Prediction-tuned synthetic data can pass realism tests while distorting average treatment effects, and separating covariate generation from treatment/outcome modeling largely fixes it.
M., Yang, F., and Dahabreh, I
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
years
2026 4representative citing papers
Non-asymptotic analysis of prediction-powered mean estimation shows that no-regret learning for query probabilities converges to the maximum allowed constant value, independent of covariates.
Including LLM predictions as covariates in standard regression adjustment for randomized experiments reduces variance with a do-no-harm property that reverts to the unadjusted estimator when predictions are uninformative.
GLIDE is a Python library that packages multiple PPI estimators and samplers for reliable GenAI evaluation and reports annotation savings in an agentic case study.
citing papers explorer
-
Generative Synthetic Data for Causal Inference: Pitfalls, Remedies, and Opportunities
Prediction-tuned synthetic data can pass realism tests while distorting average treatment effects, and separating covariate generation from treatment/outcome modeling largely fixes it.
-
Revisiting Active Sequential Prediction-Powered Mean Estimation
Non-asymptotic analysis of prediction-powered mean estimation shows that no-regret learning for query probabilities converges to the maximum allowed constant value, independent of covariates.
-
AI-Assisted Variance Reduction in Randomized Experiments
Including LLM predictions as covariates in standard regression adjustment for randomized experiments reduces variance with a do-no-harm property that reverts to the unadjusted estimator when predictions are uninformative.
-
Industrializing Prediction-Powered Inference: The GLIDE Library for Reliable GenAI and Agentic Systems Evaluation
GLIDE is a Python library that packages multiple PPI estimators and samplers for reliable GenAI evaluation and reports annotation savings in an agentic case study.