I had a look in the stan-files, but I can’t really pretend that I understand them.
It is OK to just copy, shows that you can get it working using Rstan. I would have revised their code a bit to get rid of their uniforrn (and stupid) priors, just a few lines extra in their stan-code
gaussian and cauchy
Which makes sense, since Cauchy is the same as T(df = 1), so heavier tail than T(df = 2)
If we write this out into a nice model notation, what does it look like? Something like this. Reactioni∼Normal(μ,σ)μi=α+αsubject+βDaysiα∼Normal(245.50)αsubject∼Normal(0,30)for j = 1, 2, 3...18β∼Normal(20,50)σ∼Normal(20,50) \begin{aligned} \text{Reaction}_{i} & \sim \mathrm{Normal}(\mu, \sigma) \\ \mu_i & = \alpha + \alpha_\text{subject} + \beta\text{Days}_i\\ \alpha & \sim \mathrm{Normal}(245. 50)\\ \alpha_\text{subject} & \sim \mathrm{Normal}(0, 30) & \text{for j = 1, 2, 3...18}\\ \beta & \sim \mathrm{Normal}(20,50)\\ \sigma &\sim \mathrm{Normal}(20,50)\\ \end{aligned} Reactioniμiααsubjectβσ∼Normal(μ,σ)=α+αsubject+βDaysi∼Normal(245.50)∼Normal(0,30)∼Normal(20,50)∼Normal(20,50)
for subject = 1, 2, ..., 18
To add a comment, just highlight the relevant section of the text
It works, great!