This isn't a good signal imo. The world is undergoing a lot of cultural mixing at the moment so making the assumption that someones last name indicates their cultural participation is a dangerous one. For instance, I have a very French name but I also have near zero French cultural implications.

Eh, I think from a Bayesian standpoint its valid to think that if somebody has a last name that correlates with a group that doesn't drink, and that person also argues against drinking that they're more likely to be a member of that group.

But that's the key part "more likely", and not something definitive. It's the thinking that that inference is definitive is where it crosses the line from trying to be understanding about another culture to becoming a racist argument.

I agree though that its definitely flirting with that dangerous assumption.

That's why I said "not that I know". The parent comment said that people may speculate that someone who refuses to drink might be alcoholic, but there are also people who don't drink for religious reasons, obviously. In any case, people can decline a drink for any reason at all.