Hi folks
A regression question - perhaps better placed on a different forum, but I thought I'd try here first :)
Partial correlations allow one to assess the unique contribution of a given set of predictors to the variance in Y outcome.
Let's say predictors x1, x2, x4, and x5 all stuck in a regression model accounted for 30% variance in Y outcome. Then you conduct partial correlations to assess their unique effects. Would it be possible for some of those predictors (let's say x1 and x2) to cause no unique variance? (i.e., weak and non significant correlations to Y outcome) - even though they were significant predictors in the overall model?
I think so but I am driving myself a bit mad!
Hopefully someone will be able to answer! : )