Run sensitivity models for confirmatory pupil-window analyses — fit_gazepoint_pupil_window

Run a compact set of sensitivity models for confirmatory pupil-window analyses. Supported model families are the main linear mixed model, a weighted linear mixed model, a fixed-effects linear model, and a weighted fixed-effects linear model. Weighted models use the prepared valid-sample count column as weights by default.

Usage

fit_gazepoint_pupil_window_sensitivity(
  data,
  outcome_col = "pupil_model_outcome",
  subject_col = "pupil_model_subject",
  condition_col = "pupil_model_condition",
  window_col = "pupil_model_window",
  weights_col = "pupil_model_weight",
  model_types = c("lmm", "weighted_lmm", "lm", "weighted_lm"),
  include_condition = TRUE,
  include_window = TRUE,
  include_interaction = TRUE,
  random_intercept = TRUE,
  random_window_slopes = FALSE,
  fallback_on_singular = TRUE,
  REML = FALSE,
  optimizer = "bobyqa",
  maxfun = 2e+05,
  drop_missing = TRUE,
  ...
)

Arguments

data: Pupil-window model data, usually produced by prepare_gazepoint_pupil_window_model_data().
outcome_col: Outcome column.
subject_col: Subject column.
condition_col: Condition column.
window_col: Window column.
weights_col: Optional weights column.
model_types: Character vector of model types to fit. Supported values are "lmm", "weighted_lmm", "lm", and "weighted_lm".
include_condition: Logical. Include condition fixed effects when more than one condition level is available.
include_window: Logical. Include window fixed effects when more than one window level is available.
include_interaction: Logical. Include the condition-by-window interaction when both condition and window are used.
random_intercept: Logical. Include a subject random intercept for LMM model types when feasible.
random_window_slopes: Logical. Attempt subject-level random window slopes for LMM model types when feasible.
fallback_on_singular: Logical. If TRUE, LMM model types may fall back from random-window-slope models to random-intercept models when needed.
REML: Logical. Passed to lme4::lmer().
optimizer: Optimizer passed to lme4::lmerControl().
maxfun: Maximum optimizer iterations passed to lme4::lmerControl().
drop_missing: Logical. If TRUE, rows with missing or non-finite model inputs are removed before fitting.
...: Additional arguments passed to fit_gazepoint_pupil_window_lmm().

Value

A list containing fitted models, formulas, fixed effects, a comparison table, settings, and model-status information.