View article

Studies of adaptation rely increasingly on examination of traits that appear to be correlated in several taxa. Many statistical methods have been proposed to provide a probability value for the observed associations based on a presupposed model of the evolution. Flaws in these methods include inappropriate substitution of statistical uncertainty for logical uncertainty, the a priori characterization of the unknown evolutionary process under examination, axiomatic definitions of the traits themselves and inadequate representation of the unique historical relationships that are the crux of the study. Specific tests are discussed to demonstrate their unreasonable foundations and limited validity. An example is offered to show that the relationship between traits· in given taxa is better understood by direct examination of specifics than by statistical description of generalities.