Skip to content

vif using GLMMadaptive

1 message · John Fox

#
Dear Grisenia,
I guess that you're using the vif() function in the car package (there are, I believe, other implementations as well). The reference to RStudio is irrelevant -- you'd almost surely get the same error regardless of the which programming editor you use with R.

Aliased coefficients imply that there's either perfect collinearity among the regressors in your model or a flat likelihood at the maxium, making some coefficients unidentified. In this situation, at least some of the VIFs are infinite.

In addition, the meaning of VIFs in models with interactions is ambiguous --- subject to variation depending upon inessential changes to the model, such as how contrasts for factors like sex are defined. To understand the problem, see the Fox and Monette paper referenced in ?vif. 

I hope this helps,
 John

--------------------------------------
John Fox, Professor Emeritus
McMaster University
Hamilton, Ontario, Canada
Web: socialsciences.mcmaster.ca/jfox/