2
$\begingroup$

I am rather new to R. I am trying to run a GLMM - binomial logit.

I have three independent variables (x1, x2, x3) and a dependent variable (y) - all numeric.

m <-glmer(y ~ x1:x2:x3 + (1 | participant), data=mydata, family=binomial)
  1. How can I check for the model's assumptions?
  2. Which model can be appropriate in case the assumptions are not met?

Your help is much appreciated!

$\endgroup$
4
$\begingroup$

For a binomial glmm, the main assumptions are:

  • the outcome/response is binary. You said the variables including y are "all numeric"
  • the random effects are approximately normally distrubuted. The main thing here is that you have sufficient number of participants for the software to reliably estimate a variance.
| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Does this answer your question ? If so, please consider marking it as accepted. Thanks ! $\endgroup$ – Robert Long Jul 19 at 4:47
0
$\begingroup$

In terms of model misspecification and the assumptions regarding the distribution of the residuals, GLMM is a bit problematic compared to LMM because in the binomial case the residuals are stacked on 0 and 1. The DHARMa package in R offers the solution by using a simulation-based approach to check for the GLMM residual distribution and plot tests for normality, uniformity, and dispersion. Check it out https://cran.r-project.org/web/packages/DHARMa/vignettes/DHARMa.html For model comparison, you can use bootstrapped likelihood values and choose the best model accordingly https://rpubs.com/hughes/22059.

Also, for more info on GLMM, I advise you to see https://bbolker.github.io/mixedmodels-misc/ecostats_chap.html

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.