highly significantp << 0.01
Four eigenvalues of roots were extracted from the computation, and the first two discriminant roots were highly significant (Bartlett’s test of eigenvalues p << 0.01) and accounted for practically all (99.1%) of the original variance of data.