Shifted log normal distribution with varying shifts by participant

Interfaces brms
4
  1. Yes, but I don’t know if the prior is reasonable for your data. To get more info on what to set priors, use get_prior.
  2. Yes, you can use brms’s ID syntax for it and replace (1 | participant) by (1 | ID | participant), where you can replace ID by whatever you like. See the second brms paper as published in R journal for detials.
  3. the lognormal distribution has its parameters on the log scale that is, the priors are in fact on the scale of log(RT)
  4. use lognormal instead of normal if you want to use a lognormal prior. I would argue this is non-sensible here, since you model the log(mean(RT)) and not mean(RT) as per (3).
  5. See the output of get_prior for default priors.
1 Like