Abstract
A Q-technique for evaluating canonical variates is described. It is shown to have computational, and statistical, advantages over the usual R-technique.

This publication has 1 reference indexed in Scilit: