Holy cause and effect issues, Batman!
First, almost all of the factors listed by Miko (crime, illegitimacy and deviancy) linearly increase with population density. They also linearly decrease with increasing income. The chances are pretty good that you're not going to see a lot of petty crime among people making substantial sums of money, after all. I'm also not sure why the author focuses on ideology rather than partisan affiliation, since party plays a much stronger role in vote choice than ideology. But I digress.
I just grabbed the National Elections Studies cumulative 1948 to 2000 dataset with over 40,000 respondents and regressed ideology (extremely liberal through extremely conservative on a 7-point scale) on income (listed by percentile ranges, scored 1 to 5). For those that know regression, I'm well aware that the dependent variable is not truly continuous, but it's close enough for AH BB work. I find, not unexpectedly, that a one point improvement in income percentile results in a 0.8 point positive change on the ideology scale (t = 9.219, p < 0.001). In essence, the more one makes, the more conservative one becomes.
Why should we be shocked and amazed, then, that those in the lowest income percentiles -- commonly "liberals" living in high density, high crime urban areas -- would support candidates with policy views similar to their own?
Talk about making a mountain out of a molehill. I'd like to see Professor Olson sell that smoldering pile to a group of political scientists.
-- Todd/Leviathn