Page 1 of 1

detecting multicollinearity

Posted: Tue Jun 29, 2010 7:46 am
by mscarlatos
I am unable to pull up Eviews Help anymore for some reason. Could someone tell me whether or not Eviews can run a collinearity test between two or more variables?

Re: detecting multicollinearity

Posted: Tue Jun 29, 2010 7:52 am
by EViews Gareth
From an estimated equation there are a few diagnostics to help you determine collinearity between the regressors - Variance Inflation Factors, and Variance Decomposition.

Re: detecting multicollinearity

Posted: Tue Jun 29, 2010 10:22 am
by mscarlatos
Dear Mr Evans: Thank you but could you be a bit more forthcoming: I run a regression but where do I look for VIF and Variance Decomposition? I ask since I cannot find them under any of the Views or Procs. Thank you.

Re: detecting multicollinearity

Posted: Tue Jun 29, 2010 10:26 am
by EViews Gareth
View->Coefficient Diagnostics.

Re: detecting multicollinearity

Posted: Tue Jun 29, 2010 12:06 pm
by mscarlatos
Under view, I see "coefficient tests" under which I do not see anythign measnuring multicollinearity unless "redundant variables" measures that. However, without even regressing, I find under "quick"..."group statistics"..."correlations." Does the latter serve as a measure of collinearity?

Re: detecting multicollinearity

Posted: Tue Jun 29, 2010 12:33 pm
by startz
What version of EViews are you using?

Re: detecting multicollinearity

Posted: Tue Jun 29, 2010 12:34 pm
by mscarlatos
Eviews6

Re: detecting multicollinearity

Posted: Tue Jun 29, 2010 12:48 pm
by startz
The multicollinearity diagnostics were introduced in EViews 7.

Just out of curiousity, why are you concerned about multicollinearity?