[R] Covariance matrix for first canonical variate

[Frank E Harrell Jr]

Does anyone know how to get a meaningful estimate of the covariance 
matrix for the first canonical variate coefficients for the right-hand 
(X) side of a canonical regression?  cancor returns coefficients but not 
precisions.  This would have to be subject to some constraint because of 
identifiability problems (arbitrary scaling).  I think that SPSS 
attempts to do this but seems to condition on the Y-coefficients which 
ignores a major source of uncertainty.

Thanks
Frank
-- 
Frank E Harrell Jr   Professor and Chair           School of Medicine
                      Department of Biostatistics   Vanderbilt University