Kevin Drum, summarizing research by Larry Bartels:
Using data from voting records in the early 90s, it shows that the responsiveness of senators to the views of the poor and working class is....zero. Or maybe even negative. And that's true for both parties. The middle class does better — again, with both parties — and high earners do better still. In fact, they do spectacularly better among Republican senators. And this disparity has almost certainly gotten even worse over the past two decades.
I haven't read the paper yet, so I don't know what the caveats are. But if this sort of result holds up it calls a lot of political economy models into question. I may comment more if I find time later in the week (I'm studying for comps, so blogging time is limited).