I once heard of a study that compared the crime rate with the number of refrigerators sold. The curves, plotted against time, were perfectly correlated thus proving that crime was being caused by refrigerators!

Reminder: If two things, A and B, are correlated it can mean that either: A causes B; B causes A; A and B are both being caused by something else; or coincidence.

Another vital part is that the mechanism needs to be understood. What is the mechanism for A causing B, exactly? Clearly, the example in this post fails the last test and maybe another one as well.