Abstract:We propose a new method to estimate causal effects from nonexperimental data. Each pair of sample units is first associated with a stochastic 'treatment' - differences in factors between units - and an effect - a resultant outcome difference. It is then proposed that all such pairs can be combined to provide more accurate estimates of causal effects in observational data, provided a statistical model connecting combinatorial properties of treatments to the accuracy and unbiasedness of their effects. The article introduces one such model and a Bayesian approach to combine the $O(n^2)$ pairwise observations typically available in nonexperimnetal data. This also leads to an interpretation of nonexperimental datasets as incomplete, or noisy, versions of ideal factorial experimental designs. This approach to causal effect estimation has several advantages: (1) it expands the number of observations, converting thousands of individuals into millions of observational treatments; (2) starting with treatments closest to the experimental ideal, it identifies noncausal variables that can be ignored in the future, making estimation easier in each subsequent iteration while departing minimally from experiment-like conditions; (3) it recovers individual causal effects in heterogeneous populations. We evaluate the method in simulations and the National Supported Work (NSW) program, an intensively studied program whose effects are known from randomized field experiments. We demonstrate that the proposed approach recovers causal effects in common NSW samples, as well as in arbitrary subpopulations and an order-of-magnitude larger supersample with the entire national program data, outperforming Statistical, Econometrics and Machine Learning estimators in all cases...

Estimation and Inference of Heterogeneous Treatment Effects using Random Forests

Wasserstein Random Forests and Applications in Heterogeneous Treatment Effects

Estimating Treatment Effects with Causal Forests: An Application

Heterogeneous treatment effect estimation for observational data using model-based forests

Shrinkage Bayesian Causal Forests for Heterogeneous Treatment Effects Estimation

Data-Driven Estimation of Heterogeneous Treatment Effects

Targeting relative risk heterogeneity with causal forests

Assessing the properties of patient-specific treatment effect estimates from causal forest algorithms under essential heterogeneity

Comparison of causal forest and regression-based approaches to evaluate treatment effect heterogeneity: an application for type 2 diabetes precision medicine

Harnessing Causal Forests for Epidemiologic Research: Key Consideration

Causal Rule Forest: Toward Interpretable and Precise Treatment Effect Estimation

Estimating individual treatment effect: generalization bounds and algorithms

Heterogeneous Treatment Effects in Panel Data

Modelling hetegeneous treatment effects by quantitle local polynomial decision tree and forest

Flexible machine learning estimation of conditional average treatment effects: a blessing and a curse

Distilling interpretable causal trees from causal forests

Hybrid Censored Quantile Regression Forest to Assess the Heterogeneous Effects

What can the millions of random treatments in nonexperimental data reveal about causes?

A groupwise approach for inferring heterogeneous treatment effects in causal inference

Accounting for Missing Covariates in Heterogeneous Treatment Estimation

Debiased Causal Tree: Heterogeneous Treatment Effects Estimation with Unmeasured Confounding