Gilens compiles a massive data set of public opinion surveys and subsequent policy outcomes, and reaches a shocking conclusion: Democracy has a strong tendency to simply supply the policies favored by the rich.  When the poor, the middle class, and the rich disagree, American democracy largely ignores the poor and the middle class. 

To avoid misinterpretation, this does not mean that American democracy has a strong tendency to supply the policies that most materially benefit the rich.  It doesn't.  Gilens, like all well-informed political scientists, knows that self-interest has little effect on public opinion.  Neither does this mean that Americans strongly object to the policy status quo.  They don't, because poor, middle class, and rich tend to agree.  Gilens' key conclusion is simply that when rich and poor happen to disagree, the rich generally get their way.

...

Both left and right are likely to misread Gilens.  The left will probably imagine that he's saying that American democracy is a vast conspiracy to promote the material interests of the rich.  To repeat, Gilens explicitly disavows this conclusion: His claim is not that American democracy primarily **benefits** the rich, but that it primarily **listens** to the rich. 

The right, on the other hand, will angrily reject Gilens' findings as rehashed Marxism in statistical garb.  (To quote The Communist Manifesto, "The executive of the modern state is but a committee for managing the common affairs of the whole bourgeoisie.")  If you read the whole book, though, you'll be amazed by how many leftist oxen he gores.  Most shockingly: Gilens concludes that the president most responsive to all Americans regardless of income was... George W. Bush!

...

I find Gilens' results not only intellectually satisfying, but hopeful.  If his results hold up, we know another important reason why policy is less statist than expected: Democracies listen to the relatively libertarian rich far more than they listen to the absolutely statist non-rich.  And since I think that statist policy preferences rest on a long list of empirical and normative mistakes, my sincere reaction is to say, "Thank goodness."  Democracy as we know it is bad enough.  Democracy that really listened to all the people would be an authoritarian nightmare.

Emphasis mine. Read the whole thing. Via Why Is Democracy Tolerable? Evidence from Affluence and Influence, Bryan Caplan | EconLog | Library of Economics and Liberty.