[jonnathanson]

Yep. Give me five minutes, a pen, and some paper, and I can jot down a list of at least fifty potential confounding variables. I trust that someone, somewhere has strong and well-structured data, hinting at a causal relationship and a direction of causality. If so, I'd like to read that source.

Now, I'm not naive. I know that it would be damned near impossible to structure a longitudinal test that a) identifies all the significant variables, b) isolates the most appropriate ones, and c) controls for the least appropriate ones. But absent that sort of test, I'd like to read the tests that we have.

There may indeed be a connection between socialization and multi-factor health outcomes. Maybe even a causal link. I'm certainly willing to consider that. But let's avoid blanket arguments such as 'We've evolved to be social creatures, therefore, we need friends as much as food and water.' There are a lot of interesting ideas in this article, and the article weakens those ideas when it relies on folksy generalizations.