America is less polarized than it seems. Politicians need better data.

Polarization, a scholar has quipped, is a problem that no one can explain or solve. 

We know that a vast ideological gulf now separates most Democratic from most Republican politicians. We also know that highly polarized politicians poorly represent many of their constituents, who remain comparatively moderate. But we’re not sure why polarization has arisen or what we can do about it.

As a pathology of American politics that has developed over decades, polarization probably has no single cause or cure. One major driver of polarization, however, is just starting to come into focus: Many politicians may be representing their constituents poorly because they don’t know what their constituents want.

Unlike other potential drivers, this one also points to an obvious response: We should educate politicians about their constituents’ preferences so they no longer have to guess what voters think. 

Historically, information about public opinion was largely unavailable for the districts and municipalities that elect most politicians. Recently, though, new tools have made this data accessible to many more levels of government. These tools have the potential to substantially improve representation at the federal, state and local levels.

Politicians’ knowledge of their constituents’ views is worse than you’d expect for professionals whose careers depend on public approval. 

A pioneering 2018 study found that Republican state legislators err by 10 to 40 percentage points in their perceptions of voters’ positions on issues, including abortion, gun control and immigration. Moreover, these politicians almost always err in a conservative direction, mistakenly believing voters are far more right-wing than they actually are. 

Nor are legislative staffers any more discerning than their bosses. According to a 2019 study, both Democratic and Republican staffers wildly misstate congressional district opinion. Concerning repealing Obamacare, for example, true voter support hovers near 50 percent in most districts. But Democratic staffers usually underestimate this figure by 20 to 30 points while Republican staffers tend to overestimate it by a similar amount.

If politicians’ skewed perceptions of their constituents’ preferences contribute to polarization, could correcting these biases lead to better representation? An innovative 2008 experiment suggests that it could. 

A pair of researchers conducted a large-scale survey of New Mexicans’ views on a pending fiscal proposal. This poll was big enough to generate estimates of public opinion in each State House district. The scholars then circulated the results to some (randomly chosen) State House members but not others. 

Among the legislators who were informed about their voters’ preferences, their votes on the fiscal proposal were highly responsive to public opinion. Conversely, among the legislators who denied this data, there was no correlation between their votes and their constituents’ attitudes toward the bill.

Unfortunately, this experiment is the exception, not the rule. In general, politicians lack policy-specific information about their voters’ preferences. Presidents often have access to this data at the national level. But at every subnational level, this data isn’t widely available. The representational gains that would follow if politicians were more knowledgeable about their constituents’ views therefore don’t materialize because most politicians aren’t well-informed.

In recent years, though, this situation has begun to change. 

The American Values Atlas now reports state-level opinion, derived from a major national survey, on issues including immigration, gay rights and abortion. The American Ideology Project relies on several national surveys to estimate the overall ideologies of individuals at several geographic levels. 

And a new site called TrueViews, which I helped to produce, uses many national surveys to estimate people’s preferences on many issues within many geographic units. In sum, TrueViews covers dozens of policies and geographic units ranging from municipalities to districts to states and periods from 2009 to the present.

For the first time, the data make it possible for politicians of all stripes to learn instantly what their voters want. As these capabilities become better known, they could plausibly dampen polarization and bolster representation.

For politicians who would like to abide by their constituents’ preferences but frequently don’t know what they are, the new public opinion tools supply this exact information. They enable politicians who want to be faithful delegates to be faithful delegates.

What about politicians who would rather advocate other positions, like those of their parties or donors? For them, the new tools raise the cost of defying the will of the people. 

Candidates running against these politicians could highlight their divergence from voters’ preferences. Journalists could run stories about this mismatch. Activists could make it the target of protests. In some cases, these efforts might persuade these politicians to heed public opinion. In others, these politicians might be ousted from office and replaced by rivals more mindful of voters’ views.

The new tools have the ancillary benefit of puncturing some common misconceptions about American politics. We often hear that the country is divided into red and blue (and maybe purple) states. But on many issues, public opinion doesn’t correspond to this familiar partisan cleavage. There are policies that people heavily support in most states that are unpopular nearly everywhere, and there are policies on which the states line up in unusual ways. 

Likewise, the gradient from urban to suburban to rural areas is far from universal. Some cities have pockets of surprising conservatism. Many suburban belts are internally quite heterogeneous. And rural America is anything but monolithic, often resembling a checkerboard more than the uniform red landscape depicted by electoral maps.

A generation ago, an immigration scholar wrote “The American Kaleidoscope,” a book about American political culture. As the new tools illustrate, this title also applies to American public opinion. 

Variable but nonrandom, related but not identical to partisanship, the product of myriad social, historical, and economic forces, the spatial patterns of voters’ views are absorbing, even entrancing. Take a look and see for yourself.

Nicholas Stephanopoulos is the Kirkland and Ellis Professor of Law at Harvard Law School and a member of the academic team behind TrueViews.