Contents
0:12 Talk Introduction
1:00 Speaker Disclaimer and Slide Access
1:22 Data Driven Decision Making
2:07 Main Takeaways
2:31 Intended Audience of the Talk
2:51 Speaker Introduction
3:20 Talk Agenda — The Story Behind the Data
3:55 The Limitations of Correlations
4:10 Spurious Correlations — Example
6:00 Mediator Variable — Example
6:46 Mediator Variables Help To Infer Causation
8:13 Simpson’s Paradox and Data Misinterpretation
12:46 Uncontrolled Confounders as a Cause for Misinterpretation
13:45 Incorrect Use of Metrics as a Cause for Misinterpretation
14:05 Controlling Confounders via a Weight Adjustment (ACE)
14:56 Simpson’s Calculator Interactive Demo
16:58 Simpson’s Paradox for Continuous Data: Lord’s Paradox
17:57 Simpson’s Paradox: Summary
18:55 Randomized Control Trials — Gold Standard for Causality
20:21 Visualizing Parameters With Graph Models
22:21 Simpson’s Paradox Graph Model
23:15 Required Assumptions of Causation
23:44 Summary in the Context of Pearl’s Causality Ladder
24:38 Counterfactuals
25:21 Applicability of Causal Inference and the Story Behind the Data
26:37 Main Takeaways and Next Steps in Causal Inference
27:27 Acknowledgements
27:40 Resources
28:37 Q&A 1 — How Do You Prove Relationship Direction in Causal Graphs?
Video URL: https://www.youtube.com/watch?v=tMcCzoeuf30
Contents 0:12 Talk Introduction 1:00 Speaker Disclaimer and Slide Access 1:22 Data Driven Decision Making 2:07 Main Takeaways 2:31 Intended Audience of the Talk 2:51 Speaker Introduction 3:20 Talk Agenda — The Story Behind the Data 3:55 The Limitations of Correlations 4:10 Spurious Correlations — Example 6:00 Mediator Variable — Example
6:46 Mediator Variables Help To Infer Causation 8:13 Simpson’s Paradox and Data Misinterpretation 12:46 Uncontrolled Confounders as a Cause for Misinterpretation 13:45 Incorrect Use of Metrics as a Cause for Misinterpretation 14:05 Controlling Confounders via a Weight Adjustment (ACE) 14:56 Simpson’s Calculator Interactive Demo 16:58 Simpson’s Paradox for Continuous Data: Lord’s Paradox 17:57 Simpson’s Paradox: Summary 18:55 Randomized Control Trials — Gold Standard for Causality 20:21 Visualizing Parameters With Graph Models 22:21 Simpson’s Paradox Graph Model 23:15 Required Assumptions of Causation 23:44 Summary in the Context of Pearl’s Causality Ladder 24:38 Counterfactuals 25:21 Applicability of Causal Inference and the Story Behind the Data 26:37 Main Takeaways and Next Steps in Causal Inference 27:27 Acknowledgements 27:40 Resources 28:37 Q&A 1 — How Do You Prove Relationship Direction in Causal Graphs?