The New Science of Cause and Effect
summary
Argues that data science needs an actual causal engine: a graph‑based Structural Causal Model plus surgical interventions and do‑calculus that let one answer interventional and counterfactual queries rather than just correlations. Explains two fundamental laws (counterfactuals from a complete model and graphical d‑separation implying independencies), the three‑level Ladder of Causation, seven practical pillars (external validity, selection bias, mediation, transportability), and gives algorithms with completeness results for identifiability and combining heterogeneous studies.
AI-written summary · about
chapters (21)
- 0:00Speaker Introduction
- 1:00Introduction
- 1:18Talk Proverb
- 2:13Talk Outline
- 5:11Causal Models and the Cognitive Revolution
- 10:50Typical Causal Questions & The Limitation of Standard Grammar of Science
- 18:15The Ladder of Causation (3 Level Hierarchy)
- 26:36Simpson's Paradox
- 31:00Explainability Deep-Learning Style
- 34:26Distinguish Seeing from Doing
- 36:19The Two Fundamental Laws of Causal Inference
- 38:15Reading Independencies
- 40:02Structural Causal Model (SCM) Inference Engine
- 42:33The Seven Pillars of Causal Wisdom
- 44:44Pillar 5: External Validity and Sample Selection Bias
- 45:51Pillar 5: The Problem in Real Life
- 47:06Pillar 5: The Problem in Mathematics
- 49:16Conclusion
- 51:56Q&A 1: Opinion on Natural Experiments to Discover Causal Connections in Data
- 57:38Q&A 2: Opinion on the Popularization of Statistics in News Media
- 1:01:30Q&A 3
related talks
01h 01mThe Mathematics of Causal Inference, with Reflections on Machine Learning and the Logic of Science
29m🧠🧹 Causality - Mental Hygiene for Data Science
45mCausality and Statistics, a Love Not So Impossible After All
06mCausal Machine Learning – What’s in it for Data Scientists?
43mA Practical Guide to Causality in Python (For The Perplexed)
01h 06mIntroduction to Causal Graphical Models: Graphs, d-separation, do-calculus
24mA Friendly Introduction To Causality
31mA Gentle Introduction To Causal Inference