problem with interaction in lmer even after creating an "interaction variable"
a_lampei <anna.lampei-bucharova <at> uni-tuebingen.de> writes:
Thank you very much, First, sorry for posting on wrong mailing list, I did not know that there exists a special one for lmer. Yes, there are collinearities in the data. Still, I would like to have the variables in one model to compare explained variability. Is there some option, or it is simply impossible? Thank you, Anna
The development version of lme4 has an (experimental) feature that automatically removes collinear columns of the model matrix; you could try that. Further discussion on r-sig-mixed-models ... Ben Bolker