Causation: notice when a claim says one thing produced another

When you read “X caused Y,” first make the proposed claim explicit: What changed, what result is attributed to it, and in what setting? The task here is to inspect the claim and its support, not to prove a cause from a household anecdote.

The repaired display and the new hooks

In this invented scene, a family hangs paper artwork on a community display. Last week, several pieces fell. This week, none did. One adult says, “The new hooks fixed it.”

Before repeating that explanation in the event newsletter, Tessa lists what changed. The hooks are new, the artwork is lighter, and the display is now indoors. She can confirm that the pieces stayed up during this event. She has not isolated which change accounts for that observation.

Tessa's partner proposes a cautious sentence: “The artwork stayed up at this indoor event after several setup changes.” Their child adds that they like the new display's height. That preference does not need to be part of the causal claim. The family can enjoy the result while leaving its explanation open.

This fictional example is deliberately unfinished. It does not secretly establish that the hooks, the paper, or the room caused the difference.

Find the claim's moving parts

For an article or product statement, you might note:

• The proposed cause, described specifically rather than as “the new thing.”
• The outcome said to change, including how it was measured.
• The comparison used to support that statement.
• Other changes or explanations the report discusses.
• The limits the authors put on their conclusion.

Look for the passage that connects the evidence to the explanation. If the article only gives events in sequence, ask what else supports the stronger claim. If the author uses cautious language, preserve it when you summarize. If you cannot reach the underlying report, do not present the proposed cause as something you personally verified.

Causation, correlation, and confounding

These entries offer different reading tasks. With causation, identify the proposed effect of one thing on another. With correlation, identify the relationship between measurements. With confounding, examine a proposed shared cause that could complicate that relationship. The randomized-controlled-trial entry helps you look for a particular assignment method in a study report.

Do not turn those labels into a rule that only one kind of document deserves reading. Ask what this source claims and whether its reported methods address that claim. A short glossary cannot settle the validity of a complicated causal analysis.

Explain an observation without assigning blame

In family conversations, keep “what happened” separate from “whose fault it was.” A child can describe that something fell without being required to identify the cause. A parent can correct an overconfident explanation. Partners can agree on a practical setup for next time without pretending their choice proves why the previous setup failed.

This page is a reading aid for ordinary claims. It does not provide a basis for diagnosing a person, attributing a health condition to a family member, or changing care.

Source

[1] Australian Bureau of Statistics. Correlation and causation. Released February 2, 2023; accessed September 5, 2026. “The difference between correlation and causation,” causation paragraph, supports [1]. The display example is original fiction.

Related guides

• Correlation
• Confounding
• Randomized controlled trial
• Observational study