jamovi

Hi there!

I am running a Generalized Linear Mixed Model with the GAMLj module. It works fine as long as I stick to a simple model - but as soon as I include all of my six covariates I get this message:

Note. The model did not converge. Results may be misleading or uninterpretable.
Note. Model failed to converge with max|grad| = 0.00540781 (tol = 0.002, component 1)

Am I simply using too many variables for too little data, or what does this mean?
Help would be greatly appreciated!

Kind regards
Stefanie
_____________________________________________________________________________________
Here are some more infos about my model, if needed:

• My outcome is binomial, hence I am using a logistic model.
  My main IV is the timepoint (with 4 timepoints) and I also want to insert three categorical and three continuous covariates.
  I am inserting all IVs and their interaction with the timepoint as fixed effects, plus the intercept of each individual subject as random effect.
  The number of subjects varies, as some were lost to follow-up - there are 72 at timepoint 1 and 43 at timepoint 4.

Convergence of mixed models is not guaranteed in general, and for generalized mixed in particular. Reasons may vary quite a lot. It's not a jamovi problem, it's a problem related to the statistical model you're estimating. I'd suggest looking into the literature to improve the chances of your model converging. This is a good start
https://rstudio-pubs-static.s3.amazonaw ... 01d51.html

if you share the omv file, we can help you out more effectively

Great, thanks for the quick reply and resources Marcello! I'll try and see if I can get it to work