Adding in American, no children, works with computers, and less than 30 drops it to 10%; education is a bit tougher to code (ideally, you would want to index it by age). Most of the work is being done by "American," "less than 30," and "works with computers." (No children, conditioned on being less than 30, does almost nothing.)

While this is usually true, I suspect that these measures will not all be independent (i.e. there will be some people who are outsiders to Less Wrong in many ways) and that combined with the fact that there are huge supermajorities (rather than simply majorities) of these populations will mean that most LWers are in fact highly typical.

Either way this doesn't address the core claim of whether these characteristics select for Libertarianism. I don't have the time or experience with spreadsheets to figure this out immediately but we could certainly look at that as an empirical question. Looking back my earlier reply was really outside the spirit of your first post so I apologize for that.