> On the other hand, if your experiment is expecting one correlation, and instead you find another... then the p-value hunting was the correct move...? at best

> On the other hand, if your experiment is expecting one correlation, and instead you find another... then the p-value hunting was the correct move...? at best you'd have to make corrections to your p values to account for the fact you're conducting multiple tests. those corrections weaken the results. (i did not check this paper to see if they did the corrections.)