causalis.scenarios.iv.refutation.iv_confounders_balance

Module Contents

Functions

iv_confounders_balance

Compute confounder balance diagnostics between instrument groups.

Data

__all__

API

causalis.scenarios.iv.refutation.iv_confounders_balance.iv_confounders_balance(data: causalis.data_contracts.iv_causal_data.IVCausalData) → pandas.DataFrame

Compute confounder balance diagnostics between instrument groups.

Produces a DataFrame containing expanded confounder columns (after one-hot encoding categorical variables if present) with:

• confounders: name of the confounder

• mean_z_0: mean value for rows with Z=0

• mean_z_1: mean value for rows with Z=1

• abs_diff: abs(mean_z_1 - mean_z_0)

• smd: standardized mean difference (Cohen’s d using pooled std)

• ks_pvalue: p-value for the KS test (rounded to 5 decimal places, non-scientific)

Parameters

data : IVCausalData The IV causal dataset containing exactly one binary instrument column.

Returns

pd.DataFrame Balance table sorted by |smd| (descending).

Examples

from causalis.scenarios.iv.refutation import iv_confounders_balance

Assuming ‘causal_data’ is an IVCausalData object

balance_df = iv_confounders_balance(causal_data) balance_df.head()

causalis.scenarios.iv.refutation.iv_confounders_balance.__all__

[‘iv_confounders_balance’]