Cargo Cult : Correlation vs. Causation
During WWII, isolated and pre-industrial island cultures experienced soldiers and supplies arriving in large numbers, often by airdrop. It was the first time for these cultures to be introduced to big technological advances. During this period, the soldiers would trade with the islanders, materials for food.
After the war, the soldiers left. Cargo cults arose, the islanders tried to imitate the behaviors of the soldiers, thinking that this would cause the cargo planes to drop clothes and materials from the sky.
The islanders built wooden airplanes and radars like mocks from straw. They also imitated the marches of the soldiers while holding sticks for rifles, all of that in waiting for things to drop as it used to be.
What the islanders did not have was any prior knowledge about airplanes, and how electromagnetic fields work. They based their belief only on correlation between two observations.
From statistical point of view, what the islanders did is similar to what many businessmen and scientists are doing until today: Drawing conclusions based on correlation and mixing it for causation.
Remedies : To avoid falling for the cargo cult trap, there are many powerful statistical methods that can be used to indicate causation, including AB testing (the business equivalent of the scientific method). In a future article we will talk further on AB testing.
Check also fooled by correlation here.
About me : I am an experienced data scientist who believes in simplicity. You can connect with me on LinkedIn, the link is in the Bio.
