Does the color of your uniform or sparring equipment affect your chances of winning a sparring match? Do red-clad fighters win more often than do blue-clad fighters?
A May 19, 2005 Nature journal paper that suggested that boxing, Taekwondo, and freestyle and Greco-Roman wrestling Olympic competitors who were clad in red more often beat competitors who were clad in blue for evolutionary reasons.
In the study, evolutionary anthropologists Russell Hill and Robert Barton of the United Kingdom's University of Durham wrote, “Red coloration is a sexually selected, testosterone-dependent signal of male quality.” The report found, "a consistent and statistically significant pattern in which contestants wearing red win more fights,” roughly 55% of the time, based upon 2004 Olympic game outcomes, according to the authors. "Our results suggest that the evolutionary psychology of colour is likely to be a fertile field for further investigation."
However, in research for his upcoming 2010 book, Proofiness: The Dark Arts of Mathematical Deception, science writer Charles Seife of New York University reanalyzed the 2005 data and also looked at the outcomes of the 2008 Olympics to see if red-clad fighters kept to their winning ways. The result was that the number of red-clad and blue-clad fighters was about equal.
It turns out that in the 2005 study, the authors wrongly assumed red jackets were randomly assigned to Olympic athletes However, Seife found that, in the Olympics, an athlete's position within the (competitor's brackets) determines the uniform color that he wears in each round of the competition. In many tournaments, athletes are placed so that the best competitors, as determined by preliminary rounds or by another form of ranking, are unlikely to face each other in early rounds of the competition. This arrangement is not strictly random; thus, it potentially creates a bias toward one color or another. In fact, blue-jackets won more often in some 2008 Olympics wrestling competitions because of the distribution of blue jackets to winners of early rounds.
The original red versus blue study was a good example of one class of errors that come from misunderstanding randomness by seeing a pattern when there was really none to see, and assume that something is purely random when it is not.