Configure a reproducible audit

Usage

audit_config(
  membership = c("paired", "available"),
  weighted = FALSE,
  baseline_predictors = character(),
  annotation_status = c("accepted", "adjudicated"),
  probability_sample = FALSE,
  confidence = 0.95,
  seed = 104729L,
  permutations = 999L,
  cluster = NULL,
  test_families = NULL
)

Arguments

membership: "paired" (default) or "available" wave membership.
weighted: Use supplied participant weights for descriptive outcomes and sample benchmarks. Dialogue metrics retain observed speaker weights.
baseline_predictors: Explicit pre-deliberation columns used in selection and assignment diagnostics. Baseline items use item:episode_id:item_id.
annotation_status: Included review statuses. Defaults to accepted and adjudicated; rejected annotations cannot be included.
probability_sample: Explicit declaration enabling sampling inference.
confidence: Confidence level for supported intervals.
seed: Reproducible seed, without changing the caller's random state.
permutations: Monte Carlo draws for supported assignment tests.
cluster: Optional specification from cluster_inference().
test_families: Named list of metric-name vectors. Each family includes every matching metric row across events; Holm adjustment retains raw p-values.

Value

A configuration list recorded in audit_dp() output.

Examples

audit_config(permutations = 99, membership = "paired")
#> $membership
#> [1] "paired"
#> 
#> $weighted
#> [1] FALSE
#> 
#> $baseline_predictors
#> character(0)
#> 
#> $annotation_status
#> [1] "accepted"    "adjudicated"
#> 
#> $probability_sample
#> [1] FALSE
#> 
#> $confidence
#> [1] 0.95
#> 
#> $seed
#> [1] 104729
#> 
#> $permutations
#> [1] 99
#> 
#> $cluster
#> NULL
#> 
#> $test_families
#> NULL
#>