Hmm, this is a common mantra, but I'm not sure how useful it is. For sure a formal experiment with controls will give you causation (if it exists) but a lot of the time we do not have formal experimental data.

In the smoking wars there were a lot of researchers who had found a correlation between smoking and lung cancer.

Naturally enough lawyers for the tobacco companies said "you have not proved anything!". Some said there was a third hidden variable, a personality thing, and people who had a personality which tended to make them smoke also had the genes for lung cancer and would have got lung cancer anyway and the data did not prove that this was not the case.

Only a formal experiment where we take a bunch of people and divide them AT RANDOM into two groups and force one group to smoke for 20 years and the other to not smoke would be sufficient to provide the causative information.

So this experiment was never done (ethics) so the link between smoking and cancer was not proven - but you still have to decide

Are you going to smoke?

Well, you do have the correlation.

Also, you might consider that human lungs were probably not designed for continually sucking in smoke - there is a feeling of irritation.

Also, when dead smokers are autopsied the lungs are kinda dark and gungy compared to a non smokers lungs - so something is happening.

And lung cells in a dish become cancerous when exposed to compounds from cigarette smoke.

etc etc etc.

Given these observations it is REASONABLE to believe that smoking causes lung cancer without there being formal statistical proof.

Yes, if you find a correlation between two variables which common sense (whatever that is) tells you are probably unrelated and then you prance around saying you have 'discovered' something then you are wrong to do that - but that is just being an idiot surely?