How to compute rCCA correlation significance

Analysis
1

The example on rCCA does not show how to compute correlation significance. Can this function be used to do that?

Thanks!

2

The answer is no, you could extract the components (canonical variates) from the rcc result and use the base stat function cor.test().

3

Thanks! is the No because of the regularization?

If so, if I don’t apply regularization because of n > p + q, i.e. setting lambda 1 and 2 as 0, should the function be the right one?

4

hi @blueskypie
Apologies for the delay in my answer. I had not seen the link of the function in your earlier message! (hence the ‘no, rcca cannot do this’). I looked at the p.asym() function. The theoretical setting seems to be for N >> P+Q (cancor() function for the CCA) so I am not sure it would work for rCCA. Since you do not need much regularisation, I would suspect the two results would look similar (CCA vs rCCA + p.asym()).

Kim-Anh