I've also heard that the control, those who don't drink, are often not drinking because they used to be alcoholics thus scewing the longevity stats for non-drinkers.
The red wine vs beer discrepancy could be down to wealth.
The drinker vs non-drinker could be down to the fact that seriously ill people fall into the non-drinker category.
Most people who drink mostly do so whilst getting together with friends.