Simulate from the M2 response + RT + gaze generative model

Simulates the same level-1 likelihood used by [fit_multimodal_m2()] and retains all person/item hyperparameters and realized latent parameters as truth. Channel-specific dropout is applied only after the complete generative data have been created.

Usage

simulate_multimodal_m2(
  n_person = 100L,
  n_item = 10L,
  mu_item = c(difficulty = 0, time_intensity = 4, gaze_intensity = 3.5),
  sd_person = c(ability = 1, speed = 0.5, gaze_process = 0.5),
  cor_person = matrix(c(1, 0.3, -0.3, 0.3, 1, -0.25, -0.3, -0.25, 1), 3L, 3L, byrow =
    TRUE),
  sd_item = c(difficulty = 0.75, time_intensity = 0.35, gaze_intensity = 0.6),
  cor_item = matrix(c(1, 0.25, 0.2, 0.25, 1, 0.3, 0.2, 0.3, 1), 3L, 3L, byrow = TRUE),
  nu_range = c(0.5, 0.8),
  gaze_shape = c(shape = 2, scale = 6),
  dropout = c(response = 0, rt = 0, gaze = 0),
  seed = 20260814L
)

Arguments

n_person

Number of persons.

n_item

Number of items.

mu_item

Means for item difficulty, log-time intensity, and log-gaze intensity.

sd_person

Person-side standard deviations for ability, speed, and gaze-process propensity.

cor_person

Person-side correlation matrix.

sd_item

Item-side standard deviations.

cor_item

Item-side correlation matrix.

nu_range

Uniform range for item time-discrimination parameters.

gaze_shape

Shape/scale parameters for inverse-gamma generation of negative-binomial shape parameters.

dropout

Named probabilities for response, RT, and gaze missingness.

seed

Reproducibility seed.

Value

An `eye_multimodal_m2_simulation`.