Avg_comparisons for brms ordinal model ususing R

Modeling
,
1

I have an ordinal model in brms-R

fit_model <-  brms::brm(Feasibility ~ 1 + cs(Valence) + (1|id) + (1|idscript),family = acat(link = "logit"),prior = prior_ma, data = DATA,warmup = 2000, iter = 4000,  seed=123, cores = C_n)

This is link to the model in rds format:
My model

I have 9 categories in Likert scale and 18048 observations data, so this is pretty big model.
This is the plot of the conditional effects:

conditional effects
conditional effects1250×622 13 KB

I would like to use marginaleffects::avg_comparisons for better understanding my effects and to get the Credible intervals for the contrast between different categories in the different valence conditions.

This is an example to results table I produce for similar model for a smaller data

exapmle.pdf (31.0 KB)

However the model is to big and my computer is lack of RAM to execute avg_comparisons for all categories in parallel. Is there a way to extract/perform marginaleffects::avg_comparisons (or different optional code) on each category separately? I’m no a coder person so I depend on a good package code.

Thanks in advance!!!

2

You could try a more limited number of draws:

avg_comparisons(model, ndraws = 500)
3

Thanks, I’m trying it now. What is the meaning of this limitation? It also limited the power of the statistical analysis?