Run M3 parameter-recovery and stress evidence — multimodal_m3

Repeatedly simulates and fits M3 across pupil-signal and pupil-missingness scenarios, summarizing bias, RMSE, posterior SD, interval coverage and MCMC diagnostics. Small `reps` values are smoke tests; scientific promotion requires a predeclared larger grid.

Usage

multimodal_m3_recovery(
  reps = 3L,
  pupil_signal = c("informative", "weak", "null", "redundant", "confounded"),
  pupil_missingness = c("mcar", "quality", "device"),
  n_person = 80L,
  n_item = 10L,
  seed = 20260815L,
  fit_args = list(chains = 2L, parallel_chains = 2L, iter_warmup = 500L, iter_sampling =
    300L, refresh = 0L, init = 0)
)

Arguments

reps: Replications per design cell.
pupil_signal, pupil_missingness: Scenario vectors.
n_person, n_item: Design size.
seed: Base seed.
fit_args: Named sampling arguments for `fit_multimodal_m3()`.

Value

An `eye_multimodal_m3_recovery` object.