Is collinearity of the independent fixed effects (covariates, so not the variables I'm interested in) a problem when you create a linear mixed model? Is there a good resource I can point to to support a claim that is or isn't a problem?
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
1
The difference between a linear mixed effects model and a linear model is that the former has random effects and the latter does not. Both have fixed effects and the same considerations apply regarding collinearity. If you have perfect collinearity then you won't be able to fit either type of model because the model matrix for fixed effects will be singular. If you have just high correlations among covariates them you may be able to fit the model but just as with linear models there can be problems with linear models such as stability and large standard errors. It would be worth considering dropping variables completely or applying dimension reduction techniques such as principal components analysis.
answered Jun 26, 2020 at 7:43
-
$\begingroup$ @Henke if this answers your question please consider marking it as the accepted answer $\endgroup$ Commented Jul 6, 2020 at 5:01