Post Hoc Ergo Propter Hoc: The After-Therefore-Because Trap

The rooster crows, and the sun rises. Did the rooster cause the dawn? Post hoc ergo propter hoc — Latin for "after this, therefore because of this" — is the oldest causal fallacy in the logical tradition, and one of the hardest to shake. We observe that B followed A, and our minds immediately begin constructing a causal story. The sequence feels like evidence. In many cases it is. In many others, it is coincidence, confounding, or a common cause driving both — and the causal story we tell is wrong.

The Logical Structure

The fallacy has a simple form:

1. A occurred.
2. B occurred after A.
3. Therefore, A caused B.

The inference from temporal sequence to causal relationship is invalid because temporal precedence is a necessary condition for causation — causes must precede their effects — but it is not sufficient. Countless A→B sequences occur without A causing B: the stock market rises after the Super Bowl is won by a particular conference; a patient recovers after taking a homeopathic remedy; a government takes office and the economy improves. Sequence is everywhere. Causation requires more than sequence — it requires a mechanism, controlled evidence, and the elimination of alternative explanations.

Superstition: The Personal Laboratory

Superstitions are essentially personal post hoc inferences that have survived social transmission. The athlete who wore a particular pair of socks on the day of a memorable win subsequently wears them for every important game. The logic is explicitly temporal: I wore them, we won; therefore wearing them helps us win. This is post hoc reasoning in its purest form.

Behaviourist B. F. Skinner demonstrated in 1948 that pigeons could be made "superstitious" through random reinforcement schedules. Pigeons that received food at random intervals began repeating whatever behaviour they happened to be performing when food arrived — turning counterclockwise, bobbing their heads, shuffling from foot to foot. They had learned a spurious association between their own behaviour and the reward, because the reward followed the behaviour. The pigeons were doing post hoc reasoning without the Latin.

Human superstitions follow the same mechanism, compounded by our greater capacity for storytelling, social reinforcement, and motivated reasoning. We remember the times the lucky item worked and forget — or explain away — the times it didn't. This selective memory keeps the post hoc inference alive despite accumulating disconfirming evidence.

Medical Quackery and the Regression Trap

The post hoc fallacy is the structural engine of most medical quackery. A patient with a cold takes a herbal supplement and feels better three days later. The supplement gets the credit, even though the cold was going to resolve in approximately that timeframe regardless. The treatment followed; the recovery followed the treatment; cause and effect feel established.

What makes this particularly insidious is its interaction with regression to the mean: people tend to seek treatment when their symptoms are at their worst. From the peak of severity, outcomes can only improve — and any treatment applied at that moment will appear to have caused the improvement. The post hoc inference is built into the natural history of illness. This is one reason why anecdotal evidence of treatment efficacy is so unreliable, and why controlled trials — where the regression effect applies equally to treatment and placebo groups — are necessary.

Vaccine hesitancy is partly driven by post hoc reasoning of a specific type: a child receives a routine vaccination, and shortly afterward begins showing signs of autism or developmental difference. The vaccination preceded the symptoms; therefore, the inference goes, the vaccination caused them. Large-scale epidemiological studies consistently fail to find any causal link between vaccines and autism. But the personal post hoc narrative — "it happened right after the shot" — is emotionally compelling in ways that population statistics are not. The individual sequence feels like evidence even when the aggregate data contradicts it.

Policy Attribution: Who Gets Credit for the Economy?

Political life runs on post hoc reasoning. When an economic expansion follows a government's election, the government claims credit. When a recession follows, the opposition blames the policies. In both cases, the causal inference from temporal sequence to policy effect is largely unwarranted — economies are driven by long-running structural trends, global market conditions, monetary policy, and business cycles that extend across administrations and often reflect decisions made years or decades earlier.

The unemployment rate typically lags policy changes by 12-18 months. A president inaugurated during a recession will often see unemployment peak well into their first year, regardless of their policies; conversely, they may inherit an expansion that continues independently of what they do. Crime rates, similarly, follow demographic and social trends that unfold over decades — but politicians in office during a decline routinely attribute it to their particular policing or social policies, and the post hoc logic is rarely challenged because the public narrative demands clear attribution.

This matters because misattributed causes generate misguided policies. If a crime reduction is attributed to a zero-tolerance policing strategy rather than to a demographic bulge aging out of the peak crime years, resources will be directed at extending and expanding that strategy — which may produce harms — rather than understanding the actual mechanisms at work.

The Scientific Response: Controlled Conditions

Science developed the controlled experiment precisely to defeat post hoc reasoning. In a randomised controlled trial, participants are assigned to treatment and control conditions at random. Both groups experience whatever happens after the trial begins; the difference in outcomes between groups is therefore attributable to the treatment, because everything else — including temporal coincidences and regression to the mean — affects both groups equally.

David Hume, in his philosophical analysis of causation, argued in the 18th century that we never directly perceive causation — we perceive constant conjunction and temporal priority, and our minds supply the rest. His insight is still the foundation of the problem: causation is a theoretical inference, not a direct observation. Post hoc reasoning mistakes the observation (B followed A) for the theoretical inference (A caused B). The gap between them is where scientific method lives.

Distinguishing Post Hoc from Genuine Cause

Temporal sequence alone is not sufficient for causation, but it does provide a starting point for investigation. The additional tests worth applying:

• Mechanism: Is there a plausible causal pathway from A to B? What is the biological, physical, or social mechanism?
• Dose-response: Does more A produce more B? Does less A produce less B?
• Counterfactual: Would B have occurred without A? Controlled experiments are the gold standard for answering this.
• Consistency: Does A reliably precede B across different contexts, populations, and times? Or did the association appear once?
• Confounders: Is there a third variable C that causes both A and B, producing the association without any direct causal link between them?

The last point deserves emphasis. The classic formulation — "correlation does not imply causation" — is really about the confounders problem. Ice cream sales and drowning rates are positively correlated; hot weather drives both, but no one thinks ice cream causes drowning. Visible confounders are easy to dismiss. Hidden confounders — socioeconomic status, educational level, underlying health — routinely masquerade as the effects of variables we are looking at for other reasons.

Post hoc ergo propter hoc is not just a logical error; it is a fundamental feature of how human cognition works. We are prediction machines. We learn from sequences. The problem is that in a complex world, sequence is everywhere, most of it coincidental, and we lack the intuitive equipment to distinguish genuine causal signals from noise without deliberate methods. The next time an improvement follows an action you took — or a disaster follows one someone else took — the right question is not "did this cause that?" but "how would we know if it did?"

Sources & Further Reading

• Hume, D. An Enquiry Concerning Human Understanding. 1748. Section VII: "Of the Idea of Necessary Connexion."
• Skinner, B. F. "'Superstition' in the Pigeon." Journal of Experimental Psychology 38, no. 2 (1948): 168–172.
• Sagan, C. The Demon-Haunted World: Science as a Candle in the Dark. Random House, 1995.
• Wakefield, A. J., et al. "Ileal-lymphoid-nodular hyperplasia, non-specific colitis, and pervasive developmental disorder in children." The Lancet 351 (1998): 637–641. (Retracted 2010 — paradigmatic case of post hoc vaccine attribution.)
• Pearl, J., & Mackenzie, D. The Book of Why. Basic Books, 2018.
• Wikipedia: Post hoc ergo propter hoc