1
$\begingroup$

Seeking an R package capable of modeling ordinal data including random effects (like clmm) while relaxing the proportional odds assumption. Although clmm handles random effects, it does not allow relaxation of the proportional odds assumption. Any recommendations for alternative packages would be greatly appreciated.

$\endgroup$

1 Answer 1

1
$\begingroup$

Random effects work better with Bayesian models. The rmsb package uses Stan to fit PO models with or without random effects (single level) with non-proportional odds for a subset of the variables (partial proportional odds model, Peterson & Harrell 1990). It also fits a constrained partial proportional odds model (Peterson & Harrell 1990). An extensive vignette is at https://hbiostat.org/

Note that in some situations relaxing the PO assumption makes things worse. See this. For other resources see this.

$\endgroup$

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