> For me, the takeaway from this affair is that there is no one-size-fits-all solution to make statistics impossible to hack. Getting rid of p-values is appropriate sometimes, but not other times. Demanding large sample sizes is appropriate sometimes, but not other times. Not trusting silly conclusions like “chocolate causes weight loss” works sometimes but not other times. At the end of the day, you have to actually know what you’re doing. Also, try to read more than one study.