In case anyone interested hasn’t seen it, this week’s Nature has an article stressing the need to get away from the usual significance tests we see in virtually all the junk food science that gets published. Expressed as “if p<0.05 it must be significant”
https://www.nature.com/articles/d41586-019-00857-9
Back last Summer, when Brian Wansink, the King of Junk Food Science was being pushed out from Cornell’s food science lab, Five Thirty Eight did a mock version of the kinds of spurious correlations that food studies often find. They found these strikingly good correlations.
If all you think matters is that p<0.05, eating cabbage correlates to having an innie bellybutton and drinking Lemonade correlates to believing “Crash” deserved to win the Best Picture Oscar. There’s lots more here.
I’m mostly hoping that with Brian Wansink gone, maybe some sanity will show up in the food sciences. I just kinda doubt it.