Thursday, September 3, 2009

Quackery at its best

A friend passed this "scientific" article on to me on the alleged "counter intuitive" finding that relates drinking with exercise. It quotes a study that finds that
the amount of time people devoted to exercise tended to inch up along with the number of alcoholic drinks they had each month.
Unfortunately, this particular study which the post quotes simply represents in my view a spurious finding which might have resulted from missing or unobserved variables.The study asserts causation between drinking and excercise when common sense and personal experience lead us to doubt the link.

There could be an unobserved or missing variable in the model, say age, which correlates highly with both drinking and exercise. Young single professionals for instance engage in social activities that involve drinking and exercise (they have more time for both since they are not raising kids). There is no causation between the two activities. A person's age, occupation, marital status and a host of other factors (not mentioned in the study) could be directing the activities.

Policy analysts, please beware, analytical quackery abounds in the media, even from reputable sources.

No comments:

Post a Comment