Bootstrap condition comparisons for Gazepoint-derived outcomes — compare_gazepoint_conditions

Compares two conditions using a transparent percentile bootstrap. The helper is intended for trial-level, participant-level, or event-locked summaries produced by gpbiometrics. When participant_col is supplied, observations are first averaged at the participant-by-condition level to reduce pseudo-replication. When paired = TRUE, only participants with both conditions are retained and resampled as paired units.

Usage

compare_gazepoint_conditions_bootstrap(
  data,
  outcome_col,
  condition_col,
  participant_col = NULL,
  condition_levels = NULL,
  paired = FALSE,
  by_cols = NULL,
  statistic = c("mean_difference", "median_difference", "standardized_mean_difference"),
  n_boot = 2000,
  conf_level = 0.95,
  seed = NULL,
  na_rm = TRUE
)

Arguments

data: Data frame containing the outcome and condition columns.
outcome_col: Numeric outcome column.
condition_col: Two-level condition column.
participant_col: Optional participant/unit identifier. If supplied, the bootstrap uses participant-by-condition means.
condition_levels: Optional two-element character vector defining the reference and target condition. The estimate is condition_levels[2] - condition_levels[1].
paired: If TRUE, perform a paired participant-level bootstrap. Requires participant_col.
by_cols: Optional columns used to run separate comparisons by subgroup.
statistic: Statistic to bootstrap: "mean_difference", "median_difference", or "standardized_mean_difference".
n_boot: Number of bootstrap resamples.
conf_level: Confidence level for percentile intervals.
seed: Optional random seed.
na_rm: If TRUE, remove rows with missing/non-finite outcomes or missing condition labels.

Value

Object of class gazepoint_bootstrap_condition_comparison, stored as a data frame. Bootstrap samples are stored in the bootstrap_samples attribute.