How one derives significance test for pearsons correlation coefficient?

49 Views Asked by At

I am exploring statistics and probability. What upsets me, only ready to use algorithms are present in the books. But no example how one derives a significance test, where from the test statistics come from etc. At the moment I am interesten in Pearsons coefficient significance test, but really derivation of any popular test will be helpful for me.

1

There are 1 best solutions below

0
On

A rigorous paper on the correlation test is here.

More generally, any hypothesis test is derived using a very basic strategy:

  1. Formulate the null hypothesis $H_0$ as a statement about a statistic
  2. Pick an underlying distribution for the data (or decide you are doing a nonparametric test)
  3. Determine the sampling distribution of the statistic if we assume $H_0$ is true.
  4. Determine the rejection region from the sampling distribution.

How do you choose a test statistic? That is the art of designing a good test, and always an area of active research!