Multivariate Regression vs Canonical Correlation

I know the difference between correlation and regression, however, I'm a little confused with the difference between Canonical Correlation (CCR) and MultiVARIATE Regression (MVR) (>1 DV). Seems like little info is available about the usage of MVR and a lot more available for CCR. Seems also that CCR is used for the same purpose I thought MVR should be used. I'm not too famliier with either, though I can execute and interpret both in SPSS (cancorr macro and MANOVA syntax for CCR, and Multivariate GLM for MVR).

Can someone please explain to me briefly the practical application in marketing research of CCR and MVR so I better understand which to use, when. Which procedure (CCR or MVR) is most common in marketing research (and I mean for practitioners and not for academic work).

Thanks soooo much


Super Moderator
Multivaraite regression (sometimes reffered to as redundancy analysis) is an extension of multiple regression when you have >2 DVs and >2 IVs. It is a more fomarl analysis as it produces f-values and p-values that help determine the signifiance of relationships between your variables. CCA is also a multivariate tool, allowing for all sorts of data types. It is less formal in that there are fewer assumptions. It uses two data matraces to produce of bi-plots or triplots that display these assocaitions. Overlaying vectors show the strength and relatedness of these associations.