LogFAQs > #952207079

EmbraceOfDeath posted...

It is. The scientific method is based on trying to disprove your hypothesis by proving the null hypothesis. It's only if you can't prove the null hypothesis that confidence in your original hypothesis is then strengthened.

Several things.

1. This is a bit unrelated to what that poster said. Usually people start from what is well known and accepted because they have very strong priors. That is the opposite of "distrust in accepted thought"
2. That's not really the way falsification works. While it's generally a good way to do science, the reason why it's clearly not the only way to gather evidence is highlighted by your error. You never "prove" the null. In a hypothesis testing framework, you find evidence that is either consistent with the stated null, or inconsistent. Just as you can show that evidence is consistent with the null, you can show that it is consistent with a stated hypothesis, and both conditions carry evidentiary weight. I have a really good reference that discusses this pretty exhaustively if you are interested to read more about it.

---

=E[(x-E[x])(y-E[y])]