Originally Posted by xsdg

As someone who knows something about statistics, I have to note that this is complete and utter BS. I'll assume you haven't heard of the statistical maxim that "correlation does not imply causation"? Well, just because two statistical measures are correlated does _not_ necessarily mean that you can change one thing and expect the other to change accordingly.

As an example of this concept, suppose that a man claps each time he sits in his chair. Additionally, his pet bird chirps once whenever his chair squeaks (which happens to happen each time he sits in it). The number of claps measured each day will likely be strongly correlated with the number of times the bird chirps each day. An improper inference would be to say that the man's clapping causes the bird to chirp -- if the man starts clapping every _other_ time he sits down, the bird's chirping will not be affected.

I won't go into more details here, but google will undoubtedly have a lot of links for you. Two good ones are http://www.stat.tamu.edu/stat30x/notes/node42.html and http://en.wikipedia.org/wiki/Correla...mply_causation .