4 thoughts on “How much overdispersion is too much in typical GLMMs?

  1. This is very useful! I remember reading somewhere that a paper on the package/method was in preparation. Any development on this side? I use your package all the time and recommend it almost weekly to people wanting to assess model fit.


    • Hi, thanks for the kind word. About the paper … yes, it’s on my list of urgent things to do … unfortunately it’s not the only item there. But I hope to be able to put out something more formal soon.


  2. Thanks for the interesting post. Try with fewrr observations and large means. There is a stronger risk for type 1 error when the number of observations is e.g. 20 and the frequencies are high (I believe…). 250 observations largely cancel out the errors across values of the covariate?


    • Interesting point. I have made a few simulations about this, and I don’t really see anything unusual happening when varying the intercept, other than the expected result that overall power increases for larger intercepts.

      I did not understand what you mean with 250 observations cancel each other out?


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s