By Rose Jacobs | Illustrations by Nate Kitch
March 2, 2020
By Rose Jacobs | Illustrations by Nate Kitch
March 2, 2020
Reed Schroer, a 70-year-old Lutheran pastor from Rhodes, Michigan—about an hour north of Flint—never saw eye to eye with his brother-in-law. Over 50 years, they argued about religion, taxes, and organized labor. But they also had what both would describe as a good relationship, built on an interest in scripture, a devotion to the rural communities of northern Michigan, and, ultimately, their love for Schroer’s sister.
In 2016, that balance was upset by events in Washington, and Schroer found himself increasingly frustrated by the “conspiracy theories and right-wing talking points” his brother-in-law embraced, and less and less able to convince himself that the man’s rhetoric, which Schroer found odious, belied his core beliefs—or that the two could even find a way to talk about it. The men went more than eight months without speaking. “He always drove me nuts, but in the past I might get so frustrated I would storm out of the house to cool off, and he’d chase me down the street saying, ‘I love you, man,’” says Schroer. Now, this communion has been replaced by silence and alienation.
Schroer’s experience resonates across the United States today, in families and institutions. As the Democratic Party battles over whether a moderate or liberal presidential candidate stands the better chance of winning the White House in November 2020, many Americans are asking a similar but broader question: Has the country ever been so divided?
Academics, for their part, are attempting to measure what often feel like widening gaps. In 2017, Stanford’s Matthew Gentzkow looked at a series of Pew Research Center surveys of Americans’ views on policies ranging from government regulation to welfare, immigration, and the environment, and noted that fewer individuals in 2014 than 10 years earlier held positions that put them across the political divide from their own, self-identified political party. “Most Americans hold relatively moderate views on, say, immigration,” he writes. “But the frequency of Republicans holding pro-immigrant views, or Democrats holding anti-immigrant views, has decreased substantially.” Where Americans’ political views and social attitudes, charted, might once have looked like a bell curve, with the majority gathered at a moderate center, the line increasingly shows two separate humps where Democrats and Republicans congregate.
Gentzkow, along with Brown’s Jesse Shapiro and Amazon’s Matt Taddy, has also looked at the polarization of politicians, as measured by the language they use. Republicans, they find, are more likely to talk about death taxes, where Democrats use the term estate taxes. This might not be surprising to anyone who witnessed the “antiabortion” versus “pro-life” semantic split following Roe v. Wade, and yet the researchers find that this polarization is a relatively modern phenomenon. “Partisanship of language has exploded in recent decades, reaching an unprecedented level,” they write. “From 1873 to the early 1990s, partisanship was nearly constant and fairly small in magnitude. . . . Beginning with the congressional election of 1994, partisanship turned sharply upward.” They can now guess a politician’s party on the basis of a one-minute speech with 73 percent accuracy, compared with about a 55 percent chance over the 120 years from the late 19th century to the late 20th.
Nor do divides appear confined to politics and policy. Chicago Booth’s Marianne Bertrand and Emir Kamenica examined three national surveys that probe Americans’ consumption habits, leisure time, and social attitudes. They find that different groups of Americans—rich and poor, black and white, men and women, politically liberal and conservative, college educated and not—tend to eat different food, watch different television programs, pursue different hobbies, and adopt different social attitudes. The algorithms the researchers developed for their study were able to predict people’s income bracket with nearly 90 percent accuracy on the basis of the brands of products and services they bought; they could do the same for gender by looking at what TV shows and films people watched and what magazines they read; and they could predict race with 75–85 percent accuracy using self-reported stances on topics such as marriage, law enforcement, and government spending.
Yes, then, the nation appears to be divided. But whereas Gentzkow, Shapiro, and Taddy saw some gaps growing over time, Bertrand and Kamenica’s research undermines the idea that cultural division is a contemporary phenomenon. Their data sets stretch back to at least 1990—with some, namely the time-use and social-attitude surveys, beginning in the mid-’70s—and reveal that, on almost all measures, these divides have not widened over the decades.
It is as easy to use purchases to predict a person’s income the year Bill Clinton was first elected president as when Donald Trump took office. In 1992, richer people bought Grey Poupon mustard; in 2016, they bought Apple products. Meanwhile, it is just as easy to use TV-show preferences to predict education levels in 1994 and 2016. College graduates in 1994 watched Unsolved Mysteries; in 2016, they watched the HGTV program Love It or List It. And both in 1976 and now, views on law enforcement can predict a person’s gender. (The prediction algorithms the researchers employed have the most trouble forecasting education levels and race on the basis of how people spend their free time: knowing how much time a person who is employed full time spends on leisure activities only helped the algorithms predict race and education with 50–60 percent accuracy.)
Still, there is an exception to the rule that cultural divides have held roughly steady over the past half-century. When it comes to political ideology, while the gap between what conservatives and liberals buy, read, and watch has remained roughly constant since the ’70s, it is easier to predict a person’s political persuasion today on the basis of his or her social attitudes than it was 20 years ago, and it’s much easier than it was 40 years ago. Similar to Gentzkow, Bertrand and Kamenica find that liberals and conservatives are decreasingly likely to sit at the center of a national bell curve; instead, they have diverged on issues related to premarital sex, homosexuality, abortion, environmental spending, defense spending, confidence in the executive branch of government, and more.
Bertrand and Kamenica point out that cultural gaps in the categories that they studied, between rich and poor or black and white, for instance, are worrisome in part because they might dampen social and economic mobility. The real-world effects of growing partisanship are less obvious, but research is beginning to probe how a politically divided populace plays out in areas ranging from corporate finance to macroeconomics to medicine and law.
For example, growing partisanship may affect spending. Individual investors appear to react to elections, according to a 2017 study by University of Colorado’s Yosef Bonaparte, Alok Kumar at the University of Miami, and Jeremy K. Page, with individuals moving toward riskier assets if their political party is reflected in the White House. Past research has produced mixed results about whether Republican households tend to increase spending under Republican presidents and Democratic households under Democratic presidents. Whereas Arizona State University’s Christos Makridis, for example, finds that conservatives increased spending on nondurable goods following Donald Trump’s presidential victory, confirming a trend observed by Yale’s Alan Gerber and Gregory Huber in 2009, Princeton’s Atif Mian, Chicago Booth’s Amir Sufi, and Nasim Khoshkhou at Argus Information and Advisory Services argue that while politics influence individuals’ economic expectations, the correlation breaks down when it comes to household spending.
Chicago Booth’s Elisabeth Kempf and Cornell’s Margarita Tsoutsoura recently looked at how partisanship might affect companies on a larger scale, examining the potential effect on credit ratings and, subsequently, company investment. They used voter-registration records and press releases about debt-rating changes from the three main ratings agencies—Standard and Poor’s, Moody’s, and Fitch—to determine whether an analyst’s political affiliation affects her assessment of a company’s financial outlook.
Aside from underscoring that the US is not alone in experiencing rising polarization, the evidence that populations are often divided from the top down supports the idea that grassroots empathy can be an important countervailing force.
The researchers looked at the months surrounding President Trump’s election in 2016, and find that analysts registered as Democrats were more likely to issue downgrades to the companies they covered after November 8 than were Republican analysts. This effect was greater with analysts who voted more frequently. This result is in line with their wider analysis of political affiliation and presidential elections going back 18 years, which suggests that analysts whose politics do not align with the sitting president’s are more likely to downgrade companies’ debt than analysts who share a political party with the president.
Credit analysts have company in their susceptibility to bias: political scientist Eitan Hersh and psychiatrist Matthew Goldenberg, both at Yale, find in a 2016 paper that a doctor’s treatment plan and level of concern for a patient dealing with a politicized health issue such as abortion or drug use can be affected by that doctor’s politics. Several studies detect similar patterns in judges. But Kempf and Tsoutsoura’s results suggest that even supposedly rational actors who have an incentive to make accurate ratings calls cannot escape their own biases. “These are highly sophisticated people in a competitive environment,” says Kempf. “Most people would not expect this sort of thing to affect them.”
As the evidence indicates, partisanship’s impact reaches beyond household-, clinic-, or courtroom-level decisions. Credit ratings have a direct impact on a company’s cost of borrowing, which in turn influences corporate investment decisions that can affect suppliers, employees, and the wider economy. “Understanding the potential implications of [increased political polarization] for the US economy is important,” write Kempf and Tsoutsoura, pointing out that political bias among corporate managers and asset managers would also have ripple effects in the economy.
There may also be multiple mechanisms by which political bias among individuals affects the economy at large, as the research reveals a surprising degree of political homogeneity within ratings agencies and within sectors. Moody’s and Fitch were much more likely to employ Republican analysts, and Standard and Poor’s to employ Democrats; Democratic analysts were more likely to cover utilities groups and financial companies, while Republican analysts focused more often on the energy sector. “When you see this kind of sorting within companies, you can imagine workplaces becoming echo chambers,” says Kempf. “That can be a good thing as far as workplace cohesion, but it can also reduce diversity in opinions and exacerbate biases.”
Political identity is one of many levers that can affect an analyst’s decisions. A person’s hometown, race, gender, or creed might too. Identities such as these are predictors of voting: Stephanie Chen of the London Business School and Chicago Booth’s Oleg Urminsky conducted a survey in 2017 that finds people who said their British or English identity was central to their sense of self were more likely to support the United Kingdom’s plans to leave the European Union than those who said those identities were not central. The researchers also find, through US surveys, that people who strongly identified with one political party or another were more likely to vote for that party’s candidate even if they personally were dissatisfied with the individual politician they were voting for.
The research provides a roadmap of sorts for unity. Political divides—wide though they are, and in some cases growing—might be bridged if people are able to recognize the degree to which they have come to their own politics via personal identity and personal experience—and that individuals on the other side are simply doing the same. University of Michigan’s A. Yesim Orhun and Booth’s Urminsky published a paper in 2013 suggesting we are ripe for this sort of recognition, since people seem to believe others make decisions in the same way that they themselves have, whether the decisions involve choosing presidential candidates or scenic postcards.
Ahead of the 2008 US presidential election, Orhun and Urminsky probed what students who supported candidate Barack Obama (then Democratic senator of Illinois) thought about people who supported candidate John McCain (then Republican senator of Arizona). They find that strong supporters of Obama imagined that McCain backers also felt very positively about McCain; similarly, Obama backers who despised McCain assumed McCain backers would despise Obama, and Obama backers with some positive views of McCain assumed McCain backers would have some positive views of Obama. A second study of registered voters demonstrated similar patterns, even when people changed their view of a candidate (spurred, in this case, by a televised political debate). Survey respondents tended to assume that other people, even supporters of the candidate they opposed, changed their opinions in a similar fashion. The pattern was replicated outside the political sphere, as Orhun and Urminsky asked study participants to rate the aesthetic value of two posters or two postcards, and guess how other people felt about the options. “People rely on their own evaluations to make sense of others’,” write the researchers.
Might these findings start to explain why, despite the decades of cultural division, the US has not come apart at the seams? In a 2019 book, Democracies Divided, Thomas Carothers at the Carnegie Endowment for International Peace and Andrew O’Donohue, a research fellow at the Istanbul Policy Center, asked experts in nine countries, including Brazil, Indonesia, Kenya, Poland, and Turkey, to create case studies of political polarization at home. The researchers, looking for patterns across borders, note that the phenomenon is consistently linked to divisive leaders and the social media they leverage. Aside from underscoring that the US is not alone in experiencing rising polarization, the evidence that populations are often divided from the top down supports the idea that grassroots empathy can be an important countervailing force.
If it is, however, Gentzkow’s 2016 paper, which reviews research findings about polarization, offers little hope for unity in the US. Some of the research he discusses suggests that, in addition to a widening gap in political views, dwindling empathy—the kind of empathy identified by Orhun and Urminsky—is contributing to a sense that America is more divided now than ever.
Gentzkow cites research—published in 2012, by Stanford’s Shanto Iyengar; Gaurav Sood, then a postdoctoral scholar at Princeton; and Yphtach Lelkes, now at the University of Pennsylvania—that looked at a 1960 study of how Americans perceive people who are in political parties different than their own. Iyengar, Sood, and Lelkes compared the results with YouGov polls from 2008 that asked the same questions. In 1960, only one in five people surveyed thought individuals from the opposing political party were “selfish”; in 2008, nearly half of survey respondents felt this way. In 1960, more than a quarter of people called their political opponents “intelligent”; in 2008, fewer than one in six respondents did.
“Americans may or may not be further apart on the issues than they used to be. But clearly what divides them politically is increasingly personal, and this in many ways may be worse,” writes Gentzkow.
For Schroer, this is an apt description of his situation. “I’m anything but a liberal,” he says.
I’m a Second Amendment kind of guy. I think too much government is not a good thing. I never worshipped Obama. What broke my heart with my brother-in-law is that I thought we had the same framework: Patience is a virtue. Showing courage in the face of oppression is a righteous thing. It’s better to give than to receive. The basic Christian value set. If he said those were still his lodestars and could make the case that his politicians were working for something like that for the country, there would be something to talk about. I’m still waiting for that conversation.
Instead, in Schroer’s view, his brother-in-law has thrown the rules of engagement out the window, with “notions of what is evidence, what constitutes proof, how you evaluate an idea” all useless. The sense that they’ve arrived at their present positions on recognizable, if different, paths is gone. “With him, everything was tactical or sentimental,” Schroer says. “I finally said, ‘I don’t want to do this anymore. I don’t want to be in the world you seem to live in and I certainly don’t want to get dragged into it.’ I refused to continue the argument.” Surmounting this personal gap, played out in the aggregate, may be the key to our future.
Marianne Bertrand and Emir Kamenica, “Coming Apart? Cultural Distances in the United States over Time,” Working paper, July 2018.
Yosef Bonaparte, Alok Kumar, and Jeremy K. Page, “Political Climate, Optimism, and Investment Decisions,” Journal of Financial Markets, June 2017.
Stephanie Chen and Oleg Urminsky, “The Role of Causal Beliefs in Political Identity and Voting,” Cognition, July 2019.
Matthew Gentzkow, “Polarization in 2016,” Toulouse Network for Information Technology white paper, 2016.
Matthew Gentzkow, Jesse Shapiro, and Matt Taddy, “Measuring Polarization in High-Dimensional Data: Method and Application to Congressional Speech,” Working paper, June 2015.
Alan Gerber and Gregory Huber, “Partisanship and Economic Behavior: Do Partisan Differences in Economic Forecasts Predict Real Economic Behavior?” American Political Science Review, August 2009.
Eitan Hersch and Matthew Goldenberg, “Democratic and Republican Physicians Provide Different Care on Politicized Health Issues,” Proceedings of the National Academy of Sciences, October 2016.
Shanto Iyengar, Gaurav Sood, and Yphtach Lelkes, “Affect, Not Ideology: A Social Identity Perspective on Polarization,” Public Opinion Quarterly, December 2012.
Elisabeth Kempf and Margarita Tsoutsoura, “Partisan Professionals: Evidence from Credit Rating Analysts,” NBER working paper, December 2019.
Christos Makridis, “Sentimental Business Cycles and the Protracted Great Recession,” Working paper, December 2017.
Atif Mian, Amir Sufi, and Nasim Khoshkhou, “Partisan Bias, Economic Expectations, and Household Spending,” Working paper, April 2018.
A. Yesim Orhun and Oleg Urminsky, “Conditional Projection: How Own Evaluations Impact Beliefs about Others Whose Choices Are Known,” Journal of Marketing Research, February 2013.