Fit AOI-window model-family sensitivity checks — fit_gazepoint_aoi_model

Fit a compact set of sensitivity models for AOI-window outcomes. The main model is a binomial GLMM. Additional checks can include an empirical-logit LMM, a weighted proportion LMM, and a fixed-effects quasibinomial GLM.

Usage

fit_gazepoint_aoi_model_sensitivity(
  data,
  success_col = "aoi_glmm_success",
  failure_col = "aoi_glmm_failure",
  denominator_col = "aoi_glmm_denominator",
  proportion_col = "aoi_glmm_prop",
  subject_col = "aoi_glmm_subject",
  condition_col = "aoi_glmm_condition",
  window_col = "aoi_glmm_window",
  model_types = c("binomial_glmm", "empirical_logit_lmm", "proportion_lmm",
    "quasibinomial_glm"),
  include_condition = TRUE,
  include_window = TRUE,
  include_interaction = TRUE,
  random_intercept = TRUE,
  optimizer = "bobyqa",
  maxfun = 2e+05,
  nAGQ = 0,
  empirical_logit_correction = 0.5,
  drop_missing = TRUE
)

Arguments

data: AOI GLMM data returned by prepare_gazepoint_aoi_glmm_data().
success_col: Success-count column.
failure_col: Failure-count column.
denominator_col: Denominator column.
proportion_col: Proportion column.
subject_col: Subject column.
condition_col: Condition column.
window_col: Window column.
model_types: Character vector of model types. Supported values are "binomial_glmm", "empirical_logit_lmm", "proportion_lmm", and "quasibinomial_glm".
include_condition: Logical. Include condition fixed effect when possible.
include_window: Logical. Include window fixed effect when possible.
include_interaction: Logical. Include condition-by-window interaction when both condition and window are included.
random_intercept: Logical. Include subject random intercept in mixed sensitivity models.
optimizer: Optimizer for lme4 mixed models.
maxfun: Maximum optimizer evaluations.
nAGQ: Number of adaptive Gauss-Hermite quadrature points for the binomial GLMM.
empirical_logit_correction: Small correction added to success and failure counts for empirical-logit models.
drop_missing: Logical. Drop rows with missing model variables.

Value

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