Hi,

I have completed my first analysis and write-up of some clinical trial data using brms - was hoping to get some experienced eyes to look at it. My model which has only categorical population-level predictors is specified as:

```
# Specify priors and run Stan model -------------------------------------------
prior1 <- c(prior(normal(0, 10), class = b),
prior(normal(30, 5), class = b, coef = "Intercept"))
model1 <- brm(formula = MADRS ~ 0+ Intercept + Group+Time+ Group:Time + (1|ID),
family = gaussian,
prior = prior1,
data = dressMADRS,
chains = 3, cores = 3,warmup = 500, iter = 10000,
sample_prior = TRUE);
```

In my write-up I have defined the model as:

y \sim \operatorname{Normal}(\mu, \sigma_{t})

\mu = \beta_{0} + s_j + (\overrightarrow{\beta_{Time}} \times \overrightarrow{X_{Time}}) + (\overrightarrow{\beta_{Group}} \times \overrightarrow{X_{Group}}) + (\overrightarrow{\beta_{Int}} \times \overrightarrow{X_{Int}})

\overrightarrow{\beta_{Age}},\overrightarrow{\beta_{Time}},\overrightarrow{\beta_{Int}} \sim \operatorname{Normal}(0, 10)

\beta_{0} \sim \operatorname{Normal}(30, 5)

s,\sigma_{t},sd(s) \sim \operatorname{Half Student}(3,0,mad(y))

with the following text description:

Overarrows are used to indicate dummy coding of categorical fixed effect variables and sj

corresponds to the random efffects of participant j. Relatively non-committal priors are used. For the overall intercept a normal prior with mean of 30 with sd of 10 gives 95% coverage of the range 20-40. 20 is the minimum entry criteria for the study while a patient with score > 40 would typically be excluded. For fixed effects, normal priors are used with an sd of 10 which allows for a range of effect sizes of size comparable to the literature. sd priors are specified as default half student t priors as suggested by Gelman (2006) using the median absolute deviation (mad) of y as the scale factor (BĂĽrkner (2017)). For MCMC modelling Stan is acccessed through the brms interface (BĂĽrkner (2017)). For computing Bayes Factors, a ROPE of range |sd/10| is used (Kruschke and Liddell (2018)).

With thanks in advance!