How Change Happens

by Cass R Sunstein

  Many white people quickly associate positive words like joy, or an evidently European American (Caucasian) face, with the upper-left corner when it says white and good—but have a much harder time associating joy with the left corner when the words there are black and good.5 So too, many white people quickly associate terrible with the left corner when it says black and bad, but proceed a lot more slowly when the left corner says white and bad. And when the picture in the middle is evidently of a European American (Caucasian), white people are a lot faster in associating it with the word good than when the picture is evidently of an African American.

  It is tempting to think that racial prejudice is deeply engrained and that nothing comparable can be found in the political domain, at least with respect to the two major parties in the United States. (To be sure, we might expect to see strongly negative implicit attitudes for Nazis or Communists.) To test for political prejudice, Shanto Iyengar and Sean Westwood, political scientists at Stanford University, conducted a large-scale implicit association test with two thousand adults.6 They found people’s political bias to be much larger than their racial bias. When Democrats see joy, it is much easier for them to click on a corner that says Democratic and good than on one that says Republican and good. Implicit bias across racial lines remains significant, but it is significantly greater across political lines.

  Love and Marriage

  If you are a Democrat, would you marry a Republican? Would you be upset if your sister did? Researchers have long asked such questions about race and have found that, along important dimensions, racial prejudice is decreasing.7 At the same time, party prejudice in the United States has jumped, infecting not only politics but also decisions about marriage. In 1960, just 5 percent of Republicans and 4 percent of Democrats said that they would feel “displeased” if their son or daughter married outside their political party.8 By 2010, those numbers had reached 49 percent and 33 percent.9 Interestingly, comparable increases cannot be found in the relevant period in the United Kingdom.10

  In 2009, by contrast, 6 percent of Americans reported that they “would be fine” if a member of their family married someone of any other race or ethnicity, a sharp change from as recently as 1986, when 65 percent of respondents said that interracial marriage was not fine for anyone or not fine for them.11 Asked specifically about marriages between African American and white partners, only 6 percent of white respondents and 3 percent of African Americans recently said that “they could not accept a black-white interracial marriage in their family.”12 Similarly, a recent Gallup survey found that 87 percent of people approve of interracial marriage, compared to 4 percent in 1958: a dramatic shift in social norms, showing the opposite trend line from that observed for partyism.13

  Hiring

  The IAT measures attitudes, not behavior. Growing disapproval of marriage across political lines suggests an increase in prejudice and hostility, but it might not map to actual conduct. To investigate behavior, Iyengar and Westwood asked more than one thousand people to look at the resumes of several high school seniors and say which ones should be awarded a scholarship.14 Some of these resumes contained explicitly racial cues (“president of the African American Student Association”), while others had explicitly political ones (“president of the Young Republicans”).

  In terms of ultimate judgments, race certainly mattered: African American participants preferred the African American scholarship candidates 73 percent to 27 percent. For their part, whites showed a modest preference for African American candidates as well, though by a significantly smaller margin. But party affiliation made a much larger difference. Both Democrats and Republicans selected their in-party candidate about 80 percent of the time.15 Even when a candidate from the opposing party had better credentials, most people chose the candidate from their own party.16 With respect to race, in contrast, merit prevailed.17 It is worth underlining this finding: racial preferences were eliminated when one candidate was clearly better than the other; by contrast, party preferences led people to choose a clearly inferior candidate.

  A similar study asked students to play the role of college admissions director and to decide which applicants to invite for an on-campus interview, based on both objective criteria (SAT scores, class rank) and subjective evidence (teacher recommendations).18 Among partisans with strong party identification, there was significant evidence of partyism: 44 percent of the participants reviewing someone from the opposite party selected the stronger applicant, whereas 79 percent of the participants in the control (in which participants had no knowledge of the applicant’s party affiliation) selected the stronger applicant.19

  Trust

  In a further test of the relationship between partyism and actual behavior, Iyengar and Westwood asked eight hundred people to play the trust game,20 well-known among behavioral scientists. As the game is played, player 1 is given some money (say, ten dollars) and told that she can give some, all, or none of it to player 2. Player 1 is then told that the researcher will triple the amount that she allocates to player 2—and that player 2 can give some of that back to player 1. When player 1 decides how much money to give player 2, a central question is how well she trusts him to return an equivalent or greater amount. Higher levels of trust will result in higher initial allocations.

  Are people less willing to trust people of a different race or party affiliation? Iyengar and Westwood found that race did not matter—but party did. People are significantly more trusting of others who share their party affiliation.

  Partyism can motivate partisans to be especially inclined to share negative information about the opposing party—or even to avoid its members altogether when forming a group.21 In one experiment, participants were asked to decide whether a strongly worded opinion piece, including hyperbole and name calling, that blamed congressional gridlock on one of the two political parties should be posted on a news organization’s website.22 The researchers found significant evidence of partyism: 65 percentage of people were willing to post the article if it was critical of the opposing party, but only 25 percent were willing to share it if it criticized its own party.23 They also found that the intensity of a participant’s partisan feelings correlated with their willingness to share a critical article.24

  In a second experiment, the researchers asked participants to pick a team of three people out of a list of four to join them in completing a puzzle game.25 Participants were informed of the partisan identities and education levels of the potential teammates; the least educated team member was always an independent. More than half the participants selected the least educated player for their team rather than choosing a better-educated member of the opposing party!

  An Objection

  From these studies and various others,26 it seems clear that partyism is widespread and on the rise in the United States. We can imagine reasonable disputes about the precise magnitude of the phenomenon, but not about its existence and significance. But there is an obvious objection to the effort to compare racism to partyism, and indeed to the very effort to describe partyism as seriously troubling. The objection is that people have legitimate reasons for objecting to people because of their political beliefs. If we think that Fascism or Communism is hateful, we will not object to those who are unenthusiastic about Fascists or Communists.

  For some people, a degree of suspicion and hostility across political lines is a product of legitimate disagreement, not of anything untoward. Racism and sexism result from devaluation of human beings on the basis of an immutable or at least irrelevant characteristic. Perhaps the same cannot be said for party affiliation. In fact the very idea of political prejudice, or any kind of corresponding ism, might seem badly misdirected. Perhaps we are speaking here not of any kind of prejudice but of a considered judgment about people who hold certain convictions. On certain assumptions, that is the precise opposite of prejudice.

  To come to terms with this response, we need to begin by distinguishing between daily life and politics as s
uch. It is hardly unreasonable to have a strong negative affect toward Fascists or Communists because of their political views. But if people dislike each other because of an affiliation with one of the major parties in the United States, something does seem badly amiss. To be sure, some characteristics or even commitments of one or another party might seem troublesome or worse. But both parties are large and diverse, and it is odd to think that outside of the political domain, members of one party should actually dislike members of another party as human beings.

  Of course this judgment turns on substantive conclusions. If you believe that Republicans are essentially racists and sexists, antipathy toward Republicans is understandable, and so too if you believe that Democrats are unpatriotic socialists who seek to undermine the United States. But if you believe that across the two parties, good-faith disagreements are possible and pervasive, then partyism will be hard to defend, not least if it seeps into daily life.

  In the political domain, of course, intensely held differences are common, and some kind of “we-they” attitude may be difficult or impossible to avoid. For members of Congress, such an attitude is, in a sense, built into the very structure of the two-party system. A degree of antipathy—at least if it is not personal—may reflect principled disagreement, not prejudice at all. It may be hard to avoid a measure of antipathy toward people with whom you intensely disagree, most of the time, in your day job. The problem is that good-faith disagreement is far from uncommon in politics, and in the face of such disagreement the task is to seek to identify ways to move forward (or not), rather than to discredit arguments because of their source. With respect to politics itself, something like partyism may be a product of principle, but it also has destructive consequences, as we shall shortly see.

  Causes

  What causes partyism? We do not yet know the answer, but some helpful clues have started to emerge.

  From Ideological Disagreement to Partyism?

  It is tempting to think that the growth in partyism is a product of the increasing intensity and visibility of ideological disagreements. Let us assume that at some point in the past—say, 1970—one or another of the two parties, or perhaps both, had a “wider tent.” Let us assume, in fact, that the conservative wing of the Democratic party was more conservative than the liberal wing of the Republican party and thus the two parties had significant ideological overlap. If so, we would not expect to see much in the way of partyism.

  This hypothesis could be tested in multiple ways. We could attempt to track ideological differences between the parties and test whether growth in ideological distance turned out to be correlated with increases in partyism. A strong correlation would not be definitive, but it would be at least suggestive. It would indicate that strong negative affect, across political lines, would have something to do with increasingly intense substantive disagreements. And if this turned out to be so, the rise of partyism would, in a sense, turn out to be rational, at least in the sense that prejudice and antipathy would be a product of something concrete and real. The role of partyism in the private domain would remain hard to defend, but in politics, at least, its recent increase would be comprehensible.

  But a better way to test the hypothesis would be to see whether the intensity of people’s policy preferences predicts partyism. In other words, when people have very strong views about political issues, and when those very strong views suggest clear divisions across party lines, are they more likely to show a negative affect toward the opposing party? Surprisingly, the connection between ideological polarization and negative affect is relatively weak.27 As polarization between the parties grows, negative affect does not grow with it. It appears that people’s partisan attachments are a product of their identity rather than their ideology. When Republicans dislike Democrats, or vice versa, it is largely because they are on the opposing side; substantive disagreements matter, to be sure, but they are not primary.

  Campaigns

  Do political campaigns create partyism? It is natural to suspect that they do, first because they make party differences salient, and second because part of the point is to cast the opposing side in a negative light. Iyengar and Westwood find strong support for this hypothesis. In particular, exposure to negative advertising contributes to a growth in partisan animus, and political campaigns themselves have that effect.28 Apparently campaigns serve to “prime” partisan identity and to support stereotypical and negative perceptions of both supporters and opponents.

  Your Media, My Media

  In a fragmented media market, it is easy for people to segregate along partisan lines. Recall the phenomenon of group polarization (see chapter 2), which should increase with the presence of echo chambers and information cocoons. In the United States, Fox News has an identifiable conservative orientation; MSNBC has an identifiable liberal orientation. Some talk shows are easy to characterize in terms of the political commitments of the host. If a show or a station characterizes one group of people as “the other side,” and if those on that side are described as malicious, foolish, or power-hungry, then viewers or listeners should experience a rise in partyism.

  Social media can also be used for purposes of partisan self-sorting; many people end up in the equivalent of echo chambers as a result of Facebook and Twitter. That might well contribute to partyism. We do not have clear data on this speculation, but some is emerging.29 It is reasonable to suspect a fragmented media market with clear political identifications contributes a great deal to partyism.

  Political Polarization

  Suppose that a society is divided on some proposition. The first group believes A, and the second group believes not-A. Suppose that the first group is correct. Suppose finally that truthful information is provided, not from members of the first group but from some independent source, in support of A. It would be reasonable to suppose that the second group would come to believe A. But in important settings, the opposite happens. The second group continues to believe not-A—and even more firmly than before. The result of the correction is to increase polarization.

  The underlying studies do not involve party differences as such, but they explore something very close to that, and they suggest the following proposition: An important consequence of partyism is to ensure that people with a strong political identification will be relatively immune from corrections, even on matters of fact, from people who do not share that identification. Because agreement on matters of fact is often a precondition for political progress, this phenomenon can be extremely destructive.

  In a relevant experiment, people were exposed to a mock news article in which President George W. Bush defended the Iraq war, in part by suggesting (as President Bush in fact did) that there “was a risk, a real risk, that Saddam Hussein would pass weapons or materials or information to terrorist networks.”30 After reading this article, they read about the Duelfer Report, which documented the lack of weapons of mass destruction in Iraq. Subjects were then asked to state their agreement, on a five-point scale (from “strongly agree” to “strongly disagree”), with the statement that Iraq “had an active weapons of mass destruction program, the ability to produce these weapons, and large stockpiles of WMD.”31

  The effect of the correction greatly varied by political ideology. For very liberal subjects, there was a modest shift in favor of disagreement with this statement; the shift was not significant, because very liberal subjects already tended to disagree with it.32 But for those who characterized themselves as conservative, there was a statistically significant shift in the direction of agreeing with the statement. “In other words, the correction backfired—conservatives who received a correction telling them that Iraq did not have WMD were more likely to believe that Iraq had WMD than those in the control condition.”33 It follows that the correction had a polarizing effect; it divided people more sharply on the issue at hand than they had been divided before.

  Another study confirmed the more general effect. People were asked to evaluate the proposition that
cutting taxes is so effective in stimulating economic growth that it actually increases government revenue. They were then asked to read a correction. The correction actually increased people’s commitments to the proposition in question. “Conservatives presented with evidence that tax cuts do not increase government revenues ended up believing this claim more fervently than those who did not receive a correction.”34

  Or consider a test of whether apparently credible media corrections alter the belief, supported and pressed by former Alaska Governor Sarah Palin, that the Affordable Care Act would create “death panels.”35 Among those who viewed Palin favorably but had limited political knowledge, the correction succeeded; it also succeeded among those who viewed Palin unfavorably.36 But the correction actually backfired among Palin supporters with a high degree of political knowledge. After receiving the correction, they became more likely to believe that the Affordable Care Act contained death panels.37

  Liberals (and Democrats) are hardly immune to this effect. In 2005, many liberals wrongly believed that President George W. Bush had imposed a ban on stem cell research.38 Presented with a correction from the New York Times or FoxNews.com, liberals generally continued to believe what they did before.39 By contrast, conservatives accepted the correction.40 Hence the correction produced an increase in polarization.