Decide independence at significanceLevel (default 0.05 when
omitted, matching the Python binding) using the test's rule.
The core rejects a non-finite significanceLevel.
Optionalsignificance_level: number | nullStatic metadata: { name, dataTypes, symmetric, rule }.
Run the test for x ⟂ y | z, returning a plain object
{ statistic, pValue, dof, effectSize } (null for absent
fields). x/y are column names; z is an array of names.
Construct the test bound to
data(aDatasetinstance or a raw columns object{ name: { kind, values } }) with an optionalconfigobject. Pass a sharedDatasetto avoid re-copying the column arrays across the wasm boundary when several tests share the same data.