Correlation is not causation logical fallacy: definition, examples, and how to spot it

Why it misleads

A fuller explanation of how the fallacy works and why it can look persuasive.

Correlations matter because they can reveal patterns worth investigating. The fallacy is jumping straight from 'these moved together' to 'this caused that' without checking mechanisms, timing, controls, or alternative explanations.

That's like saying...

Instead of leading with the label, this analogy answers the shape of the reasoning move directly so the mistake is easier to see in plain language.

Fallacious claim

A school adds an AI tutoring tool and test scores rise the next semester, so the district concludes that the tool caused the improvement without ruling out teacher changes, cohort differences, or extra tutoring time.

That's like saying...

That's like noticing that umbrellas and wet sidewalks appear together and deciding umbrellas cause rain. The pattern might matter, but the cause still has to be established rather than assumed.

Caveat

This label is easy to overuse. The point here is not to call every weak argument by this name, but to reserve it for the exact misstep it describes.

Common misapplication

Do not use this label just because someone mentions a correlation. Correlations can be valuable clues and can support causal reasoning when mechanism, timing, controls, and alternatives are handled well.

Use the label only when...

Use this label only when someone treats a correlation, coincidence, or time pattern as if it already established that one factor caused the other. If the real problem is that a feedback loop is treated as if it fully explains, proves, or justifies a result, even though the loop may be contingent, breakable, or not sufficient for the claimed conclusion, the better label is Circular cause and consequence.

Often confused with

These near neighbors are easy to mix up, so use the comparison to see the exact difference.

Comparison

Circular cause and consequence

Why people mix them up: Both often look like causal mistakes at first glance.

Exact difference: Correlation is not causation happens when someone treats a correlation, coincidence, or time pattern as if it already established that one factor caused the other. Circular cause and consequence happens when a feedback loop is treated as if it fully explains, proves, or justifies a result, even though the loop may be contingent, breakable, or not sufficient for the claimed conclusion.

Quick split: What evidence actually rules out coincidence, reverse causation, or a third factor? Then compare it with What evidence actually rules out coincidence, reverse causation, or a third factor?

Comparison

Post hoc ergo propter hoc

Why people mix them up: Both often look like causal mistakes at first glance.

Exact difference: Correlation is not causation happens when someone treats a correlation, coincidence, or time pattern as if it already established that one factor caused the other. Post hoc ergo propter hoc happens when someone infers that because one event happened before another, the earlier event caused the later one.

Quick split: What evidence actually rules out coincidence, reverse causation, or a third factor? Then compare it with What evidence actually rules out coincidence, reverse causation, or a third factor?

Visual argument map

This map shows where an observed pattern gets promoted into a stronger causal story than the evidence can support.

Observed pattern

A school adds an AI tutoring tool and test scores rise the next semester, so the district concludes that the tool caused the improvement without ruling out teacher changes, cohort differences, or extra tutoring time.

Claimed cause

The leap happens when someone treats a correlation, coincidence, or time pattern as if it already established that one factor caused the other.

Missing checks

Correlations matter because they can reveal patterns worth investigating. The fallacy is jumping straight from 'these moved together' to 'this caused that' without checking mechanisms, timing, controls, or alternative explanations.

Safer conclusion

What evidence actually rules out coincidence, reverse causation, or a third factor?

Practice And Repair

Extra teaching tools that show why the fallacy is persuasive, what to look for, and how to correct it.

Why it matters

Why this mistake matters

Correlation is not causation threatens rationality because someone treats a correlation, coincidence, or time pattern as if it already established that one factor caused the other.

Main reasoning problem

Someone treats a correlation, coincidence, or time pattern as if it already established that one factor caused the other.

Why this kind of mistake matters

It makes one causal pathway feel established before alternatives, confounders, and directionality are tested.

Try it

What this walkthrough follows

Step through where the example moves from a pattern or event to a stronger causal claim than the evidence can bear. It moves through Pattern, Jump, Missing test, and Repair so the reader can follow the example from setup to diagnosis to repair.

Where the slip happens

The reasoning goes wrong when someone treats a correlation, coincidence, or time pattern as if it already established that one factor caused the other.

How to use the stages

Move through Pattern -> Jump -> Missing test -> Repair in order, compare each stage to the current example, and stop as soon as the later step adds more than the earlier one has earned.

What the repair step asks

What evidence actually rules out coincidence, reverse causation, or a third factor?

Causal jump walkthrough

Step through where the example moves from a pattern or event to a stronger causal claim than the evidence can bear.

What is actually observed

The example starts with this observed pattern or event: A school adds an AI tutoring tool and test scores rise the next semester, so the district concludes that the tool caused the improvement without ruling out teacher changes, cohort differences, or extra tutoring time.

What this walkthrough highlights: The app marks any unsupported causal arrow and shows which extra observations would discriminate among paths.

Current example: A school adds an AI tutoring tool and test scores rise the next semester, so the district concludes that the tool caused the improvement without ruling out teacher changes, cohort differences, or extra tutoring time.

Check yourself

The assessment area now uses mixed 10-question sets, so the fallacy is not announced in the title before the quiz begins.

What the assessment does

You will work through a mixed set of fallacy-identification questions. Focused links from a fallacy page will quietly include this fallacy among nearby look-alikes without announcing the answer in the page title.

Open the check

Open a mixed Check Yourself set that includes this fallacy.

Questions to ask

Use these category-based prompts to audit similar arguments.

Prompt 1

What evidence actually rules out coincidence, reverse causation, or a third factor?

Case studies

Each case study explains why the example fits the fallacy and links back to its source whenever source information is available.

Noncitizen voting, already illegal in federal elections, becomes a centerpiece of 2024 GOP messaging

AP's May 18, 2024 overview of noncitizen-voting rhetoric documented how a politically useful intuition about election fraud kept being treated as if it were established by the evidence. The report is especially useful for seeing how tiny counts, suggestive language, and moral urgency can be stretched into system-wide claims. The fallacy here is Correlation is not causation: someone treats a correlation, coincidence, or time pattern as if it already established that one factor caused the other. That matters here because correlations matter because they can reveal patterns worth investigating. A better analysis would remember that the fallacy is jumping straight from 'these moved together' to 'this caused that' without checking mechanisms, timing, controls, or alternative explanations.

Associated Press · 2024-05-18

Fact-check: Trump keeps claiming noncitizen voting is a big problem. It's not

NPR's October 12, 2024 fact check on noncitizen-voting claims is a good case study in the gap between isolated anecdotes and population-level conclusions. It shows how a few suspicious stories can feel decisive even when the base rates and verified counts point the other way. The fallacy here is Correlation is not causation: someone treats a correlation, coincidence, or time pattern as if it already established that one factor caused the other. That matters here because correlations matter because they can reveal patterns worth investigating. A better analysis would remember that the fallacy is jumping straight from 'these moved together' to 'this caused that' without checking mechanisms, timing, controls, or alternative explanations.

NPR · 2024-10-12

FACT FOCUS: Here's a look at some of the false claims made during Biden and Trump's first debate

AP's June 27, 2024 fact check of the first Biden-Trump debate is a dense collection of real argumentative shortcuts: statistics pulled loose from context, emotionally loaded immigration claims, and repeated assertions that did more rhetorical than evidential work. It is one of the best single-source stress tests in the library. The fallacy here is Correlation is not causation: someone treats a correlation, coincidence, or time pattern as if it already established that one factor caused the other. That matters here because correlations matter because they can reveal patterns worth investigating. A better analysis would remember that the fallacy is jumping straight from 'these moved together' to 'this caused that' without checking mechanisms, timing, controls, or alternative explanations.

Associated Press · 2024-06-27

AI seen cutting worker numbers, survey by staffing company Adecco shows

Reuters' April 5, 2024 report on the Adecco survey is a good reminder that expectations about job loss are not the same as demonstrated causal outcomes. It is useful wherever people slide from speculative trend talk to a confident story about what one technology will inevitably do to the labor market. The fallacy here is Correlation is not causation: someone treats a correlation, coincidence, or time pattern as if it already established that one factor caused the other. That matters here because correlations matter because they can reveal patterns worth investigating. A better analysis would remember that the fallacy is jumping straight from 'these moved together' to 'this caused that' without checking mechanisms, timing, controls, or alternative explanations.

Reuters · 2024-04-05

Claims about teen mental health and social media often move too quickly from parallel trend lines to a settled single-cause story, even though researchers still debate the size, direction, and mechanisms of the effect. The fallacy here is Correlation is not causation: someone treats a correlation, coincidence, or time pattern as if it already established that one factor caused the other. That matters here because correlations matter because they can reveal patterns worth investigating. A better analysis would remember that the fallacy is jumping straight from 'these moved together' to 'this caused that' without checking mechanisms, timing, controls, or alternative explanations.

Related fallacies

Nearby entries chosen by shared categories and family resemblance.

LogFall