New orthogonal risk functions are derived for conditional OR and RR, with simulations and NHANES data showing nonparametric estimators reduce bias compared to parametric alternatives in complex settings.
Title resolution pending
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
verdicts
UNVERDICTED 6representative citing papers
Cross-fitted covariance calibration enables chi-squared hypothesis tests for penalized estimating equations that are robust to covariance misspecification under correct conditional mean.
Asymptotic linearity and influence functions of M-estimators remain identical under rerandomization versus simple randomization, with non-Gaussian limits possible unless rerandomization variables are adjusted for, plus efficiency optimality for data-adaptive estimators under rerandomization and its
A kernel-smoothed decorrelated score with cross-fitting enables valid inference for coefficients in high-dimensional classification using piecewise linear surrogate losses.
Develops shadow variable identification and SIO estimation achieving asymptotic normality and local efficiency for mediation effects under nonignorable missing confounders.
Derives non-overlap bounds for the ATE on bounded outcomes with width proportional to non-overlap size, plus a TMLE estimator and multiplier bootstrap for inference.
citing papers explorer
-
Orthogonal machine learning for conditional odds and risk ratios
New orthogonal risk functions are derived for conditional OR and RR, with simulations and NHANES data showing nonparametric estimators reduce bias compared to parametric alternatives in complex settings.
-
Hypothesis Testing for Penalized Estimating Equations with Cross-Fitted Covariance Calibration
Cross-fitted covariance calibration enables chi-squared hypothesis tests for penalized estimating equations that are robust to covariance misspecification under correct conditional mean.
-
Asymptotic inference with flexible covariate adjustment under rerandomization and stratified rerandomization
Asymptotic linearity and influence functions of M-estimators remain identical under rerandomization versus simple randomization, with non-Gaussian limits possible unless rerandomization variables are adjusted for, plus efficiency optimality for data-adaptive estimators under rerandomization and its
-
Inference with non-differentiable surrogate loss in a general high-dimensional classification framework
A kernel-smoothed decorrelated score with cross-fitting enables valid inference for coefficients in high-dimensional classification using piecewise linear surrogate losses.
-
Efficient Nonparametric Inference for Mediation Analysis with Nonignorable Missing Confounders
Develops shadow variable identification and SIO estimation achieving asymptotic normality and local efficiency for mediation effects under nonignorable missing confounders.
-
Non-overlap Average Treatment Effect Bounds
Derives non-overlap bounds for the ATE on bounded outcomes with width proportional to non-overlap size, plus a TMLE estimator and multiplier bootstrap for inference.