Post Hoc Fallacy
Assuming that because one thing happened before another, it must have caused it — confusing sequence with causation — is the source of superstition, bad science, and a great deal of bad economic policy.
The short version
- The post hoc fallacy (full name: post hoc ergo propter hoc — 'after this, therefore because of this') assumes that because one event preceded another, it must have caused it.
- Correlation and sequence are not causation: two things can be related in time without either one causing the other — they may share a common cause, or the relationship may be coincidental.
- The fallacy underlies superstition, pseudoscience, and many political claims about which policies 'caused' which economic or social outcomes.
- Establishing genuine causation requires controlled studies, the elimination of confounding variables, and ideally a proposed mechanism — not just a documented sequence of events.
What it is
The post hoc fallacy takes its name from the Latin phrase post hoc ergo propter hoc — 'after this, therefore because of this.' It is the error of concluding that because one event preceded another in time, the first event caused the second. The sequence A → B is taken as evidence that A caused B, when in reality the relationship between A and B could be coincidental, or both could be caused by a third factor C, or the relationship could be reverse-causal, or the statistical association could be an artifact of the data.
The fallacy is the conceptual root of superstition. If a baseball player doesn't shave before a game and his team wins, and he therefore doesn't shave before subsequent games — that's post hoc reasoning. If a person takes a herbal supplement and then recovers from an illness, they may attribute recovery to the supplement — without accounting for the fact that most illnesses resolve on their own. The supplement preceded the recovery; it didn't necessarily cause it. The temporal sequence feels like causal evidence because human cognition is built to detect patterns, and sequence is the most basic pattern we know.
In statistics, the distinction between correlation and causation is fundamental, and the tools for establishing genuine causation — randomized controlled trials, natural experiments, instrumental variable analysis, the identification and control of confounding variables — are specifically designed to rule out the post hoc inference. The reason these methods are necessary is that the post hoc pattern (A precedes B; therefore A caused B) is so cognitively compelling that even trained researchers can be misled by it. The history of science is full of causal claims that were later shown to be confounded — things that moved together in time for reasons entirely unrelated to one another.
The concept of spurious correlation illustrates the problem vividly: per capita cheese consumption correlates strongly with deaths by bedsheet tangling in the United States over a 10-year period. Ice cream sales correlate with drowning rates. Nicolas Cage film releases correlate with drowning in swimming pools. These are not causal relationships — they're statistical accidents or both driven by a third variable (summer, in the ice cream/drowning case). The post hoc fallacy is essentially the failure to recognize that sequence and correlation are cheap; causation is expensive to establish.
Why it matters
Post hoc reasoning is endemic in political arguments about economic policy. Presidents and parties claim credit for economic conditions that preceded them or developed independently of their policies, and assign blame for downturns to the party that happened to be in power when a business cycle turned. Research on presidential economic responsibility consistently finds that voters hold incumbents accountable for economic conditions that are often driven by factors (global commodity prices, monetary policy, business cycles) largely outside executive control. The sequence — 'economy got better/worse under President X, therefore President X caused it' — is compelling but frequently wrong.
In healthcare and pharmacology, post hoc reasoning has led to enormous harm. Patients who take ineffective treatments and recover attribute the recovery to the treatment. Parents who notice autism symptoms emerging around the same time as childhood vaccines attributed autism to the vaccines — a post hoc inference that fueled the anti-vaccine movement. The scientific literature on vaccine safety has repeatedly demonstrated no causal link; the temporal coincidence (autism symptoms often become apparent around ages 12-18 months, the same period when several vaccines are administered) created a compelling but false causal narrative. The post hoc inference, amplified across millions of families and supercharged by social media, produced a public health crisis.
The fallacy also underlies much of the reasoning in alternative medicine, where practitioners can claim their treatments 'work' because patients who use them sometimes get better. Without controlled comparison — what happens to patients with the same condition who don't use the treatment? — it's impossible to separate the treatment's effect from natural recovery, placebo response, regression to the mean, or the health improvements associated with the attention and care that accompany any treatment. Systematic reviews and meta-analyses are the tools medicine uses precisely to escape the post hoc inference.
The corrective to post hoc reasoning is demanding a causal mechanism and controlled comparison. It's not enough that A preceded B; we need a plausible account of how A caused B, and we need to compare what happened to similar people or systems where A was absent. This is the basic logic of the scientific experiment. Applied to policy, it means asking: compared to what? What would have happened without this intervention? Where has this policy been tried without producing the claimed effect? These are the questions that turn a post hoc narrative into genuine causal evidence — or reveal that the sequence was all there was.