I've tried to research this question but have had to rely on answers to only somewhat similar scenarios which have led me in different directions. For example, I've been advised to calculate variance using the eigenvalues of the dissimilarity matrix, but elsewhere advised to use eigenvalues of the stress matrix. Can anyone please explain first of all whether R squared is most appropriate or if I would be better using another metric, and how exactly I can obtain variance explained in a way that is suitable for a 2D MDS generated using cmdscale in matlab? Both the dissimilarity matrix and the original data (MUAe from a 64 channel utah array) are available. Thanks.
