A cartoon cited in Larry Bartels' classic "Homer Gets a Tax Cut."

Who does Congress represent best: the rich, the middle class, or the poor?

That question may sound absurd on first blush. Of course Congress represents the rich, and couldn't care less about the rest of us. And that may be right. But a new study casts doubt on that assumption.

The case that Congress is bought

Disgraced former Rep. William Jefferson (D-LA), who stashed gobs of money in his freezer. (Molly Riley-REUTERS)

There's a substantial segment of the research literature to back that suspicion up. Larry Bartels, a political scientist now at Vanderbilt, made major contributions to this area of research in a 2005 article and his 2008 book "Unequal Democracy."

Both find that representatives tended to mirror the ideological leanings of their rich constituents, with Republicans in particular giving the affluent a preference. "Republicans were about twice as responsive as Democrats to the views of high-income constituents," he found in the 2005 study, but added: "There is no evidence of any responsiveness to the views of constituents in the bottom third of the income distribution, even from Democrats."

Princeton's Martin Gilens has found very similar results, and condensed his research last year into the book "Affluence and Influence." "Policies favored by 20 percent of affluent Americans, for example, have about a one-in-five chance of being adopted, while policies favored by 80 percent of affluent Americans are adopted about half the time," Gilens writes. "In contrast, the support or opposition of the poor or the middle class has no impact on a policy’s prospects of being adopted."

And then there's the fact that a majority of elected officials in the United States are millionaires, as Duke's Nick Carnes has pointed out, which means that even if monied interests had no influence whatsoever, members of Congress may be inclined to better represent other rich people. All together, it suggests that legislators are pretty far out of sync with their constituencies.

Reasons for optimism

So we don't live in the Paris Commune. But do the masses still have a say?

Or does it? A strand of research has emerged since Bartels' findings that casts doubt on conclusions like these. Columbia's Robert Erikson and the University of Copenhagen's Yosef Bhatti replicated Bartels' study with a much larger data set and, like him, found that the only variable with a statistically significant effect on representatives' ideologies is the ideological character of high-income voters. But they found that classes disagreed with each other less than you might suppose. The difference in ideology between high-income and low-income people was not statistically significant. That suggests the results may be less meaningful than Bartels initially thought.

Now, three economists are presenting the most sustained critique of the Bartels-Gilens finding to date. Eric Brunner, Stephen Ross and Ebonya Washington argue that less income does not necessarily mean less representation. They argue that the basic method used by Bartels (update: worth noting this is one of several approaches Bartels uses) is completely flawed. Just because a representative and a constituent have the same score on a scale of ideology doesn't mean that constituent is being perfectly represented. Maybe the senator is super-liberal on economics and conservative on social issues, and the constituent is the opposite. That averages out to the same score, but the senator is actually the constituent's worst nightmare!

Brunner, Ross, and Washington propose a different way to go about answering this question, one which doesn't have that problem. They note that between 1991 and 2008, there were 77 different ballot propositions in California that also faced a vote in the legislature. That allows for a direct test of whether legislators are representing their constituents, since you can compare views on discrete issue areas.

They find that most of the time, legislators vote according to what both rich and poor constituents want. That's because there's mass agreement on many of these issues across the income divide (as Erikson and Bhatti found). But on average, Democratic legislators better represent low-income voters and Republicans better represent high-income ones. Because California is a Democrat-dominated state, that works out on average to a slight advantage for low income voters.

But the authors also find that income isn't an important variable once you take the party composition of districts into account. It's just that low-income people are likelier to be Democrats and high-income people are likelier to be Republicans, and legislators vote more in sync with their support base. "Republican legislators appear more responsive to the views of their higher income district residents and Democrats to their lower income constituents, not because these voters are high or low income, but because these constituents are highly partisan," they write.

Does the challenge succeed?

But yeah, Congress is still getting fliff like they're sultans. (Brad Neely / Creased Comics )

So, is Bartels debunked? Hardly. The Brunner/Ross/Washington paper has many virtues, and the data set it uses is innovative, but it's also problematic. For one thing, it's unclear that results specific to California have validity for studies of representativeness in the rest of the country. What's more, the issues in question are all state-level, and thus may be less likely to involve class politics than national issues. Finally, the study depends on the vote results for ballot propositions being an accurate representation of public opinion. Given that we know that rich people are much likelier to vote than the poor, that may not be the case.

It's also worth noting that Gilens avoids the problem that the new paper diagnoses in Bartels. He tracks individual pieces of legislation rather than focusing on ideological scores. And he finds the same thing as Bartels: the views of the rich get enacted. The poor hated the MX missile and the rich weakly favored it; they were deployed. The rich like free trade, the poor opposed it; NAFTA and GATT were enacted. That suggests the methodological problem identified by Brunner et al isn't as significant as they suggest.

Two caveats are in order, though. Gilens' work is sometimes interpreted as showing that Republicans are out of touch with the American people, but it also shows that Democrats' social views are much more representative of rich elites than most people. The rich tend to have laissez-faire attitudes toward abortion, same-sex marriage, and school prayer, while the poor are more conservative. The former attitude is better represented than the latter.

Second, representativeness may not be the be-all and end-all of a democratic system. Mark Schmitt, responding to Gilens' book, noted that the nadir of representativeness of all income groups, according to Gilens, was Lyndon B. Johnson's presidency, which featured the enactment of the Civil Rights Act, Voting Rights Act, Medicare, Medicaid, and the rest of the Great Society, and the peak was during George W. Bush's first term. Everyone hated Johnson's agenda, while Bush's enjoyed broad support, including from the poor. That just goes to show that just because the poor want something doesn't mean that it actually advances their interests.