Correlation coefficients are widely used in comparative gerontology. H
owever, due to many pitfalls, only a poor knowledge can be gained from
these studies in many cases. For instance, a few variables gathered f
rom few species may be used, with no attention to possible confounding
variables giving spurious correlations. This article describes some o
f these problems and proposes to use more thoroughly multivariate anal
ysis in comparative gerontology. Copyright (C) 1996 Elsevier Science I
nc.