Pattern in RE-grouped residual for binomial GLMM

[florianhartig]

Question via email

I'm using the DHARMa package to check the residuals of my binomial (0/1) generalized additive models for my master thesis project, and I'm having some difficulties interpreting the results. When I run the main plots, I obtain these results:

I followed what is suggested in the manual and tried to group the residuals. I have 4 natural grouping variables: 2 RE and 2 temporal (year and month), and my model include all these 4 variables and some other continuous predictors. If I understood it correctly, when I group (e.g by one of the RE) the plots indicate some underdispersion? My main doubts is related with the plot on the right (residuals vs. predictions). What does it indicate about the model? Is these residual pattern concerning?