The fun of correlation = causation...

I get the feeling that the non-drinkers are more obsessed with their work than the drinkers, therefore they get published more.

That and they tend to not have lives.

People like to mock these silly experiments, but that's how science is done. It's never so much a revealing of the truth, as it is a slow process of chipping away at the unknown.

Of course, if the study determined that beer drinkers were all geniuses, everyone would be hailing the study as the direct word of God Almighty.

Well, to some extent that's how science is done. High correlation leads to further studies to determine causation. Or it should lead to this in a perfect world.

But this is no different than the whole "working 3rd shift causes cancer". It's not a root cause nor does it actually demonstrate anything other than a need for further investigation.

good thing I'm not a scientist!

Yes - you just work with our money.

wait. Mebbe that's not a good thing...

Thing is, we can never be absolutely certain about causation. We can only find enough correlations that we can say with a measurable degree of confidence that, if X is done, Y is nearly certain to follow. Experiments can be used to increase the degree of confidence by testing for previously untested possibilities, say determining if people of a certain gene sequence and upbringing turn out stupider after drinking beer than people with a similar gene sequence and upbringing, but there may be some genetic factor that converts beer directly into brain cells and results in an individual who becomes more intelligent with every beer he / she drinks.

Even the laws of thermodynamics may turn out false, or insufficient to describe a new phenomenon, though the implications would force a re-evaluation of almost every other scientific theory developed since the days of Archimedes and his bathtub. Even the most solid scientific theory is nothing more than an enormous mass of correlations that give credence to a causation, which is why each one must have some means to disprove it, or there is no means to prove it right by performing experiments that would yield different results if it was wrong.