Define an empirical process-validation design

The design is intentionally explicit. It records measurement conditions under which recovery, uncertainty, convergence, and failure behavior will be evaluated. It does not imply that every combination is appropriate for every estimator.

Usage

process_validation_design(
  n_persons = c(50L, 150L, 500L),
  n_trials = c(10L, 30L, 80L),
  missingness = c(0, 0.05, 0.15, 0.3),
  sampling_rate_hz = c(60, 120, 300, 1000),
  aoi_error = c("low", "moderate", "severe"),
  calibration_error = c(0, 0.5, 1),
  pupil_dropout = c(0, 0.1, 0.3),
  heterogeneity = c("low", "moderate"),
  model_misspecification = c(FALSE, TRUE),
  replications = 100L,
  seed = 1L,
  label = "process_validation"
)

Arguments

n_persons

Participant counts.

n_trials

Trial/item counts per participant.

missingness

Proportion of generic process observations made missing.

sampling_rate_hz

Nominal sampling rates.

aoi_error

AOI uncertainty regimes.

calibration_error

Calibration-error regimes in user-defined units.

pupil_dropout

Pupil-specific dropout proportions.

heterogeneity

Participant-heterogeneity regimes.

model_misspecification

Logical regimes indicating deliberate mismatch.

replications

Monte Carlo replications per condition.

seed

Master simulation seed.

label

Optional design label.

Value

An `eye_process_validation_design` object.