I'm sorry, maybe I'm misinterpreting you. Are you suggesting that the biases I'm noticing are motivated by accurate assessment of the groups in question?

I think you're making a sort of Bayesian argument for racism, and I can see how consumption of a particularly odious media diet could cause that. But I don't buy the idea that it's rational to hold racist views toward, say, Barack Obama and call him names.