Monotonic issue

trial.r

# Load libraries
library(brms)
library(tidyverse)

# Load data

data <- read_csv("temp_data.csv")

# Define model

SymptomVoice_Age_f <- bf(Outcome ~ 0 + language + language : mo(Symptom) + language : mo(Symptom) : AgeS + (1 | gr(ID, by = language)))

# Define priors

SkepticPrior1 <- c(
  prior(normal(0, 1), class=b),
  prior(normal(0, .1), class=b, coef = languagedk:moSymptom),
  prior(normal(0, .1), class=b, coef = languageus:moSymptom),
  prior(normal(0, .1), class=sd)
)

# Run model

m <- brm(SymptomVoice_Age_f,
         data = data,
         family = gaussian,
         prior = SkepticPrior1,
         sample_prior = T
)

# Error:
# Error in eval(predvars, data, env) : object 'AgeS' not found

## Pseudo-Fixing it

SymptomVoice_Age_f <- bf(Outcome ~ 0 + language + language : mo(Symptom) + language : mo(Symptom) : data$AgeS + (1 | gr(ID, by = language)))


m <- brm(SymptomVoice_Age_f,
         data = data,
         family = gaussian,
         prior = SkepticPrior1,
         sample_prior = T
)

# Hypothesis testing

hypothesis(m,"languagedk:moSymptom > languagedk:moSymptom:data$AgeS", class = "bsp")

# ERROR
#Error: Some parameters cannot be found in the model: 
#  'bsp_languagedk:moSymptom:data', 'bsp_AgeS'

## Without the mo() all runs fine

# Define model

SymptomVoice_Age_f2 <- bf(Outcome ~ 0 + language + language : Symptom + language : Symptom : AgeS + (1 | gr(ID, by = language)))

# Define priors

SkepticPrior1 <- c(
  prior(normal(0, 1), class=b),
  prior(normal(0, .1), class=sd)
)

# Run model

m2 <- brm(SymptomVoice_Age_f2,
         data = data,
         family = gaussian,
         prior = SkepticPrior1,
         sample_prior = T
)