Simulate M4 multimodal sequential measurement data — simulate_multimodal

Generates deterministic synthetic response, RT, gaze, pupil, nuisance, and ordered latent-state data for software validation and methodological stress testing. Synthetic state labels are known truth for recovery only and are not psychological constructs.

Usage

simulate_multimodal_m4(
  n_person = 80L,
  n_item = 12L,
  n_session = 1L,
  n_states = 2L,
  scenario = c("clear", "weak", "null", "persistent", "rapid_switch",
    "trait_conditioned", "rt_redundant", "gaze_redundant", "pupil_redundant",
    "nuisance_confounded", "device_confounded"),
  missingness = c("none", "mcar", "quality", "gaze", "pupil_quality", "device",
    "state_dependent"),
  missing_rate = 0.08,
  seed = 20260820L
)

Arguments

n_person, n_item: Number of persons and total item trials per person.
n_session: Number of non-overlapping ordered sessions per person.
n_states: True latent-state count. May be 1 through 4.
scenario: State/data-generating scenario: `clear`, `weak`, `null`, `persistent`, `rapid_switch`, `trait_conditioned`, `rt_redundant`, `gaze_redundant`, `pupil_redundant`, `nuisance_confounded`, or `device_confounded`.
missingness: Missingness stress mechanism.
missing_rate: Base channel-missingness probability.
seed: Deterministic random seed.

Value

An `eye_multimodal_m4_simulation` containing `data`, full generating `truth`, scenario metadata, and interpretation boundary.