stats_resample_dm: Resample Observed and Predicted Statistics for Interval Estimation

Description

Internal methods to generate bootstrap-like intervals for descriptive statistics derived from either observed data or model predictions. These methods support both drift_dm objects and data.frames containing a single participant's observed data.

Usage

stats_resample_dm(object, conds, type, b_coding, ..., R, interval_level)
# S3 method for drift_dm
stats_resample_dm(
  object,
  conds,
  type,
  b_coding,
  ...,
  R = 100,
  interval_level = 0.95,
  n_sim = NULL
)
# S3 method for data.frame
stats_resample_dm(
  object,
  conds,
  type,
  b_coding,
  ...,
  R = 100,
  interval_level = 0.95,
  progress = 0,
  level
)
# S3 method for fits_ids_dm
stats_resample_dm(
  object,
  conds,
  type,
  b_coding,
  ...,
  R = 100,
  interval_level = 0.95,
  progress = 0,
  level
)

Value

A stats_dm object with added column Estimate indicating whether the row represents the lower interval bound, the original value ("orig"), or the upper interval bound. The interval level can be controlled via the interval_level argument.

Arguments

object: a drift_dm object (for model-based resampling) or a data.frame with observed data for a single participant. drift_dm_stats_types("sum_dists"), such as "quantiles" or "cafs".
conds: a character vector indicating the condition(s) for which the statistics should be resampled.
type: a character string, specifying the type of statistic to calculate
b_coding: a list, specifying the boundary coding, required when calculating the statistics.
...: additional arguments passed to stats_resample_wrapper() and simulate_data(). Must contain type and b_coding
R: an integer, number of replications (default is 100).
interval_level: a numeric between 0 and 1, controlling the width of the interval (default is 0.95).
n_sim: an optional vector, providing the trial numbers for simulating synthetic data under the model. Only relevant when no observed data is attached to the model via the obs_data entry of the model.
progress: an integer, specifying if a progress bar shall be displayed (1) or not (0).
level: a character string, specifying at which level resampling shall take place. "individual" will lead to resampling of an individual's data. "group" will lead to resampling of the entire participant.

Details

The stats_resample_dm() generic dispatches to class-specific methods. For drift_dm objects, it generates synthetic data sets under the model. For raw data, it resamples observations with replacement (i.e, it performs a bootstrap). In both cases, statistics from the resampled/generated data are used to compute intervals for the requested statistic.

Resampling is done for each condition separately.

The function stats_resample_dm() is called within calc_stats().