INKWIRE(7) argument errors INKWIRE(7)
post-hoc
Post Hoc, Ergo Propter Hoc
B happened after A, therefore A made B happen.
NAME
Definition
The Latin tag means "after this, therefore because of this". The argument takes the single fact that two events were ordered in time and promotes it into a causal claim. Nothing else is offered: no mechanism, no comparison group, no check that B ever occurs without A.
It survives because the human mind reads narrative sequence as causal sequence. A story where A causes B feels complete; the same two events with an unknown third factor between them feel unfinished, even when that is exactly what the evidence shows.
The fallacy is not the claim that A caused B. Many things do follow their causes. The error is treating sequence alone as sufficient proof, which makes the inference unfalsifiable in practice.
SIGNALS
-
S1
Sequence is the only evidence
Strip the argument down and what remains is "it came first". If removing the timeline leaves no support at all, you are looking at post hoc.
-
S2
No mechanism is proposed
Causal claims normally come with a story about how the effect is produced. Post hoc substitutes "and then" for "because", and treats the substitution as self-evident.
-
S3
Counter-cases are absorbed, not explained
When you point at people who had A and no B, the answer is usually that their case was different, milder, or not a real case. A claim that explains away every disconfirmation has stopped making predictions.
Three checks, in order
- Ask what else changed at the same time. If the effect started when the cause did, look for a third factor that started then too.
- Ask whether the effect happens without the cause. Find one population with A and no B, or B and no A.
- Ask for a dose response. Real causes usually scale: more of A should give more of B, not an all-or-nothing switch.
Where the link breaks
| Looks like a cause | What breaks the link |
|---|---|
| Ice cream sales rise, then drowning deaths rise | Both track summer temperature; nobody drowns because of a cone |
| Crime fell after a new policing programme | Crime fell over the same years in cities that never ran the programme |
| My laptop crashed right after I sent a large file | The memory leak triggers at six hours of uptime, which happens to be when you send files |
EXAMPLES
The Great Plague had barely left London before the Great Fire swept through in September 1666, and within months the disease was gone. Contemporary writers concluded that the fire had purified the air and finished the plague.
Why it fails
Plague deaths had already collapsed by the winter of 1665–66, months before the fire, and the rats and fleas that carried the disease were never part of the argument. The fire came after the decline and got credit for it.
Kindred entries
- [F] survivorship You only studied the ones that made it, then asked what they have in common.
- [F] base-rate A vivid new detail outweighs the much larger number it should be measured against.
- index — /t/fallacy/
[+] Notes and objections 3
- Is it always wrong to infer cause from order?
- No. Order is a necessary condition for cause, so it is a legitimate starting point. It becomes a fallacy when order is all you have and you refuse to look for the mechanism or the comparison group.
- How is this different from confusing correlation with causation?
- Correlation is about two variables moving together with no time direction. Post hoc is narrower and more confident: it adds the claim that the earlier event produced the later one, purely from the clock.
- What is the fastest practical check?
- Ask when the trend actually started. In most real disputes the outcome was already moving before the supposed cause arrived, and the timeline settles the argument without any statistics.
Scope — Reference material for reading arguments, not professional advice.