I wonder if other things are reproducible. Is the "1 alcoholic drink" thing been replicated when correcting for income? Lots of rich people may have 1 drink a day, and so this pushes up the average of the "1 drink a day people", but it doesn't mean you can replicate it.