Post Hoc Ergo Propter Hoc

mental-model

Sequential stories feel causal because the mind turns “after this” into “because of this.” The real test is whether the intervention made recovery more likely than it would have been without the intervention.

A child’s asthma improves after a succession of remedies. The parents credit the final remedy—even though it may simply have been the one that happened to coincide with natural improvement.

The missing counterfactual

Recovery supplies a satisfying ending, and whatever immediately preceded it becomes the apparent cause. But the observed sequence contains only one timeline: treatment, then improvement. It says nothing about the unobserved timeline in which no treatment was given.

The causal question is therefore comparative, not chronological. You must estimate how often the outcome occurs with the intervention versus without it. Natural recovery, fluctuating symptoms, and the tendency to try remedies when illness is near its worst can all produce persuasive success stories without any therapeutic effect. The Narrative Fallacy then compresses that ambiguity into a clean plot.

E1 E2 E3

Where it shows up

The last asthma remedy

Trying several remedies almost guarantees that one will be temporally closest to improvement. That privileged position can be mistaken for efficacy.

Homeopathy before recovery

A cough that was already likely to resolve improves after homeopathy, creating a testimonial that cannot reveal what would have happened without it.

Surviving bloodletting

Patients recovered after bloodletting, but recovery after treatment could not distinguish benefit from the ordinary course of conditions that often improve anyway.

Sequence is not disproof either

This fallacy does not mean that an intervention preceding recovery was useless. Timing can generate a causal hypothesis; it simply cannot settle it. The model is most important when spontaneous recovery is common, because anecdotes then blur effective treatments and inert ones alike.

E2 E3

Ask for the untreated timeline

When someone says “I did X and then I got better,” ask one concrete question: How often do comparable people recover without X? Look for a control group—ideally a randomised one—before treating the sequence as evidence that X changed the odds.

Episodes that teach this

Why We Only Hear About The Winners? Survivorship Bias Explained | FutureIQ 8,091 views
Parents try many remedies for childhood asthma, and the last thing tried before it improves naturally gets credited as the cure.
Finding A Way Out Of Misinformation | Future IQ · explained at 2:40 6,329 views
A cough was anyway going to get better, then someone takes homeopathy and believes homeopathy cured them.
The Randomness of Science - Why Randomised Controlled Trials (RCTs) Work - FutureIQ · explained at 3:44 4,109 views
After bloodletting, many patients recovered, but the transcript notes that most people may recover from many conditions whether you do anything or not.