The Kluge-Kerfoot phenomenon - a statistical artifact.

Abstract

Since the 1973 paper by Kluge and Kerfoot, many studies have reported a statistically significant positive correlation between indices of the average amount of variability found within a local population for a character and the amount of differentiation among population means for that character. The statistics used have been based on coefficients of variation in order to correct for the effects of differences in scale in different characters and in different populations. Evolutionary theory has been called upon in a number of these papers to help explain this positive association (the 'Kluge-Kerfoot phenomenon'). Data sets which had been used previously were reanalyzed. There was an inverse (and usually hyperbolic) relationship between the mean of a character and both its within- and its among-population coefficients of variation. One thus expects these latter 2 variables to be correlated since they are both correlated with the same variable. The inverse relationships found in the various data sets can be explained on the basis of one or more of the following: the effect due to having a mixture of different kinds of variables in a single study, the effect of grouping data into relatively coarse classes, the effect of measurement error, or the effect of using variables that have a discrete distribution. While there may actually be a correlation between the levels within- and among-population variability, previous studies do not provide adequate data to allow an investigation of such a relationship.-from Authors