Let's do a thread on generalized linear mixed models. I have a meeting at 1, so I'll hammer out everything I can in the next 29 minutes! (1/)

I like to build from what seems to be the "safe spot" for a lot of people: the linear model.

When you do frequentist statistical inference (e.g. hypothesis tests, confidence intervals), you assume the responses are independent, normal, heteroscedastic (equal variance). (2/)

Of course, rules are meant to be broken. So what happens if we want to loosen up these assumptions? You get the generalized linear model (GLM), the linear mixed model (LMM), and the generalized linear mixed model (GLMM). (3/)

Quick aside: the big citation you'll see associated with LMM is Stiratelli, Laird, and Ware. Fun fact: Laird is actually NAN Laird, meaning woman (not "non a number"). Yay women in statistics! (3.5/)