Acemoglu and Robinson have recently done much to academically revive the view that democratization is a struggle between pro-democratic masses and anti-democratic elites. If elites accept democracy, they do so to forestall even worse outcomes like revolution. (Here's a readable overview by Tim Harford).
Today Allan Drazen presented an interesting paper at GMU that began by questioning the assumption that masses are uniformly democratic. But the most interesting part of the paper, for me, was his Table 1A (scroll towards the end of the paper). In it, examines the determinants of five "pro-democratic" beliefs in the World Values Survey:
1. Democracy is good
2. Maintaining order is important
3. Freedom of speech is important
4. Evaluation of democracy's progress
5. Superiority of democracy
Striking result: Higher income people are markedly more pro-democratic on all five questions. (Side result: Men are more pro-democratic on all five questions too!)
I guess that Acemoglu and Robinson could claim that their "elites" are so elite that they don't appear in surveys, but frankly it looks to me like one of their core assumptions is just wrong. "Elites" are a convenient scapegoat for persistent dictatorship and unstable democracy, but that's about it - a conclusion that is not surprising when you consider the overwhelming evidence against the premise of political selfishness.
But aren't Drazen's results also a blow to my elitist skepticism about democracy? Not really. In the World Values Survey, the implicit alternative to democracy is authoritarian dictatorship, not liberty.