Let me propose another theory. In the awesome book "How not to be wrong" ellenberg walks through a scenario with dishonest roulette machines as a way to explain

Let me propose another theory. In the awesome book "How not to be wrong" ellenberg walks through a scenario with dishonest roulette machines as a way to explain Bayesian Inference and shows how powerful your priori are in interpreting new evidence that you are shown. If you hold a non neutral viewpoint on something, natural variations of a neutral truth will take an extremely long time to adjust your view of reality. It just so happens that most highly educated people get their initial priori set from the institutes of learning that have increasingly become sounding chambers for liberal ideologies. So its no surprise that it takes a preponderance of evidence to migrate those views back towards a neutral reality ( if in fact reality is neutral ). A similar situation can be established for rural and uneducated people. They get their priori set from their parents and peers who have a very distinct view of government and governance. All the other talk I see of reasons for the homogenization of the educated class into a specific political view just reads like chest thumping to me.