Logical fallacy · explained

The post hoc fallacy

The post hoc fallacy assumes that because one thing followed another, the first caused the second. From the Latin post hoc ergo propter hoc — 'after this, therefore because of this.' It is the engine behind superstition, most bad medical reasoning, and a great deal of confident nonsense about what made a business succeed.

The post hoc fallacy is the error of concluding that because one event followed another, the first caused the second. Temporal sequence is necessary for causation but nowhere near sufficient: the correlation may be coincidence, may run in the opposite direction, or may be produced by a third factor causing both.

The positions · at a glance
The fallacy at a glance
Position 01

After, therefore because

What it is

Treating sequence as proof of cause. B followed A, so A must have produced B.

Position 02

Coincidence

Alternative one

With enough events, striking sequences occur constantly by chance alone. Memory keeps the hits and discards the misses.

Position 03

Reverse causation

Alternative two

B may have caused A. Sequence alone doesn't tell you which direction the arrow points.

Position 04

A common cause

Alternative three

Some third factor may produce both — the most frequently missed explanation of a real correlation.

You take a remedy and feel better the next day. The sequence is real; the causal conclusion is not established. Most minor illnesses resolve on their own, and you would very likely have improved anyway — which means the improvement following the remedy is exactly what you would expect whether or not the remedy did anything at all.

Temporal order is a genuine requirement for causation: causes precede effects. But it is a filter, not a proof, and treating it as proof is one of the costliest errors in ordinary reasoning. At least three other explanations always compete. It may be coincidence — with enough events, striking sequences are guaranteed, and memory obligingly preserves the hits while discarding the misses. It may be reverse causation — B produced A, and the sequence you noticed was misleading. Or there may be a common cause producing both, which is the possibility people miss most often and the reason ice cream sales track drowning deaths.

Sequence is a requirement for causation, not evidence of it.The core distinction

The medical version is the most consequential, and it is why controlled trials exist. Regression to the mean guarantees that people typically seek treatment when symptoms peak, so improvement afterwards is the statistical expectation regardless of what was taken. Add the placebo effect and the natural course of illness, and post hoc reasoning will validate almost any intervention. The entire apparatus of randomisation and control groups exists to defeat precisely this fallacy — which is a good measure of how powerful it is.

Testing a causal claim means asking what else could explain the sequence. Does it hold up when repeated? Does the effect scale with the dose? Is there a plausible mechanism? Does it survive when you control for other factors? None of these is individually decisive, but together they are the difference between noticing a sequence and establishing a cause.

People also askQuick answers

What is the post hoc fallacy?

Assuming that because one event followed another, the first caused the second — from the Latin post hoc ergo propter hoc, 'after this, therefore because of this.'

What is an example of the post hoc fallacy?

Taking a remedy, feeling better the next day, and concluding the remedy worked. Most minor illnesses resolve on their own, so improvement is what you'd expect regardless of whether the remedy did anything.

Why is correlation not causation?

Because a correlation can arise from coincidence, from reverse causation, or from a third factor producing both. Sequence tells you causation is possible; it doesn't tell you it happened, or in which direction.

How do you test whether something is actually causal?

Ask whether it replicates, whether the effect scales with the dose, whether there is a plausible mechanism, and whether it survives controlling for other factors. Randomised controlled trials exist specifically to defeat this fallacy.

Sources & further reading

Fallacy classifications vary between reference works. Whether a given argument commits this fallacy often depends on context and on interpreting the speaker charitably.