# What's the difference between causation and correlation?

ACB

Sun 26 Jul, 2009 03:44 pm
@kennethamy,
kennethamy;79639 wrote:
But how does that show that we do not infer causation from correlation? The inference, of course, need not always be a valid inference. We require more than mere correlation for a valid inference to causation, since although correlation is a necessary condition of causation, it is not a sufficient condition. We require controlled studies to determine that the correlation is not just accidental correlation, and, finally, some causal nexus between the cause and the effect that explains the correlation. But, correlation is a sign of causations, and a necessary condition of causation. No one claimed that correlation and causation should be identified. Even if the two clocks strike the hour simultaneously, there need be no causation, and in fact, there is none, since the striking of the two clocks is not causally connected, but they both strike the same hour because of a common cause. The same is true of night and day. Night does not cause day, nor day, night. But they are both the effects of a common cause.

Yes, I entirely agree. I was just qualifying prothero's post #418. Sometimes we infer causation, sometimes we don't.

kennethamy

Sun 26 Jul, 2009 06:35 pm
@ACB,
ACB;79665 wrote:
Yes, I entirely agree. I was just qualifying prothero's post #418. Sometimes we infer causation, sometimes we don't.

That we do infer causation is one thing. But that we ought to infer causation is a different thing. We may sometimes infer causation when that would be committing the fallacy of post hoc ergo propeter hoc (or "false cause".But whether we ought to infer causation depends on whether we have good reason to do so (controlled studies and the like). Logic, said the American philosopher, Charles S, Peirce, is a "normative science". The rules of inference are normative, because rules are normative. Logic studies not how we do think, but how we ought to think.

prothero

Sun 26 Jul, 2009 11:25 pm
@Satan phil,
Well just for the record. I did not say everytime we observe correlation that we must or should infer causation. I said one was observed the other was inferred.