Friday, 31 July 2009

• Health Insurance Makes People Fat? FOX Thinks So

   
  

  I just saw this on FOX.  Some people did a study which determined that, among Americans who own health insurance right now, they are fatter than people who don't. I don't remember who did the study, it's probably easily accessible on google or FOX but my point is focused on the story that FOX is trying to tell about this study, and how bogus it is from the viewpoint of statistics.

  FOX is trying to say that the results of this study prove that it would be a problem to give everyone health insurance because health insurance makes people fat.  They theorize that, because insured people don't have to pay for the consequences of being fat (in money), that means that they are much more likely to carelessly let themselves go and become fat.

  Aside from that fact that is just bizarre on it's face, the people on this show don't know the difference between a simple correlation and a causal relationship. 

  A simple correlation is only one of three criteria which is necessary to claim the presence of a causal relationship. 

  Here are the three criteria necessary to prove a causal relationship (in this case the causal relationship that FOX is claiming is "Health insurance makes people fat.")

  ALL THREE OF THE FOLLOWING ARE REQUIRED

  1)  For a causal relationship to be present, you must demonstrate the presence of a correlation.  Just for the heck of it (and without even seeing the study and/or without analyzing that actual study on all the levels necessary) I will assume that we have this causal relationship based on this one study.  (I won't even talk about all the sampling problems in a study such as this because it isn't necessary to prove my point).

  AND

  2)  For a causal relationship to be present, you must be able to demonstrate a temporal relationship.  If A causes B, then A necessarily comes before B.  If A (having health insurance) causes B (being fat), then A (getting insurance) must come before B (getting fat), namely. people cannot be fat until they get health insurance. Of course we have no way of showing this to be true. It is just as likely as not that people who are fatter are more likely purchase and/or receive health insurance, and it is just as likely (for whatever reason) that fatter people are more likely to purchase or obtain health insurance than those who are not fat.

  Example:

  Maybe fat people are more worried about their health than people who are not fat, and therefore more likely to obtain insurance.

  AND

  3)  For a causal relationship to be present, you must be able to rule out alternative plausible explanations for the correlation. That means the study has to control for everything else that might make one group of people (A = those with health insurance) fatter than another group of people (B = those without). 

  In this case there are so many other variables which might explain this correlation. For example, money might have something to do with it. Class might have something to do with it.  Luck might have something to do with it. Use of alcohol might have something to do with it. Marriage and family associations might have something to do with it.  Likewise we have age, gender, race, fear, personality, disability, and temperament, just to name a few alternative explanations for this correlation in this study. Perhaps people with a certain type of personality style or temperament are more likely to both be fatter and have health insurance, for example, and that same logic could be applied to any of the other variables I have listed.

  Additional possible examples:

  Maybe older people are both fatter and more likely to have health insurance.

  Maybe married people are both fatter and more likely to have health insurance

  Maybe disabled people (who automatically get medicare) are more likely to be fat

  see what I'm saying here?

  I watch this kind of thing all the time on FOX. They take a study and they they find a way, falsely and wrongly, to show it in a light that serves their cause. They are opposed to Obama's health reform, therefore they are going to take this study and say "Oh Look!  Health insurance makes people fat, therefore it is bad, therefore we shouldn't give it to all of our citizens." Lying with statistics...there's an example, and it's common one.

  Now...Do you feel a lot smarter?  If you read and understood this, you are well on your way to getting a doctorate if you want one  

  Anyway, on this topic, how many are going to watch the new show More to Love tonight? Do you agree with the critique of some people who are saying "This TV show More to Love is bad because it will make people fat, or it will make people more likely to stay fat." 

  What do you think might be a problem with THAT claim to a causal relationship?

  • 11:21 AM
  • 519 Views
  • 38 eProps
  • 19 Comments