Summarize Bias and RMSE by Design Conditions

Compute bias and RMSE for a target parameter, grouped by design factors (e.g. condition_label, lv_cov_label, n_dyads). Use when you want to see how performance metrics vary across simulation conditions.

Usage

summarize_bias_by(
  data,
  param,
  true_param,
  ...,
  .converged_only = TRUE,
  .mcse = FALSE
)

Arguments

data

Tibble of raw results from run_bias_analysis or run_simulation (must have estimates column).

param

Parameter spec (e.g. "XA~~XB") passed to extract_param_estimates.

true_param

Column name containing true values (e.g. "lv_cov" when latent covariance varies), or a scalar for a fixed true value.

...

Bare column names to group by (e.g. condition_label, lv_cov_label, n_dyads). Uses group_by.

.converged_only

If TRUE (default), exclude non-converged rows.

.mcse

If TRUE, use simhelpers for MCSE when available.

Value

Tibble with group columns plus mean_est, bias, var, mse, rmse, n_reps.

See also

Other performance: calc_bias_rmse(), extract_param_estimates()

Examples

# After run_bias_analysis with design that has condition_label, lv_cov, n_dyads:
# summarize_bias_by(results_bias, "XA~~XB", true_param = "lv_cov",
#   condition_label, lv_cov_label, n_dyads)