Inderes Coffee Room (Part 11)

Correlation is not the same thing as causation. Even if two things change in tandem, it doesn’t automatically mean that one causes the other. It could be a common underlying factor driving both, or alternatively, it could be pure coincidence, where a statistical connection arises without any real dependency. Therefore, mere statistical correlation is not enough to prove a cause-and-effect relationship.

A classic example of the first case is ice cream consumption and drownings, which correlate, but neither causes the other; instead, the common underlying factor driving both is warm weather.

As an example of the second, consider this graph from Spurious Correlations: Popularity of the first name Sunny correlates with Salesforce's stock price (CRM) (r=0.985)

Thirdly, even if a cause-and-effect relationship had existed in the past, we cannot be certain of its continuity if, for instance, the party responsible for buybacks has changed, guidance has changed, etc.

This is a bit of a company-unrelated “zero-post,” but it’s worth being careful about what kind of conclusions one draws when conducting analysis with incomplete information.

7 Likes