Tuesday, July 17, 2007

Three Generations of AP Reporters are Enough: Matt Crenson and Height

The nineteenth and early twentieth century claim that human beings ought to be bred to become taller, more intelligent, stronger or more attractive was discredited in part by the fallacious statistical reasoning of some of its advocates and in part through its association with Nazism. From the 1920s through the early 1970s, the Lynchburg Training School and Hospital forcibly sterilized 4,000 people who were mentally challenged, some of whom would not be considered so today. The sterilization policy was legally challenged in the case of Buck v. Bell. In this famous decision Oliver Wendell Holmes declared that "three generations of imbeciles are enough".

In the nineteenth century, one of the errors that eugenicists made was the fallacy of regression to the mean. Eugenicists noticed that tall fathers had shorter sons, and concluded that people were becoming shorter. In fact, extreme realizations of a random process are usually followed by less extreme realizations. The same gene pool that produced a great person will usually produce lesser lights in subsequent generations since the exceptional person was an outlier, an exceptional realization of the distribution.

In the case of height, there are a number of reasons for height differentials. While diet and quality of health care might contribute to height, there are also genetic differences across ethnic groups. Thus, it is not a particularly interesting observation that height can correlate with wealth (since diet and health care contribute to height) but that genetic differences probably play a role as well.

In a recent AP column, Matt Crenson draws several conclusions that are as naive as those that 19th century eugenecists drew despite a century-and-a-half of improved statistical knowledge. Crenson observes that America is not the tallest country and that the average height in many European countries is taller. He then goes on to make the statement:

Many economists would argue that it does matter, because height is correlated with numerous measures of a population's well-being.

Crenson falls prey to a basic fallacy of statistical reasoning, taught in any basic statistics course: correlation does not imply causation. It may be true that height is correlated with well-being, but it is absurd to claim that well-being is the cause of height differences between US and European countries. Only an AP reporter would imagine that Americans are suffering from calorie deprivation.

Crenson contradicts himself as follows:

"Like many human traits, an individual's height is determined by a mix of genes and environment. Some experts put the contribution of genes at 40 percent, some at 70 percent, some even higher. But they all agree that aside from African pygmies and a few similar exceptions, most populations have about the same genetic potential for height."

This paragraph illustrates what is wrong with the media today. Does Crenson really believe that if genes contribute 40 to 70 percent of variability in height, then different populations with different gene pools do not have differences in height?

The median 35-45-year-old male white American's height is about 70 inches. The median 35-45-year-old Mexican-American's height is about 67 inches. The differences in height between Europeans and Americans to which Crenson alludes are 1-2 inches.

Crenson notes that there is a 1 1/2 inch difference in average height between US cities and rural areas so that rural Americans are as tall as Europeans. But, due to Crenson's being the third generation of AP reporter, he cannot figure out that there are ethnic differences between American urban areas, where somewhat shorter Mexican-American and Asians are more common, and rural areas where slightly taller whites predominate.

Perhaps Crenson's solution is to sterilize urban Americans?

No comments: