Distinction between nAGQ=0 and nAGQ=1 in glmer
I received a question about the algorithm used in glmer with nAGQ=0. I find it is easiest to show the calculations through an example, which I wrote up as a Jupyter (https://jupyter.org) notebook nAGQ.ipynb in the github repository https://github.com/dmbates/MixedModelsinJulia/ As the name of the repository indicates, this uses the MixedModels package for Julia but the methods are similar to those in glmer.