Estimate latent mean centering model with brms about funky-firefly HOT 2 CLOSED

The issue here seems to be a lack of sampling model for "x". This could help: paul-buerkner/brms#888 (comment)

from funky-firefly.

Univariate latent means model is simple with brms.

Multivariate latent means models can be specified by 1. Stacking data wrt. outcome and 2. using indicator variables for outcomes to turn parameters on and off. An example sketch of Model 4 in http://dx.doi.org/10.1037/met0000250 would be to first stack the data thuswise:

d2 <- d |> 
  pivot_longer(c(urge, dep)) |> 
  mutate(
    is_urge = if_else(name == "urge", 1, 0),
    is_dep = if_else(name == "dep", 1, 0)
    )
d2 <- d2 |> 
  left_join(d)

Then estimate this model

bform <- bf(
  value ~ 
    is_urge * alpha + 
    is_dep * depb +
    is_urge * (phi * (urge_lag - alpha)) + 
    is_urge * (beta * (dep - depb)),
  alpha + phi + beta + depb ~ (1 || person),
  nl = TRUE
)

I've confirmed that this gives exactly the same posterior summaries as in their MPlus estimation. Now on to making progress with the rest...

from funky-firefly.