2
$\begingroup$

I have an experimental design with a continous dependent variable (DV), and two factors A and B, factor A has 3 levels and factor B has 2 levels. I checked my DV distribution with the function descdist(DV), from the fitdistrplus package, and my DV is distributed as a uniform distribution. In fact when I fitted my data into a uniform distribution with the function fitdist(DV, distr="unif", method="mme"), I found a perfect overlap.

My purpose was to run a Generalized Mixed Model, but given that uniform family distribution is not supported automatically in the families of the glmer function, I don't know how to fit properly the model so that it could take in account the uniform distribution of my DV.

Have you any suggestion?

Thanks

$\endgroup$
1
  • $\begingroup$ When fitting a model, you are interested in the distribution of your DV conditional on the regressors. In practice, this mean you should check the distribution of the model residuals. Or in case of all regressors being categorical, you can also check the distributions stratified by groups. $\endgroup$ – Roland Nov 12 '18 at 8:45
2
$\begingroup$

You can have a look at the GLMMadaptive package that can fit a Beta mixed effects model. Given that the uniform distribution is a special case of the Beta distribution, this should similarly provide a good fit. For more information on how to fit a Beta mixed model, check the vignette Custom Models.

$\endgroup$

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.