I apologize if this is answered somewhere I am just missing, I did search (and am surprised to see no mention of it).

Statistics teaches us that while it is often relatively easy to prove the Correlation of variables, this does not prove Causation. In fact, Causation can be quite an elusive thing to prove; Often it just comes down to, "I cannot think of any third variable that could connect these".

So what is best practice for Skeptic.SE?

Is it good practice to mention in your answer that while these studies have shown X is correlated to Y, that does not prove causation? OR specifically that they did not account for variables A, B, and C which might of had an effect?

Is it worthwhile to comment on someone else's answer, "Correlation does not equal Causation, what if the change in Y is actually being caused by Z?". Are these theoretical arguments worthwhile even without data backing them up?

Skeptic.SE is about proving things with evidence. But if we do not include some evidence of Causation, in the applicable questions, how are we offering any evidence at all? Should Skeptic.SE perhaps be just as firm that answers address the Correlation vs Causation problem as we are that answers include citation to evidence?

| |

On skeptics.se it's good practice to show the current scientific consensus on the matter at hand, whether it is:

  • currently unknown correlation and causation (hard to prove, but certainly possible)
  • disproven correlation and causation
  • proven correlation, but unproven causation
  • proven correlation and disproven causation
  • proven correlation and causation

This is to say: we look for the available evidence and report it, but we certainly don't confuse correlation and causation.

| |

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .