Random / varying effects are labelled incorrectly using tidy

[richramsey]

I am new to broom.mixed, so this may be an elementary mistake on my part. I want to plot the random effects / varying effects as caterpillar plots following a model run in brms. However, when I get the output of tidy, the varying effects for intercepts are labelled correctly, but the varying effects for slopes are not. Specifically, the 'level' and 'term' columns are incorrect for the random slopes. The levels for participant should be the same for all varying effects of interest, but once it goes beyond the intercept things seem to get messed up. Model and table below. Item has 80 levels, participant has 11 levels. Any ideas on the problem/solution?

brms: 2.11.6 broom.mixed: 0.2.4

b4 <- brm(rating ~ 1 + painting_type_dev * motion_dev * load_dev +
          (1 + painting_type_dev * motion_dev * load_dev|participant) +
          (1|item),
        data = data_aes_dev, family = cumulative("probit"),
        prior = c(prior(normal(0, 5), class = Intercept),
                prior(cauchy(0, 2), class = b)),
        iter = 6000, warmup = 1000, cores = 4, chains = 4,
        control = list(adapt_delta = 0.99),
        inits = 0)

tidy(b4, effects = "ran_vals", robust=T, conf.method = "HPDinterval")
head(tidy_b4_ran_vals)

ran_vals.pdf

[bbolker]

Not sure yet, will look into it ... A reproducible example would help a lot.

[richramsey]

Hi, thanks for your response. I found the problem. It was the way I was labelling factors. Once I removed "_" from factor labels, everything seemed to work fine. Thanks again for repsonding. Rich

[bbolker]

thanks, you saved me a bunch of time. I should (a) document this limitation and (b) check names and give a warning if they contain "_" (if possible).

[richramsey]

Yes, that sounds good. Thanks. Fantastic package by the way.