I have generated a mixed effects logistic regression in R using the lme4::glmer() function. I've tested for overdispersion (using blmeco::dispersion_glmer()) and the estimates do not appear to be overdispersed, but what are the other assumptions that are made when using this type of model that I should test - does anyone know of a comprehensive list somewhere, especially in a format that I could cite in a scientific PhD thesis?
Asked
Active
Viewed 2,848 times
1
-
Have a look [here](https://stats.stackexchange.com/questions/185491/diagnostics-for-generalized-linear-mixed-models-specifically-residuals/187108#187108) regarding citable information on generalized mixed effects models. – Stefan Oct 30 '17 at 18:35
-
Thanks! I have read Bolker's paper but it doesn't explain much about assumption testing beyond overdispersion, other than that you should do it (presumably as the assumptions vary with the specific type of model). I have found some information of varying quality on logistic regression models, but some of the assumptions don't make sense in a mixed effects model. – Mel Oct 31 '17 at 09:49
-
Not quite sure what you are after but under the assumptions section for GLMs on [this webpage](https://onlinecourses.science.psu.edu/stat504/node/216) there are book chapters mentioned that may help you with your question. Which assumptions don't make sense in a generalized linear mixed effects model? The only assumption that could be crossed off that list is the assumption of independence since the random part in the `glmer()` model statement can take care of that, i.e. repeated measures over time or on individuals etc. – Stefan Oct 31 '17 at 18:20