This data-manipulation is outrageous but expected in a way.
Isn't there a common saying that "only trust the study you faked yourself"?
I think this is totally true in this case.
I think you once made a podcast about studies where you showed that you can proove absolutely anything with a study.
It's like:
More accidents with sober people than drunks.
More accidents on daytime than nighttime.
More accidents when driving slow than driving fast.
So drive drunk very fast in the night and you are statistically safer.. riiight..