April 17, 2002
"Mutant statistics" are stretched, twisted, distorted or mangled versions of original figures, says author Joel Best.
For example, in 1995 an undergraduate reported that 83.5 percent of downloaded images from the Internet were pornographic. The huge scope of the study - 917,410 images downloaded 8.5 million times -- implied that the study must have been exhaustive.
- The problem was the researcher did not collect a representative sample -- instead, he examined postings to only 17 of some 32 Usenet groups that carried image files.
- His findings showed that pornographic images accounted for about 3 percent of Usenet traffic, while Usenet accounted for only about an eighth of the traffic on the Internet.
- Thus only one-half of 1 percent of Internet traffic involved pornographic images -- a markedly lower figure than 83.5 percent.
Other statistics get mangled because they are difficult to grasp. Take the medical study supposedly showing that doctors referred blacks and women 40 percent less often for cardiac catheterization than whites and men. The media reported the odds ratio as the relative likelihood of receiving the procedure.
- The correct comparison would have involved figuring the chance, or relative risk, of being referred for testing.
- If 90.6 percent of whites and 84.7 percent of blacks are referred, then blacks are 93 percent as likely (84.7/90.6 = .93) to get referrals.
- Blacks and women, then, were not 40 percent less likely to receive referrals; they were 7 percent less likely to be referred.
Source: Joel Best, "Damned Lies and Statistics: Untangling Numbers from the Media, Politicians and Activists" (University of California Press), Review, Journal of Economic Policy, Third Quarter, 2001.
Browse more articles on Government Issues