Tuesday, September 7, 2010

On the Duralbility of False Statistics

Occasionally a statistic gets into the public domain through the news media that turns out to be untrue. A recent article in the Wall Street Journal illustrates this. We often hear thta the typical American will go through about seven careers in a lifetime. I have heard that claim at college graduations and from politicians. Carl Bialik explores the truth to this claim and finds that there is no support for it. Often, the source is listed as the U.S. Bureau of Labor Statistics, but the data are not tracked by the BLS.

A number of years ago I read an article in which the author searched for the source of a claim he had seen in print numerous times. The claim was that one in four coeds would be raped at a particular university in the four years they attended the school. The author said he found it hard to believe. If it were true, parents would have to be crazy to let the daughters attend the school. He tracked it down to a speech by someon who came up with the number by taking the reported number of rapes and increasing the number by some factor to account for unreported rapes.

Bjorn Lomborg, in his book, The Skeptical Environmentalist, offers another example. The press often would quote a statistic that 40,000 species are lost every year. Where did the estimate come from? Lomborg found the source in a book by Norman Myers, The Sinking Ark. Myers took a number of 100 species a year, which was a "hazarded guess," at a 1974 conference. Myers then stated that the number seemed low. He then supposes that the final 25 years of the 20th century would witness the elimination of 1 million species. This works out to 40,000 a year. But the number was not based on evidenc; it was pulled out of the air.

Too many of us assume that a number that appears in a newspaper must be true. But reporters are often not statistically trained, and apparently seldom try to find the source of a claim. Some skepticism is needed by all of us.

No comments:

Post a Comment