A matching-and-regression estimator for group average treatment effects is shown consistent, doubly robust, and asymptotically normal, with a bias-corrected version that avoids inverse propensity weighting.
Following an identi- S4 cal analysis to U1N (z), we have the variance of the first term Var N −1 PN i=1 Kh(Zi − z)Aiϵi = Op(1/(N h))
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ME 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Matching-Based Nonparametric Estimation of Group Average Treatment Effects
A matching-and-regression estimator for group average treatment effects is shown consistent, doubly robust, and asymptotically normal, with a bias-corrected version that avoids inverse propensity weighting.