The Numbers Guy on Obesity

In an article called “Obesity Study Looks Thin” in Friday’s Wall Street Journal, Carl Bialik makes a grimly amusing observation:

In 40 years, every single American could be overweight, according to a recent study.  Employing that same logic, 13 out of every 10 adult Americans by then won’t have landlines. (Note 1.)

This ridiculous prediction made by the Nature Publishing Group journal Obesity - that literally 100% of Americans will be obese by 2048 - was made by plotting points representing obesity data from the last three decades, finding the slope of the trend, then simply (and brazenly, as Bialik aptly put it) extending the line into the future.

It is no wonder the study stopped the prediction at 2048, since beyond that, even the most gullible of readers may have suspected the results: they would have showed that there would be more fat Americans than there would be Americans.   

The fact that there was little justification for dropping a straight-edge on a graph and drawing a line did not prevent the study from circulating widely.  Apparently, it appeared online not only at the Obesity web site, but on the web pages of Reuters and Matt Drudge, and it is to be printed in the October issue of Obesity.

Now, the point I want to emphasize here is not that every claim of obesity is overblown.  Clearly, many Americans eat too much and have unhealthy eating habits.  What I am concerned about is the numbers game itself.  If the prophets of doom had wished to show a trend not toward American obesity but toward American starvation, they had little to do beyond finding some pretext to draw the line with the opposite slope. 

Furthermore, I am concerned with the way these numbers are used.  This has become the pattern:  Fantastic predictions such as these are published in the news today, are digested uncritically by the public tomorrow, and end up appearing in the abstracts of bills on legislators’ desks the day after that. 

Following every whiff of a problem in the realm of personal responsibility comes the public cry, “There ought to be a law!”

It wasn’t so long ago that I used the poor eating habits of Americans as an argument, in the form of reductio ad absurdum, to defend the tobacco industry in conversations with colleagues.  My case would go something like, “People choose to smoke, so they have no one to blame but themselves for the consequences.  To blame the tobacco companies is like blaming fast food restaurants for making people fat, which is ridiculous, right?”  At the time, it seemed completely absurd to think that legislators would come to assail restaurants for using “trans-fats.”  Reports like this one from Obesity pave the way for this intrusion into our lives. 

Incidentally, the author of the WSJ article, Carl Bialik, has an interesting blog called The Numbers Guy, in which he covers the way numbers and statistics can be used to convey information... and sometimes deceive.

NOTES

1. Carl Bialik, “Obesity Study Looks Thin,” Wall Street Journal, 15 Aug 2008, p.A11.