How can I separate correlation from a cause-and-effect claim in an online post?
Quick answer
Check whether the post only shows that two things changed together, or provides evidence that one change caused the other. Ask which came first, whether other factors could explain the pattern, and whether the evidence compares suitable groups or tests an intervention. If those questions are unanswered, describe the relationship as an association, not a demonstrated cause. Separate what the evidence shows from what the post concludes.
The question, the details, the next step.
I see a post saying that people who use a certain app sleep less, so the app causes poor sleep. How do I tell whether that conclusion follows?
Start by separating the observation from the conclusion. The observation might be that app use and shorter sleep occur together in the people described. The conclusion is that app use causes shorter sleep. One does not automatically establish the other.
For this example, treat the post’s details as hypothetical unless it provides evidence. Ask: What was measured? Who was included? Did the post show that app use happened before the sleep difference? Did it compare similar people, or just report that the two features appeared together? If those details are missing, the post may support a question worth investigating, but not its confident causal wording.
A useful first note is: “The post reports a relationship between [app use] and [sleep duration]. That alone does not show that [app use] caused [shorter sleep].”
What other explanations should I check before I call it cause and effect?
Look for at least three possibilities. First, a third factor could influence both: for example, a hypothetical change in work schedule might affect both app use and sleep. Second, the direction could be reversed: people who are already sleeping poorly might use the app more. Third, the pattern might depend on who was observed or how the measures were collected.
These are questions to test, not claims that any one explanation is true. In your review sheet, write a possible alternative next to the post’s explanation, then note whether the post addresses it. If it does not, say so plainly: “The post does not explain whether [possible factor] could account for the pattern.” Avoid filling that gap with your own unsupported explanation.
Does seeing the pattern happen before and after a change prove that the change caused it?
Not by itself. A before-and-after comparison can help establish timing, but other changes during the same period might also explain the difference. Check whether the post describes a comparison that helps distinguish its proposed cause from those alternatives, such as an appropriate comparison group or a carefully described test. Then check whether the result applies to the people and circumstances the post discusses.
For instance, suppose a hypothetical post says sleep improved after an app feature was removed. That timing is relevant, but it does not tell you whether schedules, measurement, or another change also shifted. A careful summary would be: “Sleep reportedly changed after the feature was removed; the information shown here does not establish that removing it caused the change.” The evidence might support a stronger conclusion if the post provides further suitable evidence, but do not assume that it does.
Can you give me a short review sheet I can use on the post?
Copy this and fill in only what the post actually supports:
Claim in the post: [quote or paraphrase] Observed relationship: [which two things appeared together or changed] Evidence described: [what was measured, compared, or tested] Timing: [what happened first, if established] Possible alternatives raised by the post: [list, or “not addressed”] What remains uncertain: [specific missing detail] Careful wording: “The post shows/reports [observation]. It does not establish that [proposed cause] caused [outcome], because [specific gap].”
Worked hypothetical example: “The post reports that people using [app] slept less. It does not establish that the app caused shorter sleep: it does not explain whether sleep problems preceded app use or whether other factors were considered.” This wording does not prove the causal claim false; it keeps your summary within the evidence presented.
A practical checklist
Write down the observed relationship separately from the post’s causal conclusion.
Check whether the proposed cause occurred before the outcome.
Look for a plausible third factor or a possible reverse direction, and note whether the post addresses it.
Record what was measured and what comparison or test the post describes.
State the specific missing evidence and use “associated with” or “reported alongside” when causation is not established.
A few more useful answers
If two things are correlated, does that mean neither caused the other?
No. They could be causally related, but the correlation alone does not establish that. The claim needs evidence that helps distinguish causation from other explanations.
What should I say if the post gives too little information to judge?
Say what it reports, then name what is missing: for example, “The post reports an association, but does not provide enough detail here to assess whether one factor caused the other.”
Is a before-and-after comparison useless?
No. It can show that a change came before another observed change, but timing alone may not rule out other explanations. Describe what the comparison shows without treating it as proof by itself.