I’m truly stuck. I’m trying to specify a relatively complicated IRT model in BRMS. I’ve been following Paul Burkner’s Bayesian IRT paper, which is very helpful from the coding perspective. However, I’m trying to write out the model to include in my research paper, and that’s proving difficult. Given the formula below, is anyone able to provide some assistance with writing out the various levels of the model equations? I’d be very grateful!
mod1 ← bf(
response ~ exp(logalpha) * eta,
eta ~ 1 + cov1 + cov2 + cov3 + cov4+ time + (1 | i | item) + (1 + year | id),
logalpha ~ 1 + (1 | i | item),
nl = TRUE
)
The code for the above runs fine, but in order to understand the various ways to sample from this model, it would be helpful to see the various linear equations that are running within it.