Today's post comes from the kind offices of Marc Abrahams, the genius mind behind the IgNobel prizes, and the wonderful person who sends Sci the best, and sometimes the worst, hilarity that science has to offer. And today we've got a special one. A paper that shows you that you can find a correlation anywhere, if you just take the time to do the math.
Wu, F. "The correlation between tornadoes and trailer homes" Annals of Improbable Research, 1995.
It never fails. If you're going to have a movie with a tornado in it, it is absolutely essential that you also have also a whole mobile home whipping around the funnel of the thing. If it's a funny movie, there's also a good chance of seeing a surprised looking cow.
But the trailer homes. Why is it always the trailer homes? The author of this "study" wanted to know if there was a correlation between states with large numbers of trailer homes and large numbers of tornadoes. I mean, what if trailer homes just bring those big funnels of wind home with them?!
Note: The author of this study is perfectly well aware that tornadoes are formed as a result of severe supercell storms, land topography, and air rotation, and in fact have little or nothing to do with whether there are trailer homes, camcorders, or large cities in the area.
So the author did some correlations, comparing the amount of money spent purchasing mobile homes and the frequency of tornadoes. He found that both tornadoes and hurricanes are concentrated in the areas with the most prefab homes, including states like Kansas, Oklahoma, Florida, and Louisiana. Indiana, in fact, ranks second in mobile home purchases and has the most tornadoes (the author pauses to consider the rotational effects of the Indy 500).
But then of course, what if it's something else? After all, areas with lots of tornadoes in them are also areas with other features! Like plains, wind patterns...and fewer large cities.
Apparently there is another myth that tornadoes never occur near large cities. The fact that there are few large cities on the great plains couldn't be a matter of, say, infrastructure or large rivers or existing businesses. Nope, gotta be tornadoes. The author did some data collection and found that fewer large cities does tend to correlate with the trailer homes and the tornadoes, but not entirely, since states like Alaska, Montana, and Wyoming have very few large cities, but the tornadoes leave them alone. And then of course there was the big ol' tornado that hit Salt Lake City in 1999.
But then the author noticed one final thing about tornadoes. No matter where they hit, there always seems to be someone around to take shaky camcorder video, giving us our iconic images of flying trees, mobile homes, and surprised bovines. Could the tornadoes be posing for pictures? The author checked, and did find a correlation between camcorder sales and tornadoes. I knew those tornadoes were desperate for fame.
So what can we conclude from this? I'll leave it to the author:
1. Real statistics can be used to verify virtually any harebrained tale about tornadoes.
2. In case one of the ideas turns out to be right, I recommend that you build a house, make it sturdy and site it near a city, far from trailer homes, behind a big rock, and don't let anyone with a video camera come near the place.
Wise advice. But I wonder: sure, statistics can find the associations between mobile homes and camcorders and cities, but what about the others?! I will not rest until statistics can find me an association between surprised flying cows and tornado incidence. I feel it. Science has the answer.