Getting 80000 measurements and then accepting a 1-in-20 false positive rate, however, sounds like P-hacking.