I read an article in the February 27th issue of the Wall Street Journal that I thought was pretty interesting. An economist from Cornell University, named Michael Waldmen, made waves when he published a paper that tried to show a link between autism and Television watching. The paper has incited the wrath of families and other experts, who believe evidence shows that genetics play the biggest role in determining whether a child develops autistic disorders, and have led people to question whether economists like Waldmen are "stepping over the line" and getting involved with issues that they shouldn't. Part of the discussion centeres around the method Waldmen used to come to his conclusions, the statstical teqnique of the Instrumental Variable, when a random or natural variable is introduced to help researchers sort out caue and effect. Often it is difficult to determin in research whether A causes B or B causes A. For example, if it was shown that autistic children watch a lot of TV, it could be that the TV causes them to be autistic, or that something about being Autistic makes them more interested in TV. The ideal instrumental variable is one that is shown to be correlated with A but has no direct effect of it;s own on B. If the data in the study shows that the instrumental variable is linked to B, then it suggests that A must be contriuting to B. The somewhat strange instrumental variable that Waldmen chose for his study was weather; because weather data and goverment time-use studies showed that kids were likely to watch more TV on days when they were forced to stay home due to weather, Waldmen used weather as the variable to help him determin if Autism was linked to TV viewing. The researchers took precipitation and autism data from Washington, California, and Oregon. The results seemed to show that children who grew up in years of unusually high precipiation were more likely to be diagnosed with Autism. A second study using the percentage of houselholds that prescribed to able as the instrument turned out similar results. Walden himself, however, is willing to admit that his hypothesis is not all inclusive, and that there are likely other triggers besides just TV watching that might cause the disorder. His interest in the subject began when he son was diagnosed at risk to evlop autistic behavior and Walden subsequently cut down on his child's TV watching and saw an obvious improvement in his behavior.
Anyway, I found the article pretty interesting because I have Asperger's Syndrome, a form of austism, and I'm always interested in reading stuff like this about the causes of the disorder. Personally I think that if TV is a factor in causing these disorders, that it's a smaller factor than genetics. But I thought the statistical method that the professor used for his research was interesting, I'd never heard of "instrumental variables" before.