Skip to contents

Sets each donor's share of the counterfactual (its unit weight) against its share of the placebo variance behind the standard error – the draws in which it stood in for the treated unit. A donor far above the diagonal supplies more uncertainty than estimate.

Usage

donor_influence(run)

Arguments

run

An sdid_run fitted with keep = "all".

Value

A tibble with one row per donor: unit, label, weight, weight_share, n_placebo, n_trimmed, att_mean, abs_att_max, rmse_ratio and se_share, largest se_share first.

Details

Dropping a donor because it inflates the standard error is selection on the result; this table is for understanding an interval, not for choosing a pool.