"Link" is not the same as "Cause"

When you read studies--about obesity or anything else--you have to remember that being "linked to" something is not the same as a cause and effect relationship. The technical term for it is a "correlation"--something that happens more often in the same groups. For example, there's a provable, statistically significant link (a correlation) between the number of working telephones in any square mile, and the amount of crime that takes place in that square mile. Proof that telephones causes crime? Not really--a third variable is causing both things: Population density. More people per square mile own more telephones and also commit more crimes. The connection is indirect. Taking away telephones would probably have no effect on the crime rate. Another example: The longer a child's arms are, the better he is at reading. Sound odd? But it's true--because older children have longer arms. Age is the third variable in that case.

(An actual cause-and-effect can also be the other way around than what you expect it to be--rather than assuming telephones cause crimes, it could be that people who commit crimes need telephones to commit them, and so buy more. I suspect the link between obesity and lack of exercise may be of this sort: Being fat makes it socially and physically more difficult to exercise; but not exercising may not actually make you fat. Quite a few fat athletes and even more thin couch potatoes seem to be pointing to a more complex relationship than we expect in that case.)

Applying this to the cancer studies, the important point here is that the only thing those studies prove--and can ever prove--is that obesity and cancer occur a little more often (not much more often, if this article is correct) together, in the same groups of people. There's no proof that there's not a third thing that causes both--maybe an unhealthy lifestyle, genetic differences associated with both cancer and obesity, general ill health that impairs the cell self-destruct mechanisms that prevent cancer, heightened life stress making you vulnerable to both obesity and cancer, or even just the fact that people in many Western countries are likely to have enough food to get fat and live long enough to die of cancer. We don't know. Correlation can't tell us that. And the alternate theories are just as reasonable as the idea that obesity causes cancer directly.

The only thing that can establish cause and effect is a controlled experiment. Get yourself a hundred or so people, and force them to get fat. Get yourself another hundred people. Make them stay thin. Feed both groups the exact same things (only feed the fat ones more), give them the exact same environment, and make them exercise equally. Keep everybody in the dark about what is supposed to happen, so the powerful human psyche doesn't mix things up. Then watch them for a lifetime, and count the number of people who get cancer. (Most researchers use rats or something of that sort, naturally, since it would be impractical to study humans this way!) You can't prove cause any other way. And even then, there'll always be a chance that, even if you designed your experiment perfectly, you could come up with the wrong results. The more subjects you use, the smaller the chance is. With ten subjects, it's about a 30% chance. With a thousand subjects, it's down to around 1%. (General public take note: Studies with tiny numbers of participants are to be taken with an entire shaker full of grains of salt.)

Also remember: Studies that come up with positive results (that supported the hypothesis, such as "obesity is linked to cancer"), are a lot more likely to be published... which means that it could be some of the badly designed or very preliminary stuff that's coming to the forefront, while well-designed research that came up without the expected result is still sitting in someone's desk drawer, unpublished!