So you have a model and you want to check for MC using R?
Hej,
I am checking for multicollinearity. And I cant find on the web when having the outpur from R commander, on which I have to look to identify collinearity?
-GVIF- or -df- or -GVIF^2(1/(2*Df))-
Thanks for the help.
So you have a model and you want to check for MC using R?
Stop cowardice, ban guns!
yes, using R commander. I can do under 'numerical diagnostics' with Variance Inflation Factor 'test'. One output is as follows:
picupload
Which one do I have to interpret for mc?
GVIF or df or GVIF^(1/(2*Df))
What is the cut-off for mc? I read 5, but also heard 12...
First off what does the G in GVIF stand for? DF is likely degree of freedom, which you would not interpret directly for MC.
Is R commander a package? If so, what accompanying documentation does it have with it?
Stop cowardice, ban guns!
Sorry, I am not that knowledgeable about documentation etc.. We use that package in my course.
https://cran.r-project.org/web/packa...mdr/index.html
http://socserv.socsci.mcmaster.ca/jfox/Misc/Rcmdr/
There seems to be an answer here: http://stats.stackexchange.com/quest...if/96584#96584
But I still not understand which one I have to choose...
My predictors are all ordinal or nominal (categorical) and my outcome is numerical.
Here is the package I am using with the explanation. Unfortunately, I do not understand it...
http://www.inside-r.org/packages/cran/car/docs/vif
Need help!
hi,
this is a fairly complex question, see a detailed answer here: http://stats.stackexchange.com/quest...if-or-textgvif
The bottom line is imo that the best would be to square the corrected GVIF value and apply the rules for VIF . There the accepted threshold would be 5 .
regards
MasterStudent (03-14-2016)
I hoped it would be easy...
Which one is the 'corrected' GVIF?
You suggest to square the 'GVIF' myself (Not using the 'GVIF1/(2⋅df)')?
hi,
it would be the GVIF corrected by the degrees of freedom.
regrds
Fox & Monette (original citation for GVIF, GVIF^1/2df) suggest taking GVIF to the power of 1/2df makes the value of the GVIF comparable across different number of parameters. "It is analagous to taking the square root of the usual variance-inflation factor" ( from An R and S-Plus Companion to Applied Regression by John Fox). So yes, squaring it and applying the usual VIF "rule of thumb" seems reasonable.
see the R "car" package p. 157:
If all terms in an unweighted linear model have 1 df, then the usual variance-inflation factors are
calculated.
If any terms in an unweighted linear model have more than 1 df, then generalized variance-inflation
factors (Fox and Monette, 1992) are calculated. These are interpretable as the inflation in size of
the confidence ellipse or ellipsoid for the coefficients of the term in comparison with what would
be obtained for orthogonal data.
The generalized vifs are invariant with respect to the coding of the terms in the model (as long as
the subspace of the columns of the model matrix pertaining to each term is invariant). To adjust for
the dimension of the confidence ellipsoid, the function also prints GV IF1/(2×df) where df is the
degrees of freedom associated with the term.[/SIZE][/SIZE]
Tweet |