My data is structured in two nominal factors (group: 2 levels (between factor - 30 subjects for each level); perspective: 4 levels (within factor - 30 items for each level ) and 9-points Likart scale for ratings (7 scales for 7 different factors).
Because my data did no satisfy the proportional odds assumption I choose Adjacent-category model analysis for ordinal data (is it good idea?). However I straggle with the plots.
This is my model:
fit_model2 <- brm(feasibility ~ group + perspective + group:perspective + (1|participant) + (1|id_script),family = acat(link = "logit"), prior = prior_ma, data = wideDF, warmup = 2000, iter = 4000, seed=123)
I tried this code for the plot:
p3<-conditional_effects(fit_model2, c(“group”, “group:perspective”, “perspective”))
but my Y-axis is the 1-9 scale (similar to standard Anova plots) and I get this Warning messages:
1: Predictions are treated as continuous variables in ‘conditional_effects’ by default which is likely invalid for ordinal families. Please set ‘categorical’ to TRUE.
So I tried this one:
p2<-conditional_effects(fit_model2, categorical=TRUE )
and got -
Interactions cannot be plotted directly if ‘categorical’ is TRUE. Please use argument ‘conditions’ instead
this is the plot:
But it is for both groups and I need separate plots