Skip to contents

Run Bias Analysis

Source: R/run_bias.R
run_bias_analysis.Rd

Convenience wrapper for bias/RMSE/coverage designs. The design's analysis_model can include "ICC_LMM" to compare ICC estimates (from dyLMM) alongside dySEM-based models; ICC is mapped to XA~~XB for bias and coverage. ICC_LMM requires non-negative lv_cov in the population values.

Usage

run_bias_analysis(
  design,
  true_values = NULL,
  true_param_map = NULL,
  extract = c("estimates", "convergence", "coverage"),
  reps = 100,
  parallel = FALSE,
  workers = NULL,
  save_dir = NULL
)

Arguments

design

Design from create_design(); include true_values for coverage.

true_values

Named list of true parameter values (fixed across scenarios). Ignored when true_param_map is provided.

true_param_map

Named list mapping parameter labels to design column names. Use when true values vary by scenario (e.g. list("XA~~XB" = "lv_cov")). Each scenario gets list(param = design[i, col]) for coverage and bias.

extract

Character vector: "estimates", "convergence", "coverage". For bias/RMSE only (faster), use c("estimates", "convergence") to skip CI computation via performance::icc; add "coverage" when coverage is needed.

reps, parallel, workers, save_dir

Passed to run_simulation.

Value

Tibble with raw results; use summarize_bias_by(), calc_coverage() for summaries.

See also

Other simulation: run_one_scenario(), run_power_analysis()