← back to series

Causal Inference Series

A comprehensive guide to causal inference—from fundamental concepts to advanced econometric methods and causal machine learning.

What you'll learn: How to move beyond correlation to answer causal questions—does X cause Y? What would have happened if we had done Z instead? This series covers the essential frameworks, methods, and practical applications for rigorous causal reasoning in business and research.

13 articles • 6 modules • All articles published

Module 1: Foundations

1.Potential Outcomes Framework

published

Rubin Causal Model, ATE, CATE, and the fundamental problem of causal inference. Why correlation ≠ causation.

Rubin ModelATE/CATERCTsSelection Bias

2.Causal Graphs and DAGs

published

Directed Acyclic Graphs, d-separation, backdoor/frontdoor criteria, and identifying confounders vs colliders.

DAGsd-separationDo-calculusMediators & Colliders

Module 2: Classical Methods

3.Matching & Propensity Scores

published

Exact matching, propensity score estimation, matching/weighting/stratification, and assessing overlap.

Propensity ScoresCovariate BalanceCommon SupportSensitivity

4.Regression & Instrumental Variables

published

Regression adjustment, difference-in-differences, IV/2SLS, regression discontinuity, and fixed effects.

DiDIV/2SLSRDDFixed Effects

Module 3: Machine Learning for Causal Inference

5.Double/Debiased Machine Learning

published

Neyman orthogonality, cross-fitting, DML for treatment effects, and handling high-dimensional confounders.

DMLNeyman OrthogonalityCross-fittingHigh-dim Confounders

6.Causal Forests & Tree Methods

published

Causal trees, honest forests, generalized random forests, and variable importance for heterogeneous effects.

Causal TreesHonest ForestsGRFTreatment Heterogeneity

7.Meta-Learners for CATE

published

S-learner, T-learner, X-learner, DR-learner, R-learner—comparison and practical uplift modeling.

S/T/X-learnerDR-learnerR-learnerUplift Modeling

Module 4: Deep Learning for Causality

8.Neural Networks for Causal Effects

published

Treatment-agnostic representations, balancing representations (CFR, TARNET), adversarial learning, CEVAE.

CFR/TARNETAdversarial BalanceCEVAERepresentation Learning

9.Deep IV & Causal Discovery

published

DeepIV for instrumental variables, neural tangent kernels, structure learning, and DAG optimization.

DeepIVNeural Tangent KernelsCausal DiscoveryDAG Learning

Module 5: Advanced Topics

10.Time Series & Panel Data

published

Synthetic control methods, time-varying treatments, marginal structural models, and sequential treatments.

Synthetic ControlTime-varying TreatmentsMSMPanel Methods

11.Causal Reinforcement Learning

published

Off-policy evaluation, contextual bandits, counterfactual reasoning in RL, and safe policy learning.

Off-policy EvalContextual BanditsCounterfactual RLSafe Learning

Module 6: Practice & Deployment

12.Real-World Applications

published

A/B testing platforms, causal ML in tech (ads, recommendations), healthcare, and policy evaluation.

A/B TestingTech ApplicationsHealthcarePolicy Evaluation

13.Validation & Deployment

published

Model validation, sensitivity analysis, handling assumption violations, ethics, and production systems.

ValidationSensitivity AnalysisRobustnessEthicsProduction

This series is designed to be read sequentially, with each article building on concepts from previous ones. However, if you're already familiar with the basics, feel free to jump to specific topics of interest.