Dear R users Could someone explain me, why I need more observations than variables to conduct a discriminant analysis (as I was told by one of our teachers)? I have n = 66 observations, and 133 + 3 + 7 + 8 variables. Well, for the first 133 variables, I could conduct a PCA first, to reduce the amount of information. Thanks for an answer. Cheers Christoph -- Christoph Lehmann <christoph.lehmann at gmx.ch>