Hi there,
I am trying to test Multicollinearity and singularity, the data plan I have says to use assessed from the pooled within-cells correlation, does this mean I should use pearsons correlation? If so does the pearsons correlation need to be under 80 to ensure there is no multicollinearity?
Or do I need to get the VIF output?
You haven't specified anything about the data. However I would on general principles recommend using VIFs because correlations don't consider linear combinations of variables that are often important in multicollinearity situations. Best regards, David Booth
Thanks David, I have 3 DV’s and 1 IV. do I just put the iv in the independent box and the dv in the dependent box and repeat for each of the 3 dv’s? I say this because in the linear regression box you can only put 1 DV at a time.
Hi. With only 1 IV there should be no multicollinearity problem since multicollinearity is about relationships between the IVs. Best, David
Thanks for that David. I’m just unsure of what to do because multicollinearity and singularity are data assumptions of manova and CFA.
OOps Sorry I missed that you have it backward. Lack of both multicollinearity and singularity are what you want. Best, David
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude