behaviour pls
jcrofts
Posted on 10/08/08 09:38:40
Number of posts: 29
j crofts
Hi, Just suppose I have to different behavioural measures. What is the difference between running two seperate PLS analyses (one for each behaviour) and running a single PLS analysis where my behavioural data is now a (2*n) matrix? They are surely not the same; are they?
cheers, Jonathan
Untitled Post
rmcintosh
Posted on 10/08/08 12:08:18
Number of posts: 394
You are right that they are not the same thing. I can give you the mathematical reason why, but let me provide a more "conceptual" illustration. If you have two behaviors done independently, those are simple two dimensional analyses (assuming one task as well). By putting both together you allow for an expansion of the possible dimensions allowing you to identify analytically where the space occupied by the two brain-behavior measures overlaps, and where they do not. By doing them separately, you can do this exercise only qualitatively.