates of cancer is something like *900 times stronger thank the link between obesity and most diseases it supposedly contributes to. And, we also understand something more about the mechanism of how tobacco smoking causes cancer, which lends additional evidence besides the correlative evidence to the smoking/cancer link.
* I remember the numbers being something close to that according to Glenn Gaesser in "Big Fat Lies" but I don't have the actual number in front of me. Post to be updated with exact number when I can track it down.
no subject
If you truly boil out correlative issues -- like links between unhealthy diets and obesity, or between unhealthily low activity levels and obesity -- then we're effectively comparing two people with identical direct health metrics (whatever they may be -- activity, blood chemistry, whatever).
In such a comparison, it seems intuitive to me that -- all other metrics being equal -- the person with (a) greater energy reserves, and (b) better genetic/dietary resiliance to body fat in general would naturally be better off for both contracting and fighting off some diseases. Again, all other things being equal, it makes perfect sense to me that a person with higher body fat would be far better off in most cases.
I think what limits the applicability of this tho to (sadly) only the few folks who are very educated on the topic is the fact that "all other things being equal" is perhaps the exception and not the rule for most Americans. Many people probably just (wrongfully) assume that obesity is the indicator of some other health problem; that may be true more often than not, but the point is that people (including some doctors) don't realize that it's not always true.
For completeness, my understanding is that there are a few diseases where greater body fat simply makes the disease more risky just from a body-chemistry standpoint -- but my source is only secondary and may be part of the biased world. Diabetes is the example i've heard about; something about higher body fat making sugar levels harder to regulate simply for chemical reasons, all other things being equal. Does that match your research?
no subject
However, I know of other studies done where obese folks were put on a food and exercise plan and improved their insulin sensitivity without losing very much or any weight; in one study, average weight loss was on the order of 5-6 pounds and some didn't lose any weight at all, but all improved blood pressure, insulin sensitivity, and other measures of health.
The one health condition I think there's a fairly good reason to believe that it IS causation rather than correlation is in joint issues; more weight on the joints, more joint problems. However, since overweight people often have, for instance, arthritis in the wrists as well, it doesn't seem like 100 percent can be blamed on the weight, either.
no subject
The joint-issue point is a good one, but it also has to do with weight, which can also just be a person's size or musculature, right? Muscle is heavy, too... though maybe in the muscle case, the joints are more supported (by the muscle) so it evens out?
Interesting stuff. The end points seems to persist: it's best to use better metrics than size/shape/weight to determine health and/or diagnose problems; after that, your body will probably be healthiest being left at whatever size/shape/weight falls out of pursuit of those healthy metrics, and if that resultant shape includes some fat reserves, you will have some advantages against some diseases (and perhaps disadvantages with few others, but that's not certain). Nothing about that seems counter-intuitive to me.