Causal Thinking
Chapter 3 - Thinking
Previous page: Thinking Fast and Slow - Causal Thinking - Next page: Model Thinking
Back to Book content or directly to Main Page
.
Welcome to the Causal Thinking (science of cause and effect) page
.
Levels of causation
.
- Association
- Detecting regularities in our environment (patterns)
- Intervention
- Predicting the effect(s) of deliberate alterations of the environment and choosing among these alterations to produce a desired outcome (relevancy)
- Counterfactuals
- A “theory” of why it works and what to do when it doesn’t (principles)
.
.
Correlation & causation
.
What do we actually observe when we say that one thing causes another? David Hume posed this question that still has no fully satisfying answer, in ‘A Treatise of Human Nature’ (1739) and ‘An Enquiry Concerning Human Understanding’ (1748).
Hume argued that we see A followed by B repeatedly, and we come to expect B whenever we see A. This ‘causal power’ we attribute to A is a projection of our minds, not a feature of the world. Hume was not saying causation does not exist. He was saying we cannot perceive it directly. We can only infer it. And inference requires assumptions that go beyond the data.
We learn that correlation does not imply causation, but few learn the converse: absence of correlation does not imply absence of causation.
The textbook cases are familiar:
First, confounding:
- Ice cream sales correlate with drowning deaths not because ice cream causes drowning, but because summer heat increases both. A common cause Z induces correlation between X and Y even when X has no causal effect on Y.
Second, reverse causation:
- Neighbourhoods with more police have higher crime rates, but this reflects that crime causes police deployment, not the reverse. The direction of causality matters, and correlation alone cannot tell us which way it runs.
Third, selection bias:
- Among Hollywood actors, talent and physical attractiveness appear negatively correlated-the “beautiful and dumb” stereotype. But this reflects selection: to become an actor, you need either exceptional talent or exceptional looks (or both). Conditioning on the collider “became an actor” induces a spurious association between its causes.
- Correlation measures association, whether knowing X tells you something about Y.
- Causation asks whether changing X would change Y.
.
.
Types of junctions
.
Through junctions we can see the secrets of the causal process we observe. Each junction stands for a distinct pattern of causal flow and leaves its mark in the form of conditional dependence and independence. (All this is very important when considering the relationships between the Actions fields).
Three junctions
- Mediator (chain),
- Confounder (fork),
- and Collider (magnet)
are the steps from patterns to principles.
.
| Content sources |
|---|
| The Book of Why - J. Pearl & D. Mackenzie - Basic Books - 2018 |
| Een wereld vol patronen - R. Bods - Prometheus - 2019 |
.