Wednesday, April 29, 2009

Teenagers consuming one-fourth of the nation's alcohol

According to an old New York Times article, a Columbia University study that produced an often repeated sound bite about teenagers consuming one-fourth of the nation's alcohol used faulty statistical methods to arrive at that particular statistic. This is an excellent case of "figures don't lie, but liars do figure"--the researchers oversampled young people in their poll by a factor of two (out of just over 25,000 recipients, they included 10,000 people between the ages of 12 and 20) and then failed to adjust the data to reflect the skewed sample. The real figure is about 11 percent. My knowledge of statistics is pretty shaky, but even I know the basic rules of sampling; this kind of sloppy math is absolutely inexcusable.

Of course, no one's going to be ashamed of having made such a mistake, because it succeeded in bringing public attention to the "social problem" of teenage drinking. Welcome to the "it's-the-thought-that-counts" school of leftist social science. The article cites several other statistics, including percentages of teenagers that reported drinking any alcohol in the past month (about 40 percent) and that engaged in binge drinking (about 7 percent). I'd like to know what percentage of those occasional drinkers and binge drinkers actually proceed to full-fledged alcoholism later in life, but I guess that's too much to ask.