Those graphs show no error bars. It is a no go for a real, serious science. Also, there is no mentions of sampling sizes.
However, if it is indeed statistically significant, I would really like to see the theory behind !
-----
Edit - I checked the original papers and it may reaaly be a case. Hmmm...
Moderate drinking is benefitial to the health. It is a very well established fact now. But how about low-dose smoking ??? Is it possible that one or two sigarets per months can actually make you better off ?
[ October 20, 2003, 22:20: Message edited by: oleg ]