If you think American politics is a rigged shell game, where the views of the rich and powerful count but those of regular Americans don’t, then two prominent political scientists agree with you.
Three years ago, Martin Gilens of UCLA and Benjamin Page of Northwestern released a paper with a startling, disturbing conclusion: The opinions of rich people affect public policy, but the views of average, middle-class Americans don’t. They did this by leveraging an impressive, comprehensive database that Gilens spent years building: It compiled public opinion polls of Americans (broken down by income level) on various policy issues, then compared their opinions with what actually happened.
Gilens and Page found that the model that best predicted policy outcomes based on public opinion was “Economic-Elite Domination” — where the rich have influence, and the rest do not.
It’s an attractive theory, and while its validity has been directly challenged by a number of other political scientists, it’s seeped into the public consciousness. It helped explain, and motivate, the appeal of outsider anti-elite campaigns like that of Bernie Sanders, and reinforced a sense that politicians of both parties have been bought off.
But since then, new research has painted a more nuanced, and in many ways more intriguing, portrait of how well or poorly Washington represents the views of the American people. This research suggests that not all politicians ignore the views of the poor and middle class.
Democratic elected officials are much likelier to share the policy opinions of the poor or middle class on economic policy, while Republican officials are likelier to diverge from middle-class public opinion in favor of representing the views of the wealthy.
In other words, Congress isn’t just an undifferentiated mass ignoring what the public thinks. Party matters.
Using big data to see if members of Congress agree with their rich or poor constituents
The Gilens dataset has been the basis for a large and growing body literature that’s impossible to cover in full in a single article. But four papers in particular have all suggested a strong difference between how Republican and Democratic politicians represent public opinion.
The first study, by UMass Amherst’s Jesse Rhodes and Brian Schaffner, measures how representation of public opinion varied across individual members of the 112th Congress, from 2011 to 2013. (The study was published in the Quarterly Journal of Political Science.) Rhodes and Schaffner offer three models for how a given politician could represent their constituents:
Rhodes and Schaffner use an unusually rich dataset: Catalist, a private campaign database that’s leased out to Democratic candidates and progressive groups. Catalist maintains a streamlined voter registration database that incorporates consumer data, neighborhood information, and other data about over 265 million American adults.
Catalist provides researchers with a subset of this data (covering about 1.5 million adults). It estimates each household’s income, and ideological positions, using both information on their neighborhood from the census and data on consumer habits from InfoUSA, a commercial data vendor that sells information on people to bulk mailers and other marketers.
Having collected data on voters’ ideology and income, Rhodes and Schaffner then compare their political views to those of their member of Congress, as measured through NOMINATE scores (a widely used tool in political science to measure members of Congress on a left-right scale). They then charted how well legislators matched their poor, middle class, and rich constituents. The left chart shows the results for all members of Congress, while the right chart breaks it down by party:
The overall trend is that poorer constituents were more likely to see their views reflected by their member of Congress than richer constituents (though the trend reverses for very rich people) — the opposite conclusion from the one that Gilens and Page reach.
But the picture varies a lot based on party. Democratic representatives provided populist representation: They matched the views of their poor constituents more than those of their rich constituents. The opposite was true for Republicans: Republicans matched their rich constituents better than their poor ones.
But Republicans were less likely to match with their constituents’ ideologies overall — whether those constituents were poor or wealthy. A wealthy constituent would get about the same level of ideology match from a Democrat or a Republican. It’s just that the match would be on the low end compared to other constituents for a Democrat, and on the high end for a Republican.
“When lower-income individuals have a Democratic House representative, they appear to receive much more representation than when they have a Republican representative,” Rhodes and Schaffner write. “But the party of the representative matters hardly at all for wealthier Americans.”
Rhodes and Schaffner replicate their analysis with the 2012 Cooperative Congressional Election Study, a massive survey of nearly 55,000 people, which asked about specific high-profile policy proposals, like repealing the Affordable Care Act and approving the Keystone XL pipeline on which members of Congress were likely to vote. Their findings were similar: Republicans were likelier to match the opinions of the wealthy (particularly the top 97th through 99th percentile), and Democrats likelier to match the views of the poor and middle class.
The election study data suggests that the wealthy are much better off with Republican representatives — in contrast to the Catalist data, which found that both Democrats and Republicans represent the wealthy equally well — and that Democrats represent the poor and middle class about equally.
Do leaders in Congress agree with the rich or the middle class?
Three other papers similarly found that Democrats and Republicans differ strongly in how they represent different economic classes. But they complicate the story by showing that each party’s behavior is influenced by interest groups, by what constituents of their own party (as opposed to all constituents) say, and whether economic or social issues are at stake.
Matt Grossmann and William Isaac at Michigan State have put together a paper reaching similar conclusions. The authors looked specifically at the views of Democratic and Republican presidents and leaders in Congress and in the presidency, based on a data set covering 1,863 policy proposals the federal government considered from 1981 to 2002 and polling data on how the rich and middle class viewed these ideas.
That lets them see how much each party’s leaders agree with the middle class versus the rich — as well as whose views are represented by interest groups like the ACLU, Sierra Club, American Conservative Union, or Americans for Tax Reform.
These interest groups, they find, tend to lean left overall even as some individual ones are conservative. They tend to support economic and social policies backed by the middle class, and oppose foreign policies backed by the wealthy. And Grossmann and Isaac found those groups — representing certain subsets of the population — were more influential than individual voters.
“We find that the two political parties primarily represent different interest group sectors, rather than public classes, and that neither party consistently favors the views of the affluent over those of the middle class,” Grossmann and Isaac conclude.
Democratic leaders, they find, are significantly likelier to support an economic policy when the middle-class and advocacy groups support it, and significantly less likely to support it when business lobbyists support it; the wealthy’s opinions have no significant effect, though obviously they overlap considerably with those of business lobbyists, which do seem to matter. On foreign and social issues, Democratic leaders are likelier to support policies that advocacy groups back, but the opinions of the middle class, wealthy, and business lobbyists aren’t predictive.
For Republican leaders, the situation varies by issue area. On foreign policy, preferences of business lobbyists are the only significant variable; for economics, the views of business lobbyists, views of the affluent, and views of advocacy groups all matter, with GOP leaders likely to side with businesses and the affluent and against advocacy groups. On social issues, business and advocacy group preference matters, with the GOP once again leaning toward business and away from advocates.
A third paper, by Columbia’s Jeffrey Lax, Justin Phillips, and Adam Zelizer, analyzes 39 roll-call votes in the US Senate from 2001 to 2012, starting with Bush tax cuts and the Patriot Act and ending with votes on Speaker Paul Ryan’s budget and potential extensions of the Bush tax cuts in 2012.
Their dataset includes a wide variety of famous and important votes, including passage of the Affordable Care Act, the Dodd-Frank financial reforms, the Iraq War (and a later bill calling for immediate withdrawal from the country), the bank bailouts, a constitutional amendment against same-sex marriage, a path to citizenship for undocumented immigrants, and more.
The Cooperative Congressional Election Study (the same dataset Rhodes and Schaffner used) allowed Lax, Phillips, and Zelizer to estimate where each state’s voters stood on each of these issues, and compare their opinions to those of their senators. It further allowed them to break down public opinion by income level to see if, say, Montana Sen. Max Baucus (D) was voted in line with Montana’s poor or its wealthy.
Strikingly, they find that Democratic senators reflected public opinion of both the poor and the rich more than Republican senators did. The two classes agree on an awful lot, which makes that convergence possible. Once the researchers limited the sample to issues where the rich and poor disagreed (on capital gains tax cuts, for example), they found that Republicans sided with the rich 88 percent of the time, and Democrats did so only 41 percent of the time.
But that was largely driven by the views of Republicans’ constituents from their own party. Each party’s senators were more likely to represent the views of their co-partisans than those of citizens belonging to the opposing party, and this could trump Republicans’ overall alignment with the wealthy on occasion. “Yes, the Republicans ‘choose’ rich over poor when they have to make a choice, but when their own partisans align with the poor, they go with the poor over the rich,” Lax told me in an email. “So, co-partisans + poor beats the rich.”
Finally, George Washington University’s Elizabeth Rigby and Cory Maks-Solomon also examined Cooperative Congressional Election Survey data to compare US Senate votes to public opinion by income group, from 2006 to 2014. They focus on 21 particularly controversial and heavily covered votes. Unlike most of the other researchers, they focus on large gaps in public opinion between the rich and poor, with the rich being more conservative on economics and foreign policy, and notably more liberal on social issues like stem cell research, abortion, and gay marriage.
Each party, they find, represents the rich more than the poor on certain issues. Republican senators were likelier to vote with the rich than the poor on economic issues, and likelier to vote with the poor than the rich on social ones. Democrats were likely to reflect the views of poorer constituents on economic policy, but not on social issues. “We find strong evidence that Republicans over-represent the rich on economic issues—pushing them further right than they would otherwise be, as well as that Democrats over-represent the rich on moral issues—pushing them further left than they would otherwise be,” Rigby and Maks-Solomon conclude.
Intriguingly, this is because those are the only topics on which each of those parties’ bases split. Democrats broadly agree on economic questions — but disagree on social issues, with richer Democrats being more liberal. Republicans, by contrast, are unified on social issues but poorer GOPers have much more liberal economic views than rich GOPers.
“I think the most important point is that for Republican senators, they are facing a Republican electorate in which the rich are generally more conservative on economic policy issues, but more in agreement with other Republicans on social/moral issues,” Rigby says in an email.
“For Democratic senators, they are facing a Democratic constituency that generally agrees on economic policy issues (so there is not really room to better represent rich Dems on those issues) but disagree markedly on social/moral issues with Rich Dems much more liberal. So if senators are first representing co-partisans [as in the Lax/Phillips/Zelizer paper], it is really only on these that each party could even over-represent the rich. And that’s what we find.”
What does it all mean?
The four papers all show something crucial when thinking about how Congress and policymakers represent the public: party matters. Rather than Democrats and Republicans both sharing views with the wealthy at the expense of the middle class, you see a more complex pattern. Especially on economic issues, it appears that Democrats respond more to the poor and middle class, and Republicans more to the affluent.
Still, their findings are hardly identical. Saying that Republicans represent the rich, while Democrats represent the poor and/or middle class, is a simplification. Rigby and Maks-Solomon find that Democrats overrepresent the rich on social issues. Grossmann and Isaac find that opinions of advocacy and business groups are often more predictive of the positions congressional leaders and presidents take than public opinion is, even public opinion of the rich alone.
But there’s a big caveat here: It’s not actually clear that Republican and Democratic senators are purposefully taking their views from the public, even when politicians’ views match their constituents’. It’s also possible — even likely — that it’s happening the other way: Politicians are shaping the public’s views. That helps explain why Democratic constituents align their views so closely with Democratic senators, and similarly for Republican constituents.
And because of the ideological sorting that has taken place in Congress over the past 50 years — with the dying out of both conservative Southern Democrats and liberal-leaning Northern Republicans — voters’ responsiveness to elite opinion has arguably exacerbated polarization at the voter level.
Gilens acknowledges that opinions often flow from politicians to voters but insists there’s significant causal influence the other way, from public opinion (particularly of the rich, in his view) to the opinions of policymakers, as well: “Education is more strongly linked to political knowledge and attentiveness than income,” he writes in an email. “If the causal path flows from elites to citizens, we would expect education to be the strongest factor shaping the association between citizens preferences and government policy. I show that this is not the case.”
Other political scientists aren’t so sure. “Longstanding political science suggests that the path of information from governing elites to the public is stronger than the reverse,” Grossmann and Isaac note. “More troubling, affluent Americans may hear official opinions first, meaning we would observe a greater association between their opinions and policy even if the true channel of influence were from government to the affluent.”
This interpretation, if true, is immensely troubling. The problem, in the world, is not that elected officials just listen to the rich. It’s that they, for the most part, don’t “listen” to anyone — and instead the rich are the only ones who listen to politicians. In this scenario, we don’t even have a kind of elitist, limited suffrage democracy where only the rich are listened to. We have a government by interest groups in which voters-qua-voters aren’t really listened to at all.