0
$\begingroup$

Why does a test return the p value?

I am looking at the Diebold-Mariano-Test (DM Test): https://github.com/johntwk/Diebold-Mariano-Test

Basically, this test gives me a DM statistics, and also the p value. So, do I just the p-value to judge a null hypthoesis?

For instance, if p-val > alpha = 0.05, then I reject the null hypothesis, is this correct?

$\endgroup$
4
$\begingroup$

Actually, it's not correct.

In general, the null hypothesis is rejected if the p-value comes out small.

That's because the p-value is small when an observed data point $P$ (a result of a test, for example) is (far) in the tail of the null-hypothesis-based distribution.

In other words, the probability of observing such or more extreme a data point according to the null-hypothesis is very small. Hence, having reasons to believe that we've observed more or less typical point, we reject the null.

| cite | improve this answer | |
$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.