Hi, all.
I got a simple question of multiple regression results.
I want to run multiple regression analysis of VBM results with many kinds of covariates.
First time, I run the analysis with six covariates and got six beta images of each covariates.
I checked those and found that all six beta images were exactly same.
So, I replaced one covariate with another covariate.
Then I got totally different beta images.
Here is my real question.
I searched several documents about multiple regression analysis in SPM and found that those covariates should be independent each other.
I think that the reason why I got wrong results is some covariates were not independent each other.
Am I correct?
Actually, I have run correlation analysis across those covariates and among the first six covariates, one covariate was highly correlated with other five covariates.
|