Please find the (apparently obvious) problem with the results. I keep seeing articles about how study after study confirms the alcohol/longevity link, so if you can find the common problem those studies all have, (very minor) fame awaits! :)
Google "Quasi-experimentation."
not drink for health reasons
Yes this sounds plausible. Even if the person isn't unhealthy maybe their family history is such that they still don't drink.
have a less active social life etc
As a semi-introvert[1] I always wonder how true this is for the minority of us that much prefer this so called social isolation. I'd take a night alone or with a small group of friends over a bar or conference absolutely every time always. Of course that is assuming there are no other external influences - which is rarely (I'd say never but we're on HN...) the case. All that really matters is whether you are suffering from anxiety rather than loneliness[2].
[1] I'm XNTP somewhere between INTP and ENTP on the meyer brigs http://en.wikipedia.org/wiki/Myers-Briggs_Type_Indicator
[2] I can't remember the last time I was lonely or bored. Hmmm, is this really that unusual amongst similar introverts? I just consider myself somewhat curious and self-motivated.
And again, maybe the reason non-drinkers show as much longevity as they do is because those are the ones who have an increased resistance to the negative effects of abstinence. Maybe most of them die much younger. Since non-drinkers are such a minority in most places, this is actually more likely, since the die-off would be less noticeable than if everybody who drank an occasional beer dropped dead at age 40. (I still don't think this is likely either, incidentally.)
Don't forget people who have died as a result of accidents while intoxicated (driving drunk, etc.) or have been incarcerated for drunken infractions. They may be excluded from those statistics, adding a kind of bias. Not sure if the total is high enough to impact the study's results, but it's worth checking.