fallacy.wiki

A field guide to the ways reasoning breaks

Causal errors

Correlation implies causation

Also called Cum hoc ergo propter hoc

What it is

Reading a statistical association as proof that one variable drives the other.

In the wild

Ice cream sales track drownings, so ice cream drowns people.

How to answer it

Look for a third factor that could produce both, and check whether the direction could be reversed.

In depth

Origin and naming

Often given in Latin as 'cum hoc ergo propter hoc', 'with this, therefore because of this'. The warning that correlation is not causation became a fixture of statistics teaching in the twentieth century, alongside famous spurious pairings like ice cream sales and drownings, both driven by hot weather.

More places it shows up

  1. Schools with more music programs get better test scores, so music must be raising the scores — not the neighborhoods that fund both.

  2. Countries that eat the most chocolate win the most Nobel prizes, so chocolate makes you smarter.

  3. Our most engaged users are also our highest payers, so let's make everyone more engaged and revenue will follow.

  4. Cities with more police have more crime, so adding police clearly causes crime.

How to spot it

  • The whole case is 'these two numbers move together' — no experiment, no mechanism, no third variable ruled out.
  • The arrow only ever points one direction; nobody checks whether Y might be causing X instead.
  • The claim survives exactly as long as nobody asks what else the two things share.

Read more