Thus far we have seen the related statistical sins of cherry picking and data dredging. Today, let’s talk about the false cause fallacy (or “false causality” for short), which occurs when you observe two events that appear together and then leap to the conclusion that one event must have caused the other. (Here is a mundane example. The video below presents many more.) In reality, just because two events occur together does not mean that one caused the other. The causation may run in the opposite direction or some unobserved third factor might be the underlying cause of both events or there might be no direct or indirect causation at all!
It bothers me when someone decides they are going to show the general population how foolish they are and come up with something the chocolate experiment because once you ring a bell, you can’t un-hear it. Every so often I come across someone who will say, eat chocolate it’s now good for you. People only remember the sensational headlines even when they are disputed after the fact. Just look at vaccines causes autism myth. No matter how many times it is reported to be false you still have anti-vacciners out there. The whole controversy was caused by a bad study.
I completely agree. As Laura Arnold is fond of saying, the four most dangerous words in the English languare are “A new study shows …” Here is her TED Talk: https://www.youtube.com/watch?v=Y2y7BzjbSNo
Pingback: Publication bias | prior probability