conditional effect

library(rethinking)
library(brms)
options(mc.cores = 4)

# Import data
data(Trolley)
data <- Trolley

# Subset some data to make process faster
data <- subset(data, edu == "Bachelor's Degree" |  edu == "Some College"|  edu == "Master's Degree")
data$edu <- factor(data$edu)
levels(data$edu) <- c("bsc","msc","col")

# Make order and response into fewer categories
data$order <- round(data$order/6)
data$response <- round(data$response/2.4)+1
data$response <- ordered(data$response)

# Make plot as wished for
# Get usage for each dur per participant
edus <- c("col","bsc","msc")
cols <- brewer.pal(3, "Set2")
par(mfrow=c(2,2))
j <- 1
for(j in 1:4){
  resp.trolley <- data %>%
    group_by(order, edu) %>%
    summarise(
      resp = mean(response==j))

  # plot raw
  test <- subset(resp.trolley, edu==edus[1])
  plot(jitter(test$order, 0.4),test$resp,lwd=3, pch=20, ylim=c(0,max(resp.trolley$resp)*1),
       col=cols[1],xlab="Duration", ylab = paste("Proportion Rating",j))
  lines(test$order,test$resp,
        col=cols[1],lwd=0.5, lty=2)
  if (j == 3) {
    # Add legend
    legend(x = "bottomright",          # Position
           legend = c("Col", "Bsc","Msc"),  # Legend texts
           lty = c(2),           # Line types
           col = cols[1:3],           # Line colors
           lwd = 1)                 # Line width
  }
  for(i in 2:3){
    test <- subset(resp.trolley, edu==edus[i])
    points(jitter(test$order, 0.4),test$resp,ylim=c(0,0.8),lwd=3, pch=20,
           col=cols[i])
    lines(jitter(test$order, 0.4),test$resp,ylim=c(0,0.8),
          col=cols[i],lwd=0.5, lty=2)}
}

# Make BRMS model
data$response <- ordered(data$response)
m1 <- brm(response  ~ edu*mo(order) + (1|id), data= data,
          family = cumulative(threshold = "flexible"),
          chains = 2)

# Plot
conditional_effects(m1, categorical=T)

# Trying to make for conditions
conditions <- make_conditions(m1, "response")
conditional_effects(m1,categorical=TRUE, conditions = conditions)