False cause is the error of inferring that A caused B when the evidence supports only that A and B are associated. Correlation is not causation: two things can co-occur or move together because a third factor causes both, because of chance, or because the direction of cause is reversed (B caused A). The fallacy appears whenever we jump from "A and B happened together" or "A before B" to "A caused B" without ruling out alternatives. Post hoc, ergo propter hoc — "after this, therefore because of this" — is the classic form: we assume that because one event preceded another, it caused it.
The cost is misattribution. We credit or blame the wrong lever, double down on useless interventions, or abandon ones that work. In business, false cause drives bad strategy when we attribute outcomes to the wrong initiative; in investing, when we attribute returns to skill instead of luck or to the wrong factor; in policy, when we attribute change to a law or programme that was not the real driver. The corrective is to demand a mechanism (how would A cause B?) and to consider confounders and reverse causation before concluding cause.
Practical use: when someone claims "we did X and then Y happened," ask: what else changed? Could Y have caused X? Could a common cause explain both? Reserve "cause" for when you have a plausible mechanism and have tested or ruled out rival explanations.
Section 2
How to See It
False cause shows up when causation is asserted on the basis of sequence or correlation alone, when confounders are ignored, or when a single story is favoured over alternative explanations. Look for "we did X, then Y" presented as proof of cause.
Business
You're seeing False Cause when a team credits a new pricing page for a 15% conversion lift while a major competitor simultaneously left the market and traffic mix shifted. The new page may have helped, but the competitor and traffic are plausible causes. Attributing the lift entirely to the page is false cause without controlled comparison.
Technology
You're seeing False Cause when an eng team attributes a drop in errors to a new monitoring tool when the real change was a reduction in deployment frequency. Sequence (tool then fewer errors) does not establish cause. The tool might have had no effect or even made debugging harder; fewer deploys could explain the drop.
Investing
You're seeing False Cause when a fund attributes outperformance to its research process when the period was a broad bull run in its sector. Returns and process co-occurred; that does not prove the process caused the returns. Without a control (e.g. same process in different markets) or a clear mechanism, cause is overstated.
Markets
You're seeing False Cause when a regulator attributes a fall in incidents to a new rule when economic conditions, technology, or industry structure also changed. Policy and outcome are correlated; causation requires ruling out confounders and establishing a mechanism.
Section 3
How to Use It
Decision filter
"Before concluding that A caused B, ask: what is the mechanism? What else could explain the association? Could B have caused A, or could a third factor cause both? Demand evidence that goes beyond sequence or correlation."
As a founder
Attribute outcomes to causes with care. When a metric improves after a change, list rival explanations: seasonality, market shift, other changes, regression to the mean. Run experiments (A/B tests, holdouts) where possible so you compare "with" vs "without" the cause. The mistake is building strategy on false cause — e.g. scaling a channel that you think drove growth when the real driver was something else.
As an investor
Be sceptical of narratives that attribute success or failure to one factor. Ask how the cause would produce the effect (mechanism) and what would have happened without it (counterfactual). Back teams that run controlled tests and separate correlation from cause; be wary of those who present sequence as proof.
As a decision-maker
Institutionalise "could something else explain this?" in reviews. Require mechanism and consideration of confounders before accepting causal claims. When allocating credit or blame, resist the story that is simplest or most flattering; consider reverse causation and common causes. Decisions based on false cause are expensive.
Common misapplication: Dismissing all causal inference as false cause. The point is to improve causal inference — mechanism, confounders, counterfactuals — not to say we can never know cause. Use the lens to sharpen claims, not to paralyse them.
Second misapplication: Demanding RCTs for every claim. Randomised experiments are the gold standard but are not always feasible. Mechanism, natural experiments, and ruling out confounders can support weaker but useful causal claims. The bar is "better than sequence alone," not "RCT or nothing."
Feynman insisted on distinguishing what we know from what we think we know. In the Challenger investigation, he resisted narrative and showed the mechanism (O-ring failure in cold). His habit was to ask "how would that work?" and "what else could explain it?" — the same questions that block false cause. Causal claims had to pass a clear mechanism and consideration of alternatives.
Munger warns against the "man with a hammer" and narrative bias: we like stories where one cause explains the effect. He stresses alternative explanations and inversion — what would have to be true for this cause to be wrong? That discipline reduces false cause in investing and in judging management.
Section 6
Visual Explanation
Picture A and B as nodes. An arrow A → B is "A caused B." False cause is drawing the arrow on the basis of "A and B happened" or "A before B" alone. The reality might be B → A, or Z → A and Z → B (confounder), or no arrow (chance). The discipline is to draw arrows only when you have mechanism and have considered the other diagrams.
Section 7
Connected Models
False cause sits in the family of causal reasoning, correlation, and evidence. These models reinforce the danger of conflating association with cause or extend how to do better.
Reinforces
Correlation vs Causation
Correlation vs causation is the same idea: association does not imply cause. False cause is the fallacy of acting as if it does. The two reinforce each other — use both to slow down causal claims.
Reinforces
Confounding Factor
A confounding factor is a variable that causes both A and B, creating a spurious association. False cause often survives because confounders are ignored. Controlling for or reasoning about confounders is how you reduce false cause.
Tension
Narrative Fallacy
Humans prefer coherent narratives with clear causes. Narrative fallacy is the bias toward such stories; false cause is one result. The tension: narrative is useful for communication and memory but dangerous when it substitutes for causal rigor.
Tension
Necessity & Sufficiency
Necessity (cause required for effect) and sufficiency (cause enough for effect) refine causal claims. False cause ignores both: we assert cause without checking if it was necessary or sufficient. The tension: loose causal talk avoids the necessity/sufficiency question.
Section 8
One Key Quote
"We may define a cause to be an object, followed by another, and where all the objects similar to the first are followed by objects similar to the second."
— David Hume
Hume is saying we observe succession and regularity, not cause itself. Cause is an inference. False cause is the error of inferring cause from succession alone — "A followed B" — without the additional structure (mechanism, no confounders) that would justify the inference.
Section 9
Analyst's Take
Faster Than Normal — Editorial View
False cause is one of the most expensive errors in strategy. We scale what we think worked and kill what we think failed; if the causal story is wrong, we scale the wrong thing and kill the right one. The fix is to treat causal claims as hypotheses and to test them with mechanism, comparison, and counterfactuals.
Sequence is not cause. "We launched X and then Y improved" is a hypothesis, not proof. List rival explanations and, where possible, run a controlled test. The cost of one good experiment is often lower than the cost of years of strategy built on false cause.
Confounders are the usual culprit. When A and B move together, ask what might cause both. Market conditions, seasonality, and other initiatives are common confounders. Write them down and reason about them before concluding cause.
Investors should be especially wary. Performance attribution is full of false cause — we attribute returns to skill, process, or thesis when luck, sector, or timing could explain the result. Ask for mechanism and for how the team would know if they were wrong.
Section 10
Test Yourself
Is this mental model at work here?
Scenario 1
A company launches a new brand campaign. Sales rise 25% over the next quarter. The CMO attributes the lift entirely to the campaign.
Scenario 2
A team runs an A/B test: treatment group sees a new onboarding flow, control does not. Conversion is 5% higher in the treatment group. They conclude the new flow caused the lift.
Section 11
Summary & Further Reading
False cause is inferring that A caused B when the evidence supports only association or sequence. Correctives: ask for a mechanism, consider confounders and reverse causation, and use comparison (experiments, counterfactuals) where possible. Reserve causal claims for when you have done that work.
Tetlock on how to update beliefs and attribute outcomes in forecasting. Reinforces the need for mechanism and counterfactuals when judging cause.
Leads-to
Counterfactuals
Counterfactual reasoning asks: what would have happened without A? That is the right way to test cause. False cause is the failure to ask; counterfactuals are the corrective. Leads to causal inference and experimentation.
Leads-to
Randomized Controlled Experiment
RCTs assign cause (treatment) at random so that association is not explained by confounders. They are the strongest design to avoid false cause. When feasible, run them; when not, use the logic (control, comparison, mechanism) to strengthen claims.
