Why Is It So Hard To Admit When You're Wrong?

When it comes to political polarization, it's confirmation bias all the way down.


People who commit intentional murder—and only those people—should be executed. That's a view I held for virtually all of my adult life.

I am fully aware of the decadeslong debate over the death penalty. I have made it my business over the years to read the many conflicting studies on the practice's efficacy. But I didn't care if executing convicted murderers has a deterrent effect or not: I supported capital punishment because I want to do justice.

I am by nature a peaceable man; I have not hit anyone in anger since my teenage years. But my conception of what is just is informed by what I would want to do to a person who, beyond any shadow of a doubt, willfully killed my wife, another family member, or a close friend: inflict barbarous atonement for a barbaric act. One of the chief purposes of state-sanctioned execution has been to maintain social peace by forestalling blood feuds between people who would otherwise seek justice on their own.

I was not alone in advocating death sentences for murderers. Gallup reports that an average of 66 percent of Americans (and a majority of both parties) favored the death penalty for convicted murderers during the first decade of this century. By 2020, however, that number had dropped to 55 percent. Gallup has been documenting a widening gap on the issue between Republicans and Democrats over the past two decades, with a rock-solid 80 percent of Republicans still favoring the death penalty even as Democratic support has dropped to under 40 percent.

Despite that recent shift in the numbers, any rancor over the widening partisan divide with respect to the death penalty has been relatively mild compared to the growing estrangement over such issues as guns, affirmative action, climate change, and vaccinations. Research shows Americans increasingly align their opinions on hot-button issues along partisan lines and that they are likely to stick with those positions once committed.

Today, if you are a member of one of the two major American political parties, you are statistically likely to dislike and distrust members of the other party. While your affection for your own party has not grown in recent years, your distaste for the other party has intensified. You distrust news sources preferred by the other side.  Its supporters seem increasingly alien to you: different not just in partisan affiliation but in social, cultural, economic, and even racial characteristics. You may even consider them subhuman in some respects.

You're also likely to be wrong about the characteristics of members of the other party, about what they actually believe, and even about their views of you. But you are trapped in a partisan prison by the psychological effects of confirmation bias. Being confronted with factual information that contradicts your previously held views does not change them, and it may even reinforce them. Vilification of the other party perversely leads partisans to behave in precisely the norm-violating and game-rigging ways they fear their opponents will. It's a classic vicious cycle, and it's accelerating.

It also traps individuals within their preexisting worldviews. As a libertarian, conventional left/right partisan splits over many public policy issues are not particularly relevant to me. But even as my unease about the death penalty slowly mounted, I found in myself an incredibly powerful reluctance to publicly change my view and renounce prior commitments on the matter. Why is it so hard to admit when you're wrong, especially in the realm of politics?

Social scientists have a term for the phenomenon described above: affective polarization. In the U.S. context, that means Democrats' and Republicans' growing tendency to dislike and distrust each other.

Since 1978, the Northwestern University psychologist Eli Finkel and his colleagues have been trying to capture this phenomenon with a thermometer. By asking Americans to describe their feelings on a scale from cold (0 degrees) to warm (100 degrees), they've found that people feel quite warmly about their co-partisans, consistently reporting between 70 and 75 degrees. In contrast, feelings toward opposing partisans have plummeted from a mild 48 degrees in the 1970s to a frosty 20 degrees today: an emotional cold snap. "Since 2012—and for the first time on record—out-party hate has been stronger than in-party love," they write in the October 30, 2020, issue of Science.

The consequences of this big chill are apparent in several other studies, notably the work of the Louisiana State University political scientist Nathan Kalmoe and the University of Maryland political scientist Lilliana Mason. One of their more striking results is that 60 percent to 70 percent of both parties in a 2017–18 survey said they thought the other party was a "serious threat to the United States and its people"; 40 percent of respondents in both parties thought the other party was "downright evil." In another poll, 15 percent of Republicans and 20 percent of Democrats agreed with the brutal sentiment that the country would be better off if large numbers of opposing partisans in the public today "just died." And 18 percent of Democrats and 13 percent of Republicans said that violence would be justified if the opposing party won the 2020 presidential election.

Such studies suggest that there is something substantially different about the virulence of partisan sentiment in recent years and that the trend isn't going away.

Why do Americans increasingly think ill of their political opponents? To some extent, people may be taking their cues from political elites. Parsing the roll call votes of Democratic and Republican legislators reveals steeply increasing partisan polarization in Congress since the 1970s. In a 2018 Electoral Studies article on how party elite polarization affects voters, the Texas Tech political scientist Kevin K. Banda and the University of Massachusetts Lowell political scientist John Cluverius find that "partisans respond to increasing levels of elite polarization by expressing higher levels of affective polarization, i.e. more negative evaluations of the opposing party relative to their own."

The Emory University political scientists Steven Webster and Alan Abramowitz have been tracking the growing mutual dislike of Democratic and Republican partisans, and they note that the growing ideological distance between Republican and Democratic Party elites may be contributing to broader partisan polarization.

In addition, partisan affiliation used to be much less correlated to other social and political divisions. Mason and the Louisiana State University political scientist Nicholas Davis have analyzed survey data that YouGov/Polimetrix and the American National Election Studies collected from 1948 to 2012. In a 2015 working paper, the two scholars report that "the stronger and more strongly aligned our religious, racial, and partisan -identities, the more neatly our parties correspond to our ideological identities. This increased ideological consistency corresponds to an increase in partisan bias and intolerance across the electorate."

Even with the growing ideological split, partisans dramatically overestimate the substantive differences between members of the two parties.

In a 2015 YouGov survey, respondents reckoned that 32 percent of Democrats are LGBT, 29 percent are atheists or agnostics, and 39 percent belong to unions; the right figures are really 6, 9, and 11 percent, respectively. Meanwhile, they estimated that 38 percent of Republicans earn over $250,000 per year, 39 percent are over age 65, and 42 percent are evangelicals; actually, just 2 percent earn that much, 21 percent are senior citizens, and 34 percent are evangelicals.

Democrats and Republicans also regularly overestimate just how much their opponents loathe them. On a sliding scale from 0 (least evolved) to 100 (most evolved) Republicans rated the humanity of their fellow partisans at around 85 points and that of Democrats at 62 points, a 23-point difference. Conversely, Democrats gave 83 points to their political confreres and only 62 points to Republicans, a 21-point difference. Even more interesting is that the Democrats guessed that the Republicans would award them just 36 points (26 points less than the true number), while Republicans estimated that Democrats would give them a measly 28 points (34 points less than the true number).

"Democrats and Republicans equally dislike and dehumanize each other," concluded the University of Pennsylvania political scientist Samantha Moore-Berg and her colleagues in a 2020 study published in the Proceedings of the National Academy of Sciences, "but think that the levels of prejudice and dehumanization held by the outgroup party are approximately twice as strong as actually reported by a representative sample of Democrats and Republicans."

One of the more dire consequences of this exaggerated meta-perception—the perception partisans have of the other side's perception of them—is that it seems to make people more willing to support illiberal and antidemocratic policies, such as curbs on free speech and political participation.

Moore-Berg's findings were essentially replicated in a 2021 study by the University of California Santa Barbara social scientist Alexander Landry and his colleagues, who further found that "despite the socially progressive and egalitarian outlook traditionally associated with liberalism, the most liberal Democrats actually expressed the greatest dehumanization of Republicans." Democrats also expressed greater antidemocratic outgroup spite than Republicans.

The Yale political scientists Matthew Graham and Milan Svolik conducted experiments asking partisans if they would still support their party's candidates if those standard-bearers advocated various policies that violate democratic norms. Among the policies: a redistricting plan that would give their own party two extra seats despite a decline at the polls, and a proposal to reduce the number of polling stations in areas where the opposition party is strong. The researchers found that only a small percentage of voters would withhold their support from politicians from their own parties who violated such norms. "Put bluntly," they conclude, "our estimates suggest that in the vast majority of U.S. House districts, a majority-party candidate could openly violate one of the democratic principles we examined and nonetheless get away with it."

According to Kalmoe and Mason, about 20 percent of both Democrats and Republicans agreed that if their own parties break a few rules to oppose the other party, it's because they need to do it for the sake of the country.

Given that many of these views about the opposing party are mistaken, would more accurate information help solve the problem? Unfortunately, a large body of research shows that political partisans tend to see what they want to see when confronted with data and that they tend to seek out information that confirms their previously held views, even when the cost to doing so is clear.

In a 2012 experiment, partisan viewers were shown the same protest video. "When participants thought that the video depicted liberally-minded protesters (i.e. opposing military recruitment on campus), Republicans were more in favor of a police intervention than Democrats, whereas the opposite emerged when participants thought the video showed a conservative protest (i.e. opposing an abortion clinic)," observed the New York University psychologist Jay J. Van Bavel and the Leiden University psychologist Andrea Pereira in a 2018 Trends in Cognitive Science article. "Faced with the same visual information, people seem to have seen different things and drawn different conclusions depending on their political affiliations."

But are partisans really seeing different things? Perhaps they are mostly cheerleading their team rather than asserting actual beliefs. This is the thesis explored by the University of Nottingham philosopher Michael Hannon in a 2020 paper for Political Epistemology. He points to a survey of nearly 1,400 Americans conducted in January 2017. Researchers showed half of the respondents photos, simply labeled A and B, of the crowds on the National Mall during Barack Obama's 2009 inauguration and Donald Trump's 2017 inauguration. They were asked which photo depicted the crowd for each president. Forty-one percent of Trump voters said the photo with the larger crowd depicted the Trump inauguration, which was actually the one from the Obama inauguration. Only 8 percent of Hillary Clinton voters picked the wrong photo. The researchers argue that it is likely that Trump voters picked the photo with the larger crowd as a way to express their partisan loyalties and show their support for him.

More tellingly, the researchers asked the other half of the respondents which photo depicted the larger crowd. One answer was clearly correct. But Trump voters were seven times more likely (15 percent) than Clinton voters (2 percent) to assert that the much less populous photo of Trump's inauguration had more people. Remarkably, 26 percent of Trump voters with college degrees answered incorrectly. "When a Republican says that Trump's inauguration photo has more people, they are not actually disagreeing with those who claim otherwise. They're just cheerleading," argues Hannon. "People are simply making claims about factual issues to signal their allegiance to a particular ideological community."

Partisan cheerleading sounds harmless—not much different from fans rooting for a local football team, right? Nope. Hannon argues that "if our disagreements are not based on genuine reasons or arguments, then we cannot engage with each other's views." If team loyalty is the main thing, then the upshot for Hannon is that "we cannot decrease polarization by reasoned debate."

A 2015 study in the Quarterly Journal of Political Science sought to distinguish partisan cheerleading from sincere partisan divergence. The Northwestern University political scientist John Bullock and his colleagues found that offering participants small payments for giving correct and "don't know" answers to politically salient questions reduced the partisan gap between Republicans and Democrats by about 80 percent.

"To the extent that factual beliefs are determined by partisanship, paying partisans to answer correctly should not affect their responses to factual questions. But it does," they observe. "We find that even modest payments substantially reduce the observed gaps between Democrats and Republicans, which suggests that Democrats and Republicans do not hold starkly different beliefs about many important facts." Based on these results, the researchers urge analysts of public opinion to "consider the possibility that the appearance of polarization in American politics is, to some extent, an artifact of survey measurement rather than evidence of real and deeply held differences in assessments of facts."

On the other hand is a series of experiments conducted by the Texas A&M political scientist Erik Peterson and the Stanford political scientist Shanto Iyengar. As they report in a 2021 article for the American Journal of Political Science, they asked Republican and Democratic partisans to evaluate the truth of claims about several hot-button issues, such as "illegal immigrants commit violent crime at a significantly higher rate than legal American citizens" and "40 percent of firearm sales in the U.S. occur without a background check." (In both cases, the correct answer was they don't.) The researchers found that 97 percent of the Democrats got the right answer about immigrant crime vs. 45 percent of the Republicans. But on gun sales, only 22 percent of the Democrats got the right answer, compared to 56 percent of the Republicans.

Peterson and Iyengar also gave respondents access to various news sources so that they could check for additional information on whether their beliefs were accurate. These included sources identifiably associated with both liberal and conservative partisan loyalties, so-called mainstream sources, and expert sources from peer-reviewed journals. Some 29 percent turned to co-partisan sources, 26 percent to expert sources, 38 percent to mainstream sources, and only 7 percent to out-party sources.

The researchers offered another set of respondents a small monetary incentive for providing accurate answers, along with access to the various news sources. Even with the incentive, they found, "roughly 60 to 70 percent of the initial partisan differences remain." That is evidence for some cheerleading, but it also suggests that most partisans sincerely believe factually inaccurate claims.

If unincentivized partisans are mostly cheerleading, Peterson and Iyengar hypothesized, reliance on congenial partisan news sources should decline when they are paid to provide an accurate response. On the other hand, if they are confident that their responses are already correct, partisans' preferences for biased information would be unaffected by a financial incentive. Peterson and Iyengar report that "the incentives have no effect whatsoever on news choice."

Nearly 900 participants in Peterson and Iyengar's experimental surveys also agreed to have their everyday media diets tracked using an app installed on their computers. Out in the wild, these participants likewise tended to rely on news sources that confirmed their partisan views.

The proliferation of self-consciously partisan broadcast media, such as Fox and MSNBC, and of partisan gathering places on social media platforms provides political sectarians plenty of opportunity to find information that confirms their ideological predispositions and disparages their opponents' views. In 2019, a Perspectives on Psychological Science review of 51 studies testing for political bias found that "both liberals and conservatives were biased in favor of information that confirmed their political beliefs, and the two groups were biased to very similar degrees."

"Our studies indicate that partisans are genuinely committed to the inaccurate beliefs they report in surveys," Peterson and Iyengar conclude. It's confirmation bias all the way down.

Maybe the way out of this quagmire is to examine why partisans believe their views are correct in the first place and why they're so sure their opponents are wrong.

Hrishikesh Joshi, a philosopher at Bowling Green State University, did just that in a 2020 paper called "What Are the Chances You're Right About Everything?" He begins by listing nine highly politicized propositions: abortion is wrong; a carbon tax to address global warming is a good idea; illegal immigration is a serious problem; homosexual couples should be allowed to marry; the federal minimum wage should be raised; controls on gun ownership should be increased; racial affirmative action in college admissions is not justified; African Americans are unfairly targeted by police; and there are too many regulations on U.S. businesses.

Joshi contends that these propositions are orthogonal—that is, your position on one doesn't necessarily commit you to any particular position with respect to the others. Your stance on abortion rights should not, as a matter of pure logic, suggest anything about your views on climate change. And yet we all know that by quizzing people about their views on a couple of hot-button political issues, we can often figure out where they stand on the rest of Joshi's list of nine. If they support gun control, they are likely to favor affirmative action. If they don't support gun control, they are likely to think business is overregulated.

"Since the two sides disagree with respect to a host of political issues," Joshi writes, "one side's getting it consistently right entails that the other side is getting things consistently wrong." Somehow, each side's political opponents "succeed in consistently getting the wrong answer with respect to a large domain of rationally separable political questions!" If that were so, argues Joshi, those opponents would be not just unreliable but anti-reliable: They would reliably choose the wrong answer on each separate issue. He challenges partisans to "identify psychological differences between conservatives and liberals that can plausibly ground an explanation of why one side is anti-reliable with respect to the issues of partisan disagreement."

Joshi himself explores various ways that this could happen. One possibility is that they are systematically wrong because they share a core false belief.

As an example, Joshi posits a libertarian belief in a limited-government night-watchman state. For proponents of an extensive social welfare state, such a libertarian would be anti-reliable with respect to funding universal health care, generous unemployment insurance schemes, and subsidized housing for low-income individuals. But as Joshi points out, these issues are related to the libertarian's core belief and so are not orthogonal—that is, they are rationally related to one another.

Joshi evaluates other possible explanations for the anti-reliability of partisan opponents. Is one group of partisans on average more intelligent than the other? Not really. Joshi cites a 2019 study, "(Ideo)Logical Reasoning: Ideology Impairs Sound Reasoning," that found an equal tendency among liberals and conservatives to ignore the soundness of classically structured logical syllogisms in order to reach conclusions that supported the political beliefs that they already held. (He also notes that some 2018 research by Danish psychologists finds that greater cognitive ability predicts greater social liberalism but greater economic conservatism, a combination that may sound familiar to libertarians.)

Recent research finds neither liberals nor conservatives exhibiting greater distrust of scientific expertise, Joshi notes, yet greater science literacy correlates with more polarized beliefs on such topics as climate change and stem cell research. In other words, greater scientific knowledge may facilitate defenses of positions motivated by nonscientific concerns.

Another possibility: Perhaps your opponents are consistently wrong on political issues because they are in thrall to a perverted set of morals. But a 2018 study in Political Psychology, "Deep Alignment with Country or Political Party Shrinks the Gap Between Conservatives' and Liberals' Moral Values," found that liberals and conservatives broadly share the same moral foundations and values.

Basically, liberals and conservatives do not formulate their opinions in fundamentally different ways. And that, Joshi argues, suggests that partisans cannot account for how their opponents must be anti-reliable.

"It is not plausible to suppose that there are some people who are in general drawn toward falsity," concurs the University of Colorado philosopher Michael Huemer. "Even if there are people who are not very good at getting to the truth (they are stupid, or irrational, etc.), their beliefs should be, at worst, unrelated to the truth; they should not be systematically directed away from the truth. Thus, while there could be a 'true cluster' of political beliefs, the present consideration strongly suggests that neither the liberal nor the conservative belief-cluster is it."

Joshi acknowledges that his arguments about anti-reliability of opponents do not apply to libertarians, Marxists, or others whose views on political issues are derived from a core principle. (As it happens, I agree with five of the nine propositions Joshi highlights.) Instead, his analysis applies to partisans who confidently hold virtually all the opinions on one side or the other of the conventional right/left spectrum. "The problem for the partisan is not simply that someone disagrees with her," argues Joshi. "The problem is that political beliefs are distributed across the population in such a way as to make it highly unlikely that the partisan's beliefs are all or mostly true." This implies partisans should at least become less confident in their views.

Joshi suggests that partisans seek out and engage with the best arguments for their opponents' convictions. His recommendation mirrors John Stuart Mill's admonition in On Liberty: "He who knows only his own side of the case, knows little of that. His reasons may be good, and no one may have been able to refute them. But if he is equally unable to refute the reasons on the opposite side; if he does not so much as know what they are, he has no ground for preferring either opinion."

In an intriguing 2020 study in the Journal of Experimental Social Psychology, a team of Duke psychologists notes that "Americans have become increasingly likely to dislike, distrust, and derogate their ideological opponents on contemporary social and political issues." Why? Perhaps partisans assume that their opponents do not have good reasons for their views, leading them to believe their opponents must be intellectually or morally deficient. What would happen, the Duke researchers wondered, if we provided partisans with their opponents' arguments on such issues as concealed gun carry, mandatory body-worn cameras on police, and universal health care?

The good news is that when presented with reasons favoring their opponents' views, partisans were less likely to report that their opponents lacked intellectual ability or moral character. "Our results provide evidence that reasons serve a novel function distinct from persuasion, decision change, or acquiring knowledge," conclude the researchers. "Even if the consideration of opposing reasons does not induce a change in one's position, our results indicate that presenting opposing reasons might at least make people less likely to view their opponents negatively. This, in turn, might have the potential to make people more willing to listen to opponents and more willing to engage in genuine discussion with their opponents, which might have positive implications for compromise, fruitful deliberation, and the pursuit of a common good."

With all that in mind, let's turn back to the partisan split over the death penalty. For years I defended capital punishment in arguments with friends, colleagues, random people I met in bars, and my patient wife. Although I don't think I ever persuaded anyone to come around to my view, I hope that, since I was giving reasons for my position, at least some of my interlocutors concluded that I was not entirely lacking in intellect and morals.

At the heart of many of my opponents' desire to abolish capital punishment was their revulsion toward state-sanctioned execution. Modern civilized people, they argued, simply cannot endorse such barbarity. They recognized that I did not share that view. So they would cite studies purporting to show that the death penalty did not deter murderers. I, of course, sought to persuade them using the same sort of evidence—that is, contrary research showing that the death penalty did deter would-be murderers.

For years, these back-and-forth arguments did nothing to blunt my desire to enact vengeful justice on those who killed other people with brutal malice aforethought.

Many of my opponents did, however, deploy one argument against the death penalty that pierced my conscience: the possibility that an innocent person might be wrongfully executed for a murder that he or she did not commit. As evidence, they would point to the rising number—the count currently stands at 186—of people exonerated after being confined to death row. I had to admit that this fact was disquieting. Nevertheless, I would respond by pointing out that since the Supreme Court reinstated the death penalty as constitutional in 1976, none of the more than 1,500 people who have been executed were later conclusively proved innocent.

Then in 2021, belated DNA testing of genetic material from the murder weapon in the case of Ledell Lee pointed to an unknown someone else. Lee was executed for murder by the state of Arkansas four years ago. Perhaps he was guilty, but I find the evidence that the government executed an innocent man quite compelling in his case.

My desire for retribution has not abated even a little bit. But the arguments marshalled over the years by friends, colleagues, random bar patrons, and, yes, my wife have finally convinced me that the death penalty cannot be justly administered. I was wrong to support it.

Changing my mind on this topic was wrenchingly difficult—and this despite the fact that I was joining my fellow libertarians, who for the most part oppose the death penalty administered by the state, meaning that I had little at stake in terms of my other prior commitments.

Everyday experience and political science data amply confirm that affective polarization between partisans in the U.S. is growing. The political divide is sustained and made ever wider by the fact that many Americans tend to seek out information and arguments that confirm what they already believe and to discount contradictory evidence.

Joshi persuasively challenges the notion that one wing or the other of the conventional right/left political spectrum is likely to be right about every issue. Since that is so, he counsels partisans to be less confident of their views and to seek out and engage the best arguments of their opponents. In fact, other recent hopeful research confirms that when partisans are presented with reasons favoring their opponents' views, they think better of them. The question is: How likely is it that today's partisans will stop shouting past one another long enough to realize that the other side may have a point?

The fact that it took decades for my friends and colleagues to persuade me that I was wrong about the death penalty—even in the absence of strong affective polarization—is not a good omen.