I am analysing data and using liner regression and also multiple logistic regression, but in contrast to literature I don't see the proper results. I don't know how much it might be a cause of multicollinearity. (VIF for predictor variables is around 1.3). Do I need to handle it or it is not that high? one dependent variable has 1.8 skewness in 328 subjects, might it be the problematic case? thanks a lot
As for the VIF is concerned, it should be below a threshold value of 10. You'd better also look at the tolerance values. If you are using SAS, then use /tol after the /VIF parameter. There are the same assumptions about this. Then there is the index score you have to look at. It should be too high >500.
May be a closer look may solve the problem
Hope it helps.
There is no standard about threshold of VIF.
I think it is 20. So, there is no collinearity in your data.
If you send me Excel data without virus, I will check it.
You mention me I will open the result on RG or not.
You check my papers about CPD data that has three strong collinearities.
I am very interest in collinearity.
- Badges
- Science topic