This smells of p-hacking to me. Only 2 of 6 tests showed statistical significance and it was a relatively modest effect. Not impossible that there's an actual

This smells of p-hacking to me. Only 2 of 6 tests showed statistical significance and it was a relatively modest effect. Not impossible that there's an actual causal link. But I'm not quite buying yet given the amount of normalization they did and size of the effect.