# Model Agnostic Differentially Private Causal Inference

arXiv: https://arxiv.org/abs/2505.19589 · HTML: https://mlsanity.com/paper/2505.19589/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 21 of its 50 references that are in the map.
- First posted on arXiv 26 May 2025.

## Builds on (most cited first, 21 of 21)

- [Renyi Differential Privacy](https://mlsanity.com/paper/1702.07476/) (2017)
- [Concentrated Differential Privacy: Simplifications, Extensions, and Lower Bounds](https://mlsanity.com/paper/1605.02065/) (2016)
- [Scalable Private Learning with PATE](https://mlsanity.com/paper/1802.08908/) (2018)
- [Improving the Gaussian Mechanism for Differential Privacy: Analytical Calibration and Optimal Denoising](https://mlsanity.com/paper/1805.06530/) (2018)
- [Knock Knock, Who's There? Membership Inference on Aggregate Location Data](https://mlsanity.com/paper/1708.06145/) (2017)
- [Differentially Private Release and Learning of Threshold Functions](https://mlsanity.com/paper/1504.07553/) (2015)
- [Estimating Treatment Effects with Causal Forests: An Application](https://mlsanity.com/paper/1902.07409/) (2019)
- [Accuracy, Interpretability, and Differential Privacy via Explainable Boosting](https://mlsanity.com/paper/2106.09680/) (2021)
- [Causal Inference with Corrupted Data: Measurement Error, Missing Values, Discretization, and Differential Privacy](https://mlsanity.com/paper/2107.02780/) (2021)
- [Confidence-Ranked Reconstruction of Census Microdata from Published Statistics](https://mlsanity.com/paper/2211.03128/) (2022)
- [Differentially Private Estimation of Heterogeneous Causal Effects](https://mlsanity.com/paper/2202.11043/) (2022)
- [Hyperparameter Tuning for Causal Inference with Double Machine Learning: A Simulation Study](https://mlsanity.com/paper/2402.04674/) (2024)
- [Privacy-Preserving Causal Inference via Inverse Probability Weighting](https://mlsanity.com/paper/1905.12592/) (2019)
- [Locally Private Causal Inference for Randomized Experiments](https://mlsanity.com/paper/2301.01616/) (2023)
- [Differentially Private Estimation of Weighted Average Treatment Effects for Binary Outcomes](https://mlsanity.com/paper/2408.14766/) (2024)
- [Gaussian DP for Reporting Differential Privacy Guarantees in Machine Learning](https://mlsanity.com/paper/2503.10945/) (2025)
- [Differentially Private Covariate Balancing Causal Inference](https://mlsanity.com/paper/2410.14789/) (2024)
- [Differentially Private Learners for Heterogeneous Treatment Effects](https://mlsanity.com/paper/2503.03486/) (2025)
- [Differentially Private Multi-Site Treatment Effect Estimation](https://mlsanity.com/paper/2310.06237/) (2023)
- [Privacy-preserving Quantile Treatment Effect Estimation for Randomized Controlled Trials](https://mlsanity.com/paper/2401.14549/) (2024)
- [Federated Experiment Design under Distributed Differential Privacy](https://mlsanity.com/paper/2311.04375/) (2023)

## Built on by (most cited first, 2 of 2)

- [Workload-Preserving Differentially Private Synthetic Data for Causal Inference via Maximum-Entropy Calibration](https://mlsanity.com/paper/2607.08122/) (2026)
- [Differentially Private Average Treatment Effect Estimation by Propensity Score Blocking](https://mlsanity.com/paper/2609.09536/) (2026)

## Abstract (arXiv)

Estimating causal effects from observational data is essential in fields such as medicine, economics and social sciences, where privacy concerns are paramount. We propose a general, model-agnostic framework for differentially private estimation of average treatment effects (ATE) that avoids strong structural assumptions on the data-generating process or the models used to estimate propensity scores and conditional outcomes. In contrast to prior work, which enforces differential privacy by directly privatizing these nuisance components, our approach decouples nuisance estimation from privacy protection. This separation allows the use of flexible, state-of-the-art black-box models, while differential privacy is achieved by perturbing only predictions and aggregation steps within a fold-splitting scheme with ensemble techniques. We instantiate the framework for three classical estimators -- the G-Formula, inverse propensity weighting (IPW), and augmented IPW (AIPW) -- and provide formal utility and privacy guarantees, together with privatized confidence intervals. Empirical results on synthetic and real data show that our methods maintain competitive performance under realistic privacy budgets.