In: Advanced Math
In general, if p is the number of predictor variables and k is
the number of groups for the outcome variable in a DA, how many
different discriminant functions can be obtained to differentiate
among
groups? (This question assumes that the X predictor variables have
a determinant that is nonzero; that is, no individual Xi predictor
variable can be perfectly predicted from scores on one or more
other X
predictor variables.)