> And the point is
> this: What those "half a dozen other studies," as you put it, fail to take
> into account is the fact that people already ill tend to be nondrinkers.
> Their illness isn't caused by their failure to quaff a couple a day.  It's
> the other way around, they don't drink because they're ill.  This is a
> real statistical quirk which causes the U to turn into something more like
> a J

They could have factored ill persons out of the analysis (just as they
factored poor diet and cigarette smoking out the groups drinking a lot of
alcohol each week).  But they didn't; they just speculated that this ill
group of non-drinkers could account for higher cardiovascular disease.

Rob