Correlation does not equal causation.

Ok, I'll get this out of the way now:

People who taught you that correlation does not equal causation lied to you!

Now, if you care to know more I shall elaborate:

First, someone tried to differentiate between correlation and association. They can be used interchangeably when you are using correlation to mean association (in which case there is no demonstration of cause). That is, you can say that "gender is correlated with weight." You have to code gender as numbers (1 and 2, for instance) but you can easily do a correlation on that.

The argument for correlations not equaling causation comes down to how the data was gathered, not the test used. So we generally mean that correlational research designs don't show cause (because their could be other extraneous variables). Even in this case, however, it is possible to control for most extraneous variables. These studies may not be 100% causal, but they can be pretty damn close.

If I demonstrate a perfect correlation between the number of times you say "fart-nugget" and the amount of speeding tickets you get in a year, then there may be no cause there, but you sure as hell shouldn't say "fart-nugget" unless you want a ticket.

Finally, a correlational analysis (e.g., Pearson's r) on experimentally obtained data can be used to show cause. In fact, any statistical test on experimentally obtained data can be used to show cause.

The moral: If you read that two things are correlated, don't spew out that correlation does not imply causation because there are exceptions.

(Please note that I'm not saying that a correlation betwen religious faith and antisocial behaviors means that religion causes the antisocial behavior or that antisocial behavior causes religious beliefs...just talking about semantics of that annoying armchair statistics catch-phrase)