The Politics of the Culture Wars in Contemporary America

Western societies are in the midst of a growing “culture war” between cultural socialism and cultural liberalism. The two sides in this conflict only partly overlap with the country’s partisan political divide: the culture war divides Democrats while largely uniting Republicans and independents. It presents, therefore, a risk for Democrats and an opportunity for Republicans.

In a controversy dominated by anecdotes and headlines, it is vital to systematically gather and analyze survey data on public experiences and attitudes toward culture-war issues. While this has been done for universities, this report—based principally on a new survey conducted on the Qualtrics platform—is the first comprehensive analysis of the wider American experience with, and opinion of, cancel culture, political correctness, and Critical Race Theory.

Cultural liberalism is the belief that individuals and groups should have the freedom to express themselves, should not be compelled to endorse beliefs that they oppose, and should be treated equally by social norms and the law.

Cultural socialism is the idea that public policy should be used to redistribute wealth, power, and self-esteem from the privileged groups in society to disadvantaged groups, especially racial and sexual minorities, and women. This justifies restrictions on the freedom and equal treatment of members of advantaged groups.

For some Democratic voters, a commitment to cultural socialism overrides their historical defense of free speech. Most Republicans disagree with that position. They also oppose what they perceive to be the denigration of white Americans and the nation’s past, which underlie their support for a new politics of civil rights in schools and workplaces.

Using the Qualtrics survey of mainly college-educated Americans, along with several other polls, I map public opinion across a wide range of questions pertaining to cancel culture and a set of phenomena often grouped under the label Critical Race Theory.

The main findings include:

• A majority of Americans oppose cancel culture, but a significant minority—about a third— support it, backing decisions to fire employees for legal speech that they regard as unacceptable. Cancel culture is thus not only about people being afraid to stand up for their rights; it is rooted in genuine philosophical differences in the population between cultural socialism and cultural liberalism.
• The problem of cancel culture is going to get worse, not better. Younger people are substantially more likely to support cultural socialism than older Americans, even when controlling for ideology and party identification. As today’s college graduates enter large organizations, they will mount an increasing challenge to freedom of expression.

Consider the chart below, drawn from the Qualtrics survey: two-thirds of those aged 18–25 agreed with Google’s decision to fire James Damore for questioning whether discrimination explains the underrepresentation of women in computer programming. By contrast, nearly two-thirds of those over 50 opposed the decision. 

Was Google Right to Fire James Damore?: By Age

• The strongest support for progressive illiberalism comes from the far-left fifth of America’s political spectrum, with moderate leftists much more opposed. This splits the left and means that Republicans have an incentive to highlight culture-war issues, as arguably took place in Glenn Youngkin’s Virginia gubernatorial campaign. Democrats might be advised to distance themselves, as Barack Obama has done, from both cancel culture and Critical Race Theory.

The chart below (drawn from Figure 7 in the report) shows the opinions of strong and weak Democrats, independents (“neither”), and strong Republicans to a range of cancel-culture questions. The first six questions on the left-hand side of the chart ask about cases where people have lost their jobs for legal speech, and whether people oppose the actions of the organizations that fired these individuals. The second set of five questions pertains to hypothetical scenarios such as whether a telecom supplier should disconnect a known white nationalist from the Internet, whether a public debate should be prevented about whether trans women are women or all whites are racist, whether Twitter should ban an immigration restrictionist, or whether a firm should be allowed to fire a person who makes a legal but racist comment. Across these questions, strong Democrats are often the outlier—substantially less opposed to cancel culture than weak Democrats, independents, or Republicans. 

Opposition to Selected Cancel-Culture Issues: By Party Identification

One way of visualizing how youth and far leftism interact to support cultural socialism is to look at the opinion, shown in the chart below (Figure 10 in the report), on the 2014 ouster of Brandon Eich, Mozilla’s CEO, who donated to California’s antigay marriage Proposition 8 in 2008. Controlling for age, education, and a host of other factors, the survey shows that a strong Democrat aged 35 or under has a 70% chance of supporting Eich’s ouster, while a strong Democrat over 50 has only a 40% chance of doing so. Republicans generally opposed canceling Eich, regardless of age.

Brandon Eich Should Have Been Forced Out: By Age and Partisanship

• Culture-war issues are no longer an arcane subject confined to campuses and newsrooms. They now rank above the midpoint on Republican voters’ priority lists, above religion and family values, and near the midpoint for independents. For all voters, these issues are now more important than the environment.

The chart below (Figure 39 in the report) shows that over 30% of voters ranked cancel culture and political correctness a top-three issue from a list of nine. Among Republicans, 48% ranked this issue in the top three, second only to covid/economy (63%) and immigration (60%), and ahead of moral values/religion (37%).

“Which, for You, Is the MOST Important Problem Facing the Country?” Share (%) Citing Issue as Top 3

• Diversity training appears to have a chilling effect on employees. About half of those in the Qualtrics survey had taken diversity training, and a quarter had received training that involved concepts such as white privilege and patriarchy, which are linked to Critical Race Theory or Critical Social Justice. The chart below (Figure 32 in the report) shows that people who report having attended diversity training are significantly more worried about losing their jobs or reputations for present or past speech than people who have not had this training, even when controlling for partisanship and demographics. They also feel less free to share their political views on questions like immigration or transgenderism. Alongside evidence that diversity training has a negative effect on intergroup relations at work and no impact on discrimination, this heightening of threat perceptions suggests that current forms of diversity training are counterproductive.

Self-Censorship: By Experience of Diversity Training

• Many Republicans are insulated from the culture wars because of where they live and work. Democratic and Republican voters tend to sort into neighborhoods, social groups, and workplaces that reflect their values. This helps protect conservatives and moderates from progressive illiberalism and political discrimination, forces that would cause them to selfcensor. Conservatives (including conservative Democrats) living and working in left-wing environments bear the brunt of progressive illiberalism. Democrats in Republican workplaces also report less freedom to express their views but not nearly as much as Republicans in Democratic-dominated organizations, where fewer than three in 10 Trump voters would tell a coworker how they voted.

The chart below (Figure 28 in the report) shows that nearly six in 10 Trump voters said that their workplace leans Republican, with barely two in 10 saying that their workplace leans Democratic. A similar pattern, in reverse, characterizes Biden voters. Republicans, regardless of education level, are more likely than Democrats to work in smaller businesses and are less likely to have taken diversity training.

Average Party Identification at Workplace: By Vote (%)

In sum, a divide has emerged between a cultural socialist minority and a culturally liberal majority. Among the youngest voters, cultural socialism arguably has the edge over cultural liberalism, suggesting that cancel culture is likely to worsen in the years to come. Issues of cancel culture and Critical Race Theory now rank at the midpoint in American politics and are a high priority for Republican voters and a mid-ranking issue for independents.

More than a third of all workers are concerned about losing their jobs or reputations to cancel culture. More than seven in 10 people say that political correctness has gone too far and that they self-censor their beliefs in at least some situations. Among employees with college degrees, a majority have experienced diversity training, and taking diversity training is associated with a heightened fear of misspeaking or being fired.

An overwhelming majority of voters of all political stripes oppose certain Critical Race Theory– inspired teaching methods, such as separating children by race into “privileged” and “oppressed.” However, there are large partisan gaps over whether students should be taught that the U.S. is a racist country or whether the curriculum should focus more on race and gender. Public opinion on culture-war issues tends to split the Democratic coalition while uniting Republicans, suggesting that culture-war issues are a risk that the Democrats must manage, while presenting an opportunity for the Republicans.

Western societies are in the midst of a growing conflict between what I call cultural socialism and cultural liberalism. The two sides in this culture war partly overlap with this country’s partisan political divide, but the conflict divides Democrats while largely uniting Republicans and independents. For some Democrats, a commitment to cultural socialism overrides their historical defense of free speech; among most Republicans, a perceived denigration of white Americans and the nation’s past underlies their support for a new politics of civil rights in schools and workplaces.

In a debate dominated by anecdotes and headlines, it is vital to systematically gather and analyze survey data on public experiences and attitudes toward culture-war issues. While this has been done for universities, this study is the first comprehensive analysis of the wider American experience with, and opinion of, cancel culture, political correctness, and Critical Race Theory (or CRT).

Cultural liberalism is the belief that individuals and groups should have the freedom to express themselves, should not be compelled to endorse beliefs that they oppose, and should be treated equally by the law and social norms. Cultural socialism is the idea that public policy should be used to redistribute wealth, power, and self-esteem from privileged to underprivileged groups in society—notably, historically disadvantaged racial and sexual minorities and women. The term “socialism” is used here in the European sense of egalitarianism, rather than the Marxist commitment to state ownership of the means of production. Socialism as egalitarianism allows for gradations, from partial redistribution to complete redistribution via affirmative action quotas that mirror a group’s share of the population.

Cultural socialism holds that the quest for equal outcomes among identity groups justifies restrictions on the freedom and equal treatment of members of what are called advantaged groups. Interpreting writers such as Michel Foucault or Jacques Derrida, strong cultural socialists believe that narratives, phrases, and even words reproduce intergroup power disparities. In recent years, a psychotherapeutic layer has been added to the cultural socialist worldview, claiming that some conventional sayings and words such as “anyone can make it in America” harm the mental health and safety of subaltern groups.

To cultural liberalism and socialism, we might add “cultural conservatism,” which has played an important part in the past in restricting the freedoms of groups such as gays or socialists in the name of community values such as religion or patriotism. While cultural conservatism is still a factor today in some environments, as with a mooted Tennessee law against student athletes taking the knee, or broad definitions of anti-Semitism (which could restrict criticism of Israel) imposed on university professors, threats to free speech from the right occur with considerably less frequency than threats from the left. [1]

Some people occupy a middle ground in the growing culture war—they may sympathize with a measure of cultural egalitarianism but shrink from endorsing quotas for all desirable positions in society. And few cultural liberals would argue that we should always express our true feelings to someone we have just met whose religion we disapprove of; or that we should not make at least some allowance for someone who has directly experienced a traumatic event when choosing which topics to joke about. Many cultural liberals would also agree that greater equality among groups in society is a worthy goal, even if they oppose speech restrictions and quotas.

Cultural socialism and liberalism reinforced each other during the civil rights era, when both fought against the illiberalism of cultural conservatives—a struggle that persisted on some issues, such as gay rights, into the 2010s. Nevertheless, tension arose as well, leading some cultural liberals to criticize the Berkeley Free Speech Movement for shifting from cultural liberalism to socialism after the mid-1960s. [2] Illiberalism among progressives, which first appeared among student movements on university campuses in that period, has now spread beyond the university to elite institutions, from major newspapers to publishers to large corporations, including tech firms.

While anecdotal evidence of the excesses of illiberalism, such as cancel culture, is abundant, we lack a solid foundation of survey data on which to build our understanding of the extent of this phenomenon, especially off-campus. Is the conflict between cultural liberals and cultural socialists merely an obsession of political junkies, metropolitan elites, and a very online Twitterati? Where do people stand on the divide between cultural socialism and liberalism? How pervasive are people’s experiences of progressive illiberalism? Do most people feel that cultural socialism is a malign force but are too afraid to speak out against it?

This report uses a new survey (on the Qualtrics platform), along with several existing ones, to explore this terrain (see Sidebar: Surveys Used or Cited in This Report). A few major themes rise from the data. First, while a majority of Americans oppose progressive illiberalism, a significant minority—about a third—support it. This group backs decisions to fire employees for legal speech, for example. The upshot is that cultural conflict is not primarily about people being scared to speak out, though that also comes through as an important motivation: it is a genuine battle for hearts and minds.

Sidebar: Surveys Used or Cited in This Report

Second, younger people are substantially more likely to support progressive illiberalism than older Americans, even when controlling for their political ideology. This suggests that the problem is likely to grow, not subside, as today’s college graduates enter large organizations.

Third, unsurprisingly, the strongest support for progressive illiberalism comes from the far-left fifth of America’s political spectrum. Generally speaking, cultural socialism splits the left and unites the right and most moderates. In political terms, this means that Republicans, such as Glenn Youngkin in Virginia, have an incentive to highlight culture-war issues. Democrats have an incentive to distance themselves, as Barack Obama has done, from cancel culture, as well as from unpopular practices associated with Critical Race Theory.

Fourth, culture-war issues now rank above the midpoint on Republican voters’ priority lists, above religion and family values. For all voters, these issues are now more important than the environment. Republicans are more united in opposition to progressive illiberalism than Democrats are in support, with independents more opposed than supportive. This suggests that culture-war issues are likely to benefit the political right more than the political left in the foreseeable future.

Fifth, those who have taken diversity training are significantly more worried about losing their jobs for present or past speech than those who have not done so, even when controlling for their politics and demographics. Alongside evidence that diversity training has a negative effect on intergroup relations at work and no impact on discrimination, this apparent heightening of threat perceptions suggests that current forms of diversity training should be curtailed or replaced with programs that do not carry such negative effects.

Finally, Democratic and Republican voters tend to sort into neighborhoods, social groups, and workplaces that reflect their values. This helps insulate conservatives from progressive illiberalism and political discrimination, forces that would restrict their freedom by forcing them to self-censor. Conservatives (including conservative Democrats) living and working in left-wing environments bear the brunt of progressive illiberalism. Democrats in Republican workplaces also report less freedom to express their views but not nearly as much as Republicans in Democratic-dominated organizations.

Public Opinion of Cancel Culture: How Do People Balance Cultural Socialism and Cultural Liberalism?

Attitudes toward speech restrictions in the name of protecting the vulnerable are complex but follow a fairly clear pattern. Essentially, most people agree that political correctness has gone too far and oppose cancel culture, but when it comes to the trade-off between protecting free speech and protecting the vulnerable from emotional or potential downstream material harm, opinion divides along partisan and age lines.

When people are asked about cancel culture, public opinion is generally opposed. A national poll conducted for Parents Defending Education in April 2021 shows that while 29% of people have not heard of cancel culture, among those who have, 62% have an unfavorable view while 9% have a positive view. Among Republicans, 79% have an unfavorable view, 5% are positive, and 16% are unsure or haven’t heard of it. For independents, these figures are 56% unfavorable, 8% favorable, and 36% unsure/unaware; and for Democrats, 38% are unfavorable, 18% favorable, and 44% unsure/unaware. [3] A Harvard/Harris poll in February 2021 asked people if there was “a growing cancel culture that is a threat to our freedom”: 64% agreed, including 80% of Republicans and 48% of Democrats. [4]

Yet things are not so clear-cut because people tend to endorse values that are in tension with each other. Philip Tetlock writes that people have different deep-seated value dispositions that take the form of a ranking. An individual’s policy choices tend to be resolved in favor of the option that accords with their value hierarchy—even if individuals rank two commitments, such as free speech and minority protection, relatively high. [5] This means that it is vital to map people’s trade-offs between values and not merely the values themselves. The influential “Hidden Tribes” report (by the More in Common organization) in 2018, for example, showed that “82% of Americans agree that hate speech is a problem in America today, but 80% also believe that political correctness has gone too far.” [6] What is less clear is how people trade off these competing priorities—egalitarian versus liberal—against each other.

This study focuses on differences in how adherents of the two main political parties resolve such dilemmas.

In my new survey (Qualtrics 2021), which oversamples the college-educated, 74% of the respondents overall agreed that political correctness (PC) had gone too far, with 14% opposed. This largely comports with existing research. As Figure 1 shows, there is an important 20–40-point difference between self-identified “strong Democrats” and others on a five-point partisanship scale (weak Democrats, neither Democrat nor Republican, weak Republicans, strong Republicans) on this question, which echoes the Hidden Tribes finding that there is a consensus in American society that PC has gone too far, with only a small group (8% of the population) of “progressive activists” tending to disagree. [7]

Figure 1

Political Correctness (PC) Has “Gone Too Far”: By Party

There is also an age gradient on this question, with those under 25 about 20 points less likely than older respondents to agree that PC has gone too far (Figure 2). When controlling for party identification, race, gender, and education, however, age has only a modest effect that is barely significant statistically, with respondents aged 18–25 about a half scale point less likely than the reference group, aged 36–50, to agree that PC has gone too far on a seven-point scale (strongly disagree, disagree, somewhat disagree, neither agree nor disagree, somewhat agree, agree, strongly agree). By contrast, strong Democrats are almost 1.5 scale points less likely to agree, controlling for age, race, gender, and education. Women are significantly less anti-PC, at about a third of a scale point less likely to agree. Race also has an effect but a more modest one than gender does.

Figure 2

PC Has “Gone Too Far”: By Age

However, when asked as a trade-off between cultural socialism and liberalism—“Thinking about political correctness, are you generally in favor of it (it protects against discrimination), or against it (it stifles freedom of speech)?”—opinion is considerably warmer toward PC, with 37% in favor and 41% opposed. This 41–37 split is much closer than the 74–14 split that I get when the question simply asks whether PC has gone too far. In fact, a majority of Democrats now favor PC (Figure 3). Even a fifth of independents and Republicans and nearly three in 10 of those who say PC has gone too far lean in favor of PC on this question.

Figure 3

Favor PC (It Protects Against Discrimination), or Against PC (It Stifles Freedom of Speech): By Party

Age is a more modest predictor of PC attitudes, but half the respondents under 35 are pro-PC on this question, with only about three in 10 opposed (Figure 4). The 36–50 age group is evenly split, while the over-50s break nearly five to three against. Controlling for race, gender, and education, age again makes only a modest difference. This time, it is the over-50s who stick out, as being about a quarter scale point less pro-PC than the 36–50 group, which is statistically significant at the modest p<0.05 level. Women are over a third of a scale point more pro-PC than men. Nonwhites do not differ from whites on this measure, reflecting other survey data that show that a slight majority of African-Americans see PC as demeaning to blacks rather than necessary to protect them, and do not differ from whites on this question. [8]

Figure 4

Favor PC (It Protects Against Discrimination), or Against PC (It Stifles Freedom of Speech): By Age

Another question in Qualtrics 2021 probed people’s trade-off between competing values: “When it comes to the tension between free speech and hate speech, where do you stand?” The options were:

1. Prioritize protecting against hate speech over free speech, even if this restricts speech NOT INTENDED to be hateful.
2. Somewhat prioritize protecting against hate speech over free speech: restrict free speech only where words are INTENDED to be hateful.
3. Somewhat prioritize free speech: restrict speech only where words are LIKELY to incite PHYSICAL violence.
4. Prioritize free speech. No restrictions, or restrict speech only where words are CERTAIN to incite PHYSICAL violence.

The survey data (Figure 5) show that a majority of all political stripes came down in favor of giving priority to free speech (64%) over protecting against hate speech (36%). There was less of a political gradient in the answers than in previous questions, with under 20 points separating strong Democrats from strong Republicans.

Figure 5

Prioritize Free Speech or Protect Against Hate Speech: By Party

There is a similarly modest divide on age. Individuals 25 and under split fairly evenly between the two options, while those over 51 give priority to free speech over protection against hate speech by more than two to one (Figure 6). In a statistical model, the under-35s are significantly but modestly less in favor of free speech than the 36–50 age group. Women are modestly less pro– free speech than men, but there are, once again, no significant racial differences.

Figure 6

Prioritize Free Speech or Protect Against Hate Speech: By Age Source: Qualtrics

Another variant of this question asks: “How should we view a person accused of hate speech by a member of a historically disadvantaged minority group?” The two options are “Benefit of the doubt: the accused should be believed to be innocent until proven guilty” or “Zero tolerance: the accuser should be believed until the accused proves themselves innocent.” The response: 74% chose the former option and 26% the latter. Respondents aged 25 and under broke 65–35 while the over-50s divided 77–23. Strong Democrats favored due process by a 63–37 margin; Republicans favored due process 81–19. Democrats under age 35 backed due process by a slim 55–45 margin, illustrating the combined effects of age and party identity on progressive illiberalism.

The widest gap was between the strongest pro-PC respondents (restricts free speech vs. protects against discrimination), at 56–44, and the strongest anti-PC ones, at 85–15. Despite some variation, there is a solid majority for due process in hate-speech incidents rather than “believe the accuser” logic.

The results of the Qualtrics 2021 and other surveys show that most Americans think that both political correctness and hate speech are problems, and most oppose cancel culture. But when asked to make trade-offs, strong Democrats lean more in favor of PC than others, giving greater priority to protection from hate speech and minority harm, while Republicans are more strongly anti-PC and pro–free speech. Support of free speech is overall a higher priority for Republicans than Democrats, even though it matters to all. The youngest age group is considerably more pro-PC and less pro–free speech than older age groups, even if party identification is a stronger correlate of attitudes than age.

Notwithstanding answers to abstract questions about free speech, people’s views about concrete cases where competing values come into conflict are arguably more important. How do people think about actual situations and episodes in which individuals have been fired for their views? While many people don’t understand what the term cancel culture means, they do have views about cases that concretize the more abstract tensions between free speech and the purported harm of that speech. [9] To get at these trade-offs, Qualtrics 2021 posed a wide range of questions pertaining to actual cases of cancellation, or hypothetical scenarios involving firing and no-platforming colleagues with contentious views.

This section focuses on support for two key pillars of cultural socialism: punishment and discrimination. Punishment, or hard authoritarianism, directly penalizes or censors legal speech, often through firing or no-platforming. Political discrimination, or soft authoritarianism, involves subtler social and career penalties, including discrimination in hiring and promotion, as well as social exclusion.

I begin with support for hard authoritarianism. Figure 7 compiles results in Qualtrics 2021 across 20 questions about support for cancel culture across three distinct clusters, with each shape—square, circle, or triangle—denoting the view of individuals from each of a five-point partisanship scale.

The first cluster involves people’s views of whether individuals in real, specific cases should have been fired from their jobs. The second cluster focuses on whether people support firing or prosecuting hypothetical individuals for political donations or contentious speech. The third cluster asks whether social media platforms or utilities should ban debate on contentious subjects like the trans issue or be permitted to remove individuals whose tweets or posts are controversial but otherwise legal.

Results for each cluster are sorted from the question where there is the greatest average opposition to canceling to that with the weakest opposition to dismissing or banning dissenters. The results are somewhat affected by the wording of each question, but most of the difference in opinion is based on the substance of the question asked. Replies for independents are a reasonable guide as to where the average of the sample lies.

Cluster one begins with the firing of Emmanuel Cafferty, a Hispanic San Diego Gas & Electric worker fired for inadvertently making an “OK” sign while cracking his knuckles after a photograph by an activist ended up on Twitter (the OK hand sign is a common gesture of “all is well” in the U.S. and with similar connotations in other countries and cultures, but it acquired an association with some white power groups a few years ago). Of independents, 94% said that he should not have been fired. The most contentious case appears to be that of Steven Salaita, a University of Illinois professor who wrote numerous tweets against Israel, including: “At this point, if [Israeli Prime Minister] Netanyahu appeared on TV with a necklace made from the teeth of Palestinian children, would anybody be surprised?” This time, 44% of respondents (47% of independents) opposed firing him.

In cluster two, 94% of independents would not fire a group of “white male employees” who refused to take diversity training, but just 51% of them would oppose prosecuting someone who used a racial slur against a black person in a meeting. (Uttering a racial slur is not a prosecutable offense, but the question can serve as a test of people’s opinions.) The rightmost question in this cluster has a three-part wording so is not as easily comparable, but just 40% oppose the firing of a colleague who did research showing that diversity tends to lower social solidarity.

In cluster three, 75%–77% of independents opposed the idea of banning people with far-left or anti-immigration views from Twitter and Facebook to reduce support for defunding the police or restricting immigration. But fewer independents—61%—would ban a firm from firing someone for racist but legal speech.

Figure 7

Opposition to Cancel Culture: By Party

The first point to note is that there is only one question (about Salaita) where there is majority support for canceling—yet even here, it is borderline, with 44% of the sample (47% of independents) opposing dismissal. In two other instances—whether to prosecute for racial slurs uttered against whites (52%) and blacks (51%), the balance of opinion is evenly divided. However, on most questions, the share who oppose dismissals and bans is around 70%, with a majority of weak Democrats opposed. This pattern of public opinion represents a potential electoral opportunity for Republicans and a liability for Democrats.

The second point is that an individual’s position on a five-point partisanship scale correlates with his view on whether someone should be canceled. Strong Democrats are substantially more likely to endorse dismissal, compared with weak Democrats, independents, and Republicans. Of strong Democrats, 66% oppose the firing of Emmanuel Cafferty for being photographed inadvertently making the OK sign, compared with 94% of independents. Of strong Democrats, 36% would prevent a firm from firing someone for legal but racist speech, compared with 61% of independents. On most questions, strong Democrats are about 20 points more in favor of cancellation than independents.

Weak Democrats are slightly more favorable, and Republicans slightly less favorable, toward canceling than independents, but the effects are much smaller. Again, majority sentiment is against cancellation, and if the Democrats come to be seen as the party that supports cancel culture, they will be on the wrong side of public opinion. This political implication would seem to be that Democrats are advised to convey, as Barack Obama did, that they do not approve of cancel culture. [10]

Age is the next most important factor governing support for cancellation (Figure 8). On average, people 18–25 are 20 points more likely to back a firing or no-platforming campaign than the over-50s. About 40% of the youngest respondents oppose the firing of Mozilla CEO Brandon Eich for supporting California’s anti–gay marriage Proposition 8 in 2008; or Professor Charles Negy, who tweeted that various forms of affirmative action amounted to black privilege and should not be shielded from criticism; or Professor Philip Adamo, who read a passage from black writer James Baldwin that used the N-word. The only cases where the young are as tolerant as their elders tend to be those involving curbs on left-wing speech, such as the Salaita case, banning a far leftist from social media, or being allowed to debate whether all whites are racist.

Figure 8

Opposition to Cancel Culture: By Age

To what extent do the age effects merely reflect the fact that younger people tend to vote Democratic more than older people? To find out, I developed a cancel-culture index by combining opinions on six cancellation incidents—Brandon Eich, Charles Negy, Philip Adamo, James Damore, Gina Carano, and Steven Salaita—through factor analysis. [11]

I then ran a statistical model on this cancellation index, asking which characteristics best correlate with it, controlling for confounding effects. The results (Figure 9) sort the factors that most closely correlate with a person supporting cancellation, by their importance. They show that being a strong Democrat is by far the strongest predictor of supporting the firing of these individuals. However, younger people are significantly more likely than older people to support dismissal, even when you take into account their more Democratic Party leanings. This indicates that we are likely to see more support for cancel culture as millennials become a larger share of the workforce.

In addition, those who have attended diversity training are significantly more likely to favor a firing campaign. It’s impossible to definitively parse whether this is because diversity training sensitizes people to the feelings of protected groups and thereby increases support for dismissal because it is framed as protecting them—or whether those who are already on board with this agenda are more likely to have volunteered to attend diversity training. Yet an important piece of evidence indicates that diversity training has an independent effect—its impact on increasing support for dismissal campaigns affects people regardless of where they lie on the political spectrum. Strong Republicans who attended diversity training, for example, are significantly more pro-cancel than strong Republicans who have not.

Jews and Muslims, two small religious minorities, are more pro-cancel than those of another or no faith. The remaining correlates have small effects: weak Democrats are somewhat more procancel than average, while weak Republicans and married people are somewhat more anti-cancel.

Whites are slightly more likely to oppose cancel culture than minorities when you account for the characteristics listed above, but the effect is not statistically significant. Women and men also don’t differ in their views.

Figure 9

Predictors of Pro-Cancel Sentiment

To visualize the impact of partisanship and age, Figure 10 plots the predicted support for Mozilla CEO Brandon Eich being forced to step down in 2014 for having supported the 2008 anti–gay marriage Proposition 8 in California. For example, a strong Democrat aged 18–35 has a 0.7 chance of agreeing that Eich should have been forced out. This falls to 0.43 for a strong Democrat over age 51. The gap between young and old seems to matter most among independents and is least pronounced among strong Republicans.

Thus, there is just a six-point gap between young strong Republicans (0.24 chance of being pro-cancel) and strong Republicans over 51, among whom there is just a 0.18 probability of backing Eich’s ouster. Young weak Republicans are, however, about 20 points more in favor of canceling Eich than weak Republicans over 51. A key point is that young people who are in the political center have a 50% chance of backing the firing of Eich, compared with centrist voters over 35 who have less than a 20% chance of doing so.

In short, millennials and Gen-Z are more likely to endorse progressive authoritarianism than Gen-Xers or baby boomers. But this is more a straight-line story about age than one of sharply defined generations reflecting different formative events or the use of social media. That is, age has a linear and gradual effect, with younger people more pro-cancel than older people across the age range.

Figure 10

Brandon Eich Should Have Been Forced Out: By Age

If threats of discipline, including being fired or no-platformed, represent direct censorship and hard authoritarianism, political discrimination involves what I term “soft authoritarianism.” [12] In many ways, people’s fear of being discriminated against in hiring, promotion, or even social interaction may be more important than their fear of being terminated or no-platformed. Both fears chill speech, reducing expressive freedom. Later in the report, I will discuss which of these two fears is more pervasive; for now, the focus is on support for discriminatory behavior, rather than perceptions of the victims of discrimination.

American politics has become increasingly polarized since the 1980s, so much so that party identification is now a major factor in dating. For example, only 6% of the 93% of Ivy League female students who don’t support Trump would have no problem dating a Trump supporter. Forty percent of Americans wouldn’t want their child to marry someone who identifies with the opposing political party. In simulations, political prejudice is pervasive and openly expressed in a way that racial prejudice is not. [13]

Studies asking academics whether they would discriminate against hires, papers, grants, or promotion applications from the opposing party found that 15%–50% would so discriminate. [14] Among American social sciences and humanities academics, my report for the Center for the Study of Partisanship and Ideology (CSPI 2021)—which analyzed several surveys, mainly those that I conducted—found that 25% openly admitted that they would discriminate against a Trump supporter, with 33% saying that they “didn’t know.” [15] In a revealed-preference test, the actual share willing to discriminate was 40%. For the Qualtrics 2021 survey, I removed the “didn’t know” option, changing the question to a two-category yes/no question and found that 35% of Biden supporters in the sample openly said that they would discriminate against a Trump supporter for a job. Using the same revealed-preference test showed that 49% of Biden voters would discriminate against a Trump supporter. Even if we allow for measurement noise due to the fact that only 450 Biden supporters were polled, this suggests that these nonacademics resemble their academic counterparts in their willingness to discriminate against Trump voters for a job. Yet political discrimination from the left is not primarily about Trump: it applies also to conservatism and national populism more generally. The right also discriminates against the left. A series of studies in the U.S. and Europe from 2012 onward asking academics whether they would discriminate against hires, papers, grants, or promotion applications from individuals of the opposing party or ideology found that 15%–50% would do so. [16] In Britain, for instance, I found that one in three academics would not hire a Brexit supporter.

Figure 11 presents the full list of Qualtrics 2021 questions tapping into people’s willingness to discriminate against individuals from the other party. Broadly speaking, Trump and Biden supporters both display similar and substantial levels of bias, though Biden voters are more consistent in their views than Trump voters. Sixty percent of Biden voters are either uncomfortable (20%) or unsure (40%) about having lunch with a Trump supporter, and 67% would be uncomfortable or unsure about sitting down to lunch with a woman who opposes the idea of granting trans women access to women’s shelters. For comparison, my CSPI 2021 report found that 60% of Clinton-voting academics would be uncomfortable (26%) or unsure (34%) about sitting down with a Trump supporter, a similar result.

A total of 51% of Biden voters would prefer hiring a Sanders supporter over a Trump supporter when the two have the same merit, rather than remaining neutral between the two. A total of 45% of Trump voters would favor the Trump supporter for the job in the same question. For comparison, 50% of Clinton-voting academics in CSPI 2021 also chose the Sanders supporter over the Trump supporter or being neutral, indicating that a similar level of “soft” political discrimination exists among Democrats inside and outside academia.

There is some political skew when it comes to whether people think that firms should have the right to politically discriminate. In Qualtrics 2021, 64% of Trump voters think that companies should be able to discriminate against an Alexandria Ocasio-Cortez (AOC) supporter for a job, but just 31% of them think that firms should be able to do the same against a Trump supporter. Democrats are more consistent, with 46% believing that a firm should be able to discriminate against an AOC supporter for a job and 49% saying that a firm should be able to do so against a Trump supporter. This could partially reflect the fact that AOC is on the left wing of the Democratic Party and thus less appealing to the modal Democrat in my sample, whereas Trump has been the Republican presidential candidate on two occasions.

Finally, 27% of both Trump and Biden supporters would support firing a business executive who donated to the Biden campaign. For an executive donating to Trump’s campaign, a similar share of Biden supporters would support firing, but the number dips to 13% among Trump voters. The Cato Institute’s July 2020 survey asked the same question and found that 28% of Trump voters would support firing an executive who donated to Biden’s campaign while 39% of Clinton voters would fire an executive who donated to Trump’s campaign. [17] Here the timing of the Cato survey, close to an upcoming election, may have led to the 11-point higher share of Democrats endorsing firing. Using a similar statistical model across both surveys (based on age, race, gender, education, and party identity), the one consistent finding is that there is a bigger partisan gap over whether to fire the Trump donor than there is over whether to fire the Biden donor.

Figure 11

Political Discrimination: By Political Party

Younger respondents are significantly more politically biased, even controlling for party identification, race, and gender. Some 55% of Biden voters 25 and under would not hire a Trump supporter for a job, dropping to 39% for the 26–49 group and 29% for those over 50. Only 23% of young Biden supporters said that they would be comfortable having lunch with a Trump supporter, compared with 42% of Biden voters over 25. In general, age had about a quarter to a third as much statistical power as party identification for discrimination questions. This is somewhat below its power to predict support for cancellation, where age was about half as important as partisanship.

Firing for speech or being discriminated against is not the only expression of illiberalism. A sliding scale runs from termination through suspension to loss of privileges to social pressure. The authoritarian face of “inclusion” initiatives like diversity training or mandatory requirements to diversify university reading lists is punishment: those who choose not to comply face coercive sticks and disincentives. To what extent are supporters of inclusion initiatives willing to accept the restrictions on people’s freedoms required to achieve their progressive aims?

In CSPI 2021, [18] I discovered that more than 40% supported mandatory race and gender quotas on reading lists (such as 30% of the readings authored by women, 20% by authors of color) while barely 30% opposed them. However, about half of American academics do not wish to see any sanctions applied to non-compliers—30% endorse some form of punishment or loss of opportunity, with another 19% preferring social pressure only (Figure 12). Clearly, most academics recoil from the illiberal implications of the curriculum “decolonization” policies that many of them support.

Nonacademics surveyed in Qualtrics 2021 are more authoritarian than academics, though the questions differ: refusing to diversify a curriculum in the academic case, refusing to take diversity training in the nonacademic case. Though the questions are not identical, 32% of nonacademics would suspend or fire a group of several “white male employees” who refused to take diversity training, with a further 18% saying that they should lose opportunities at work. Among left-wing nonacademics, nearly half endorse firing (13%) or suspension (35%), with a further 22% saying that they should lose opportunities at work. Younger respondents were not harsher when ideology and party identification were taken into account. These results indicate that even where people do not support firing, there is a substantial reservoir of support among progressives for forms of coercion to conform to cultural socialist values that violate individuals’ freedom of conscience.

Figure 12

Preferred Punishment for Diversity Noncompliance: U.S. Academics and Nonacademics

Many questions take a binary approach to cancellation, but it is vital to explore the nuanced middle ground, where people are attracted to progressive ends but repulsed by authoritarian means. A further question on a hypothetical cancel campaign (similar to the one that cost software engineer James Damore his job at Google) [19] explored this subtle structure of opinion and compared it with similar studies (Figure 13). The question: “If someone in your workplace did research showing that greater ethnic diversity leads to increased societal tension and poorer social outcomes, would you support or oppose efforts by staff to let the person know to find work elsewhere?” Overall, only 19% of people supported the campaign to oust the individual, rising to 28% among left-wing respondents but also including 12% of Trump voters.

Narrowing the focus only to left-wing respondents and comparing the results with three other surveys in Britain and America shows that left-wing support for the dismissal campaign runs at only 9%–28% but opposing the dismissal effort is also a minority position, at 29%–43%. The modal response, 43%–58%, is to be unsure. There is some measurement noise on this response as weary subjects gravitate to the middle category. But the answer also reflects cross-pressure between cultural socialist and cultural liberal value commitments because, on a less controversial question involving whether someone who wants less immigration should be fired, the academic surveys showed 76%–83% opposition to a campaign to fire such an individual, with very few opting for the uncertain category. [20] Left-wing professors and lecturers are also about 15 points more pro–free speech than left-wing nonacademics.

Figure 13

Attitude Toward a Campaign to Dismiss a Researcher Who Finds that Diversity Reduces Solidarity (Left-Wing Respondents)

Younger respondents were substantially more likely to support the firing campaign than older respondents. For instance, while 19% of the Qualtrics 2021 sample favored firing the researcher, this rises to 29% of those 35 and under and 39% for people 25 and under. In fact, age was only somewhat weaker than a person’s ideology or party identity in predicting support for canceling. This echoes findings in the academic surveys that showed, controlling for other factors, that academics under 35 were twice as likely to endorse a dismissal campaign as academics over 50. [21]

Would people who oppose diversity initiatives that carry authoritarian implications for freedom of speech or conscience be willing to speak against them? If not, a committed minority of cultural socialists, leveraging societal taboos, can drive illiberal policy change, counting on fear to quell majority opposition. Timur Kuran refers to this as “preference falsification,” in which a majority is cowed into silence by a social norm. People may appear to support regime dogma while privately opposing it. Faced with harsh sanctions for speaking out, many come to believe that others support the regime when, in fact, most oppose it. [22] Cass Sunstein terms this emperor’s- new-clothes scenario “pluralistic ignorance,” arguing that this characterized mass publics during the Nazi and Soviet regimes and does so in organizational life today. [23]

To what extent do employees genuinely believe in the Critical Social Justice (or CSJ) [24] ideas that place the emotional safety and psychic equality of historically marginalized groups above employees’ right to freedom of expression and conscience? Or are they acting in the way Kuran and Sunstein suggest, secretly opposing CSJ ideas but fearing the reputational and career consequences of opposing regime ideology? In what follows, I attempt to weigh these competing arguments for why progressive illiberalism has been able to make headway in organizations.

Scenario 1: Affirmative Action Plan

In the first hypothetical scenario, survey respondents were asked, “Please imagine you are in the following situation at work. There is a proposal at a staff meeting of 20 people to have a quota of at least 30% minorities and 50% women for new hires and promotions. How would you respond in real life?” The options included:

1. Raise my hand and express OPPOSITION to it
2. Not willing to raise my hand alone but willing to raise my hand in a show of hands AGAINST it
3. Not willing to voice opposition or raise my hand in a show of hands AGAINST it but willing to vote AGAINST it in an anonymous survey
4. Not willing to voice support or raise my hand in a show of hands FOR it but willing to vote FOR it in an anonymous survey
5. Not willing to raise my hand alone but willing to raise my hand in a show of hands FOR it
6. Raise my hand and express support FOR it

The responses show two important things (Figure 14). First, people’s beliefs matter more than preference falsification: 47% of people support the affirmative action plan, and 53% are opposed. If a spiral of silence was the main factor quelling opposition to equity and diversity, we would expect people to be overwhelmingly opposed but afraid to voice their beliefs by raising their hands.

Second, preference falsification is also a factor: 41% of those opposed to the affirmative action quotas are willing to raise their hands alone, but 37% would express opposition only in an anonymous survey. By contrast, 55% of affirmative action supporters would be willing to express their support by raising their hands alone, and just 26% would express support only in an anonymous survey. These results are confirmed when people were asked to guess what share of opponents of the plan would be willing to raise their hands to express opposition and support. Respondents thought that 54% of the plan’s supporters would be willing to raise their hands alone and that 39% of the plan’s opponents would do so. As it happens, this guess is very close to the actual figures that people gave of 55% and 41%.

Figure 14

Response to a Hypothetical Affirmative Action Plan (% Support by Option)

Next, I asked those who indicated that they oppose the measure but would not raise their hands alone why they wouldn’t. Was this about being a team player, not being seen as a racist (or antiwhite), or seen as a Republican (or a Democrat)? Figure 15 shows the distribution of responses.

Nearly three-fourths of the opponents of the affirmative action plan agreed that not being seen as racist would be a motivation for not raising their hands, with only 9% disagreeing. While opponents who would remain silent were also more likely to do so for consensual reasons of being a team player, concerns about being seen to violate a norm against racism were paramount. Among supporters who would remain silent, it is less clear what the principal motivation is: 45% agreed that they wouldn’t raise their hands because this would make them look antiwhite, but 25% disagreed that this would be a motivation. People were also less concerned about being politically identified as Democrats or liberals. Responses were only modestly affected by the average ideology and politics of the person’s workplace.

Figure 15

Affirmative Action Plan: Reasons for Remaining Anonymous (%)

Scenario 2: Diversity Pledge

A second question sought to probe the willingness of people to speak up against a progressive authoritarian proposal:

"Please imagine you are in the following situation at work. There is a proposal at a staff meeting of 20 people to require all current and future employees, as a condition of employment, to sign a pledge to “increase diversity and achieve equal outcomes between groups, and to combat structures of white supremacy and patriarchy operating within our organization.” Those who refuse to sign must leave the company. How would you respond in real life?"

The six options were identical to those of the previous question. Once again, beliefs are central: 57% oppose the proposal, but 43% would support it, even as it violates individuals’ freedom of conscience (Figure 16). Preference falsification is lower among opponents than in the previous example, with 51% willing to raise their hands, compared with 41% last time. A third of opponents, however, would wish to remain anonymous, only a bit below the 37% in the affirmative action plan scenario.

Supporters of the diversity pledge are generally willing to raise their hands: 55% for the pledge, compared with 57% for the affirmative action plan. Some 27% prefer to remain anonymous, compared with 25% for the previous scenario, a very similar share. People guessed that 40% of the opponents and 54% of the supporters would raise their hands alone. This time, estimates were accurate for supporters to one point, but off by 11 points in underestimating that 55% of opponents said that they would be willing to stand up and be counted. Preference falsification is thus even less important in this case than in the affirmative action case.

Figure 16

Response to Compulsory Diversity Pledge (% Support, by Option)

Probing the reasons why people were unwilling to raise their hands alone in Figure 17 paints a very similar picture to Figure 15. Opponents of the diversity pledge were principally concerned about being seen as racist, though they were also more likely than supporters to cite the importance of being a team player as a reason for not sticking their necks out. The partisan cast of a person’s workplace mattered somewhat more this time, with 73%–74% of opponents in liberal or Democrat workplaces citing not being seen as racist as a motivation, compared with 63%–64% of opponents working in conservative or Republican-leaning organizations.

Figure 17

Diversity Pledge: Reasons for Remaining Anonymous (%)

The political and ideological tenor of people’s workplaces was somewhat important in determining whether opponents chose to remain silent. Some 53% of those opposed to the diversity pledge who work in organizations that lean conservative were willing to raise their hands alone, but in liberal workplaces, 37% would do so. In fact, more opponents working in liberal workplaces chose to remain anonymous on this issue than would speak up. For supporters of the measures, and for the affirmative action scenario, the dominant ideology of the workplace had little effect.

These findings tell us that genuine differences of belief account for the lion’s share of the variation in people’s support for culturally socialist measures that arguably violate people’s freedom of expression and conscience, and their right to equal treatment. Preference falsification plays a role, with 33%–37% of opponents choosing to remain anonymous, rising to 44% among those working in liberal workplaces. Nearly half of opponents would not raise their hands alone; about a quarter of supporters also opted to remain anonymous.

By far, the most important concern of opponents who chose not to express their disapproval was their desire not to be seen as racist—demonstrating the power of taboos around racism to quell dissent among an important minority of those who oppose illiberal progressive policies.

In both scenarios, a slight majority of respondents opposed the proposals. The tilt was 53–47 against the affirmative action plan and 57–43 against the diversity pledge. Focusing only on those willing to raise their hands alone or as a group, the balance of opinion shifts—but only modestly, to 49–51 and 54–46, respectively. In the breach, many opponents who say that they would express their views may, in fact, lose their nerve. Even if the preference falsification problem was solved, we are a long way from an emperor’s-new-clothes scenario in which most people oppose the regime but are afraid to say so. Two in three Biden voters and a similar share of individuals aged 18–25 back the affirmative action scenario, and 60% of both groups support the compulsory diversity pledge. In liberal workplaces such as newsrooms, university faculties, publishing houses, or tech firms, these results suggest that even anonymous voting wouldn’t turn the tide against illiberal measures.

Two examples from the rarefied world of England’s Oxford and Cambridge Universities illustrate that both ideas and preference falsification matter. Arif Ahmed, a philosophy professor, reports that it was very difficult for him to acquire the 25 signatures needed to get his free-speech motion to change the university’s policy from mandatory “respect” for differences to “toleration” of differences on the ballot at Cambridge, but it passed with 80% faculty support, suggesting that preference falsification around questions of racism was important. [25] On the other hand, students at Oxford’s Oriel College voted to remove a statue of the British imperial figure Cecil Rhodes from the campus, and those at Magdalen College voted to remove a photo of the Queen from a common room for her ostensible ties to colonialism, indicating that progressive ideas were held by the majority—though in neither of these cases was a free-speech issue at stake. [26] It is important to note that in the Ahmed case, many of those voting were senior or retired, and some were nonacademics. Were the vote to be held among an age-representative sample of current faculty, it arguably would have been closer.

• Summary Results of the Two Scenarios

A fine-grained analysis of attitudes toward firing and less dramatic forms of punishment indicates that ideology is the strongest predictor of attitudes toward cancellation. While only 8% of respondents would fire a group of white male employees who opted not to take diversity training and just 20% back a hypothetical firing campaign for a controversial researcher, the share who retrospectively endorse the firing of James Damore reaches 42%. It seems that while many balk at actively canceling someone, many more are willing to endorse a firing decision once it has been taken by an institution. In addition, a third of respondents feel that those who don’t take diversity training should be suspended or fired—rising to half, if we include those who say that the dissenters should lose opportunities at work. Thus, there is a considerable minority who support disciplinary action for legal speech around hot-button issues.

The term “liberal” is conventionally used by American political scientists as a euphemism for leftist, with self-described ideology assessed on a 5-point “very liberal” to “very conservative” scale. Based on this meaning of “liberal,” seven in 10 liberals endorsed the mandatory affirmative action and diversity pledge proposals that carry punishments for noncompliance. In my academic surveys, over half of liberals endorsed mandatory race and gender quotas on reading lists. The appetite for punishing dissenters varies. In the Qualtrics 2021 survey of nonacademics, fully seven in 10 liberals would punish those who don’t attend diversity training through firing, suspension, or making people lose opportunities at work. Among academics, just one in 10 would fire those who refused to abide by race/gender quotas on their reading lists or give them less desirable teaching assignments, but nearly half felt that making them attend diversity training or applying social pressure would be appropriate. Here we see that the authoritarian, politically exclusionary, face of diversity and inclusion initiatives has an important base of support, especially in liberal workplaces, though academics are less punitive than nonacademics.

Young people are somewhat less tolerant of those who dissent from progressive edicts. Some 63% of left-wing (“liberal”) respondents aged 25 and under endorse firing or suspending dissenters, compared with 46% of liberals over 50. Among left-wing academics under age 35, 24% would punish dissenters by giving them worse teaching or making them cancel their courses, compared with 7% among faculty over age 35. As the more culturally socialist millennials enter the workforce in larger numbers, progressive illiberalism is likely to be openly supported by a growing share of employees.

The layers of opinion that permit small groups of committed ideologues to move policy in an illiberal direction—a geology of cancel culture—are summarized in Figure 18. As the results for the question on whether to fire a researcher for claiming that diversity reduces social solidarity show, there is, depending on the question, a “true believer” group of 10%–30% among people on the left who support firing dissenters for transgressing progressive orthodoxy on identity politics issues, a share that rises to 20%–40% among those under 30.

Beneath the true believers lies a “conflicted” group of 40%–50% who sympathize with progressive ends but not necessarily with illiberal means. Then there are the free-speech proponents who number in the 35%–55% range, falling to 20%–45% among those under 30. But free-speech advocates are susceptible to preference falsification: among free-speech proponents in the diversity-solidarity example who also opposed the affirmative action and diversity pledge initiatives, more than 40% were unwilling to raise their hands alone. Of this 40%, nearly threequarters agreed that not being seen as a racist was a motive for staying silent, with only 5%–15% disagreeing. Assuming that the reluctant group of free-speech proponents in an actual meeting would be considerably larger, this means that activists can rely on radioactivity from the racism taboo to silence many, if not most, of their opponents.

Based on surveys, this means that, in left-leaning workplaces, a committed 10% or 20% minority can count on a conflicted group of about half the total to nod through illiberal proposals or even cancellation attempts while taboos on race, gender, or sexuality silence many of the remaining 35%–50% of free-speech proponents. Hence a committed group of illiberal activists can leverage progressive aims and taboos as a force multiplier to amplify their power by a factor of five or 10. Meanwhile, cohort change is steadily increasing the number of illiberal employees while eroding the ranks of those willing to defend free speech, thereby increasing the prevalence of progressive authoritarianism.

Figure 18

The Geology of Cancel Culture in Liberal Workplaces

General Perceptions of Freedom

The Qualtrics 2021 survey showed that three-quarters of people think that political correctness has gone too far, even if a third of those who think it has gone too far support the need for PC (Figures 1 and 2). Likewise, a majority of people report self-censoring their speech: when asked whether they agree with the statement “The political climate these days prevents me from saying things I believe because others might find them offensive,” 64% of the sample agreed, including 55% of Biden voters and 73% of Trump voters. This accords well with a July 2020 Cato Institute survey that found, using the same question, that 62% of respondents agreed, up from 58% in 2017. The Cato survey recorded a larger partisan split, with 80% of 2016 Trump voters saying so, compared with 49% of Clinton voters. [27]

A 2021 British survey found that half the people felt less free to speak their minds than five years ago; 43% said that they felt less free to speak their minds on immigration and 42% on transgender issues. [28] The Qualtrics 2021 survey shows, remarkably, an identical set of numbers: 43% say that they feel less free to express their views on immigration than five years ago, and 42% say the same for trans issues. On immigration, 51% of Trump voters and 36% of Biden voters said that they felt less free to discuss these issues than five years ago. On trans issues, the corresponding figures are almost identical, at 51 and 35.

Expressive Freedom in the Workplace

According to the British survey cited above, people were most comfortable sharing their views at home (56%), followed by pubs and restaurants with friends (45%), and then the office (24%). People are more guarded at work—why might this be? Aside from explicit codes of ethics and norms against bringing politics into the office, political discrimination and punishment—which, as mentioned earlier, I respectively term “soft authoritarianism” and “hard authoritarianism”— tend to reduce people’s expressive freedom in the workplace.

More politically unbalanced workplaces have a chilling effect on speech, especially for Republicans. The Qualtrics 2021 survey asked people to rate their workplaces on a 1 to 5 party identity scale (strong Democrat, weak Democrat, neither Democrat nor Republican, weak Republican, strong Republican) and on a 1 to 5 liberal to conservative ideological scale (very liberal, somewhat liberal, moderate, somewhat conservative, very conservative). Some 96% of Trump voters working in very liberal workplaces said that the political climate prevented them from speaking their minds on some issues, compared with 48% of Trump voters in conservative workplaces. Among Biden voters working in very conservative workplaces, 74% felt similarly restrained, as did 54% of Biden voters in liberal workplaces. Among Trump voters working in very liberal workplaces, 81% said that they felt less free to express their views on immigration than five years ago, while 67% said the same about the trans issue. Figure 19 shows the effects by 2020 vote, using a three-category workplace ideology classification.

Figure 19

Expressive Freedom: By 2020 Vote and Political Leaning of the Workplace

Political speech at work may carry hard or soft sanctions. Hard authoritarianism directly censors people, involving people being fired or disciplined for legal speech, past or present, inside or outside the workplace. Punishment also serves as a warning to others who might violate orthodoxy, leading many to self-censor. The soft authoritarianism of political discrimination produces similarly chilling effects, cowing dissenters into keeping their political beliefs to themselves in order to avoid adverse effects on their social life, reputation, or career advancement.

Fear of Being Canceled

To establish how many people are seriously worried about being punished by their employers, Qualtrics 2021 asked: “Are you worried about losing your job or reputation because someone misunderstands something you have said or done, takes it out of context, or posts something from your past online?” Some 37% said that they were worried, and 63% said that they were not. The figures did not vary much by partisanship, with 35% of Trump voters and 36% of Biden voters expressing this concern. For comparison, a Harvard/Harris poll of 1,778 people in February 2021 asked if they were afraid that if they expressed their true viewpoints on Twitter, they might be “banned or fired from my job”; the poll showed that 39% were concerned, including 47% of Republicans and 35% of Democrats. [29]

Despite the fact that there were no differences in fear of discipline between Trump and Biden voters in the Qualtrics survey, all respondents in liberal workplaces, especially Trump voters, felt more concerned. Respondents were asked to rate their workplaces on a 1 to 5 party identity scale and a 1 to 5 liberal to conservative ideological scale. In environments rated as very liberal, 57% of Trump voters worried about losing their job or career due to past or present speech, as did 45% of Biden voters (Figure 20). By contrast, those working in very conservative environments were less concerned, with 32% of Trump voters and 36% of Biden voters worried that something they do or say could lead to them losing their jobs or reputations. This indicates that a left-wing atmosphere at work is more censorious, on average, than a right-wing atmosphere, especially for Trump voters.

Figure 20

Worried About Losing Job or Career for Past or Present Speech: By Workplace

Another important dimension is age. Those under 45 in the Qualtrics 2021 survey (Figure 21) are considerably more worried about losing their jobs or reputations due to things they might say than those over 45. Nearly half the workers under 45 worry about losing their jobs and careers for speech, compared with 30% for those over 45 and 25% among the over-60s. Older people are also much less worried about losing their job for speech in the Cato survey. Part of this is a function of the young having more to lose, but this is not the whole story.

Figure 21

Worried About Losing Job or Career for Past or Present Speech: By Age

Another major aspect of the picture is workplace diversity, which affects the degree to which white employees fear for their jobs. Though I didn’t ask specifically about the racial diversity of an individual’s workplace, it was approximated through an individual’s zip code of residence or local area. Among the 161 whites living in white-minority zip codes, 50% worried about being canceled, whereas, among the 375 whites living in areas that were over 80% white, only 30% did. Figure 22 shows that, even with controls for age, party identification, gender, firm size, and the ideological climate at work, white employees living in whiter neighborhoods are less worried about being fired or shamed. In a hypothetical neighborhood with almost no whites, the few whites residing there have a 0.53 chance of being worried for their job due to speech (with other factors held constant), compared with a 0.27 chance among a white person residing in a lily-white area.

Figure 22

White Employees’ Worry About Losing a Job or Career

The size of the organization where a person works is of borderline statistical significance: 40% of people working in organizations with more than 500 employees worried, compared with 28% in organizations with fewer than 50. Larger institutions have more workers with advanced degrees and lean further left. Left-leaning institutions like universities, publishing houses, or newsrooms are thus more chilling places to work. Finally, those who post regularly on social media are more fearful of losing their jobs and reputations, even when controlling for age, education, politics, and other factors: 40% of those posting several times a day said that they worried, compared with under 30% of those not on social media.

Political Discrimination at Work

Asking whether partisans would openly share their political views with colleagues is a useful measure of how politically discriminatory the workplace is. In Qualtrics 2021, 52% thought that a Trump supporter would share their views, and 20% said that they would not, while 65% said that a Biden supporter would feel free to do so and 12% said that they would not. These numbers are quite different from the CSPI academic surveys: 70% of U.S. social sciences and humanities (SSH) academics said that Trump supporters would not share their views, and only 12% said that they would. By contrast, 91% of these academics reported that Biden supporters would openly share their views. Only 2% said otherwise.

The big difference between the surveys reflects the more politically balanced social climate prevailing in most nonacademic workplaces. In SSH academia, those on the left outnumbered those on the right in Qualtrics 2021 by a ratio of about 14 to one—a number broadly confirmed elsewhere based on the party registration of academics. [30] Moreover, 28% of SSH academics identified as far left, compared with 13% of nonacademics in Qualtrics 2021.

Unsurprisingly, the dominant political tenor of people’s workplaces had a major impact on their sense of whether partisans would feel comfortable expressing their beliefs to a colleague. What is noteworthy is that Biden voters are less affected by the political complexion of the workplace than Trump voters. For example, 57% of the employees in a Republican-leaning workplace (whatever their 2020 vote) say that Biden voters would feel comfortable sharing their views with colleagues, but only 40% of those working in Democratic-leaning workplaces say that Trump voters would feel comfortable doing so (Figure 23).

If the focus is restricted to the perceptions of partisans, 53% of Biden voters working in Republican- leaning workplaces say that Biden voters would be comfortable sharing their views, but just 29% of Trump voters in Democratic-leaning workplaces agree. In a separate question, 48% of Trump voters working in Democratic-leaning workplaces said that their workplace was a “hostile environment” for their political beliefs, compared with 24% of Biden voters working in Republican-leaning workplaces who described their workplace this way.

Political conformity is more powerful in Democratic work environments, though it is also true that Trump was arguably the more controversial candidate, so this could affect the results, compared with what might be expected for a different Republican or Democratic candidate. Another possibility is that, in elite environments, Democrats are more intolerant than Republicans— though college-educated voters did not differ much from non-college-educated voters in their assessment of whether a Trump voter would feel free to speak.

Figure 23

Would Trump Supporters or Biden Supporters Express Their Beliefs at Work?

Though Trump voters feel less free to express themselves at work, there is still a big difference between campus and off-campus work environments. In my CSPI 2021 academic surveys, just 15% of academics in left-leaning departments (the vast majority of departments) said that Trump supporters would feel comfortable sharing their views, dropping to zero among the small group of Trump-supporting academics in these departments. Left-leaning workplaces outside academia are more tolerant. In Qualtrics 2021, 42% of those working in places that they rate as liberal said that Trump voters would feel comfortable expressing their views. Some 38% of Trump supporters in liberal workplaces also said that they would be willing to express their political beliefs to colleagues. Meanwhile, 57% of Biden voters employed in right-leaning workplaces said that they would feel free to share their views.

Whereas 87% of Trump-voting academics in left-leaning American departments in my CSPI 2021 surveys described their departments as hostile environments for their beliefs, this figure is 48% for Trump voters in left-leaning workplaces outside academia (Qualtrics 2021). Whether the difference has to do with the higher share of far leftists among academics (about 25% in SSH faculties), compared with nonacademics—or because of a generally less tolerant and more discriminatory culture on American campuses—is difficult to discern. Though campus currents are affecting the outside world, Andrew Sullivan is only partially correct when he observes that “we all live on campus now.” [31]

Figure 24 shows the extent to which people working in different sectors think that Trump supporters would be willing to share their views with colleagues. Though most university employees in the Qualtrics 2021 sample are not academics, it is still the case that this is the sector in which the lowest share of employees feel that Trump supporters would share their political beliefs with coworkers. These figures share some affinity with those for Britain, where I similarly found that people were most likely to say that Brexit voters would keep their views to themselves in the university sector. [32] They also correspond to the partisan composition of political donors in America, in which academics lean 9–1 toward the Democrats and teachers 4–1 toward the Democrats, while many blue-collar occupations tilt Republican. [33]

Figure 24

Where a Trump Supporter Would Speak Freely: By Type of Work

One factor to weigh in assessing these results: Trump voters are especially likely to mistrust university academics. Pew notes a substantial change in Republicans’ perceptions of higher education from mainly positive in 2015 (54% positive, 37% negative) to mainly negative in 2017 (36% positive, 58% negative) and 2019 (33% positive, 59% negative). [34] But this conceals important variations, depending on which fields of the university are in focus.

In Qualtrics 2021, for example, 67% of Trump voters say that they trust science, technology, engineering, and math (STEM) professors (Figure 25). While this is 20 points lower than the 89% of Biden voters who answered likewise, it is still substantially higher than Trump voters’ trust in oil executives (39%). Whereas Biden voters are nearly as trusting of social sciences and humanities (SSH) professors (81%) as STEM professors, just 34% of Trump voters trust SSH professors, a massive gap of 47 points between themselves and Biden voters. A similar divide of 45 points separates Trump and Biden voters when it comes to trusting journalists. The strongly leftwing cast of both SSH professors and journalists arguably accounts for these figures, indicating that these facts are increasingly becoming public knowledge, probably via conservative outlets’ coverage of campus cancel culture and their criticism of mainstream media. [35]

Figure 25

Trust in Profession: By Vote (%)

Education has long been linked to higher social trust and trust in institutions more generally. But the relationship is affected by partisan leaning: just 29% of Trump voters with master’s or doctoral degrees (98 in my sample) trust SSH professors, but 75% of them trust STEM professors. For Biden voters with advanced degrees, 86% trust SSH professors and 93% trust STEM professors. This means that there is a 57-point gap between highly educated Biden and Trump voters in their trust of SSH professors, compared with an 18-point gap in trust of STEM professors. For journalists, there is a 55-point trust difference between the two stripes of highly educated respondents.

Democrats’ greater likelihood of attending university does not account for the partisan trust gap; quite the opposite. In fact, there is a larger partisan trust gap for SSH academics and journalists among college-educated voters (Figure 26). The same is not true for oil executives and STEM academics, neither of whom are as strongly associated with left-wing politics. K–12 teachers, another left-leaning group, may increasingly be affected. The survey by Parents Defending Education finds that, among Republicans, 59% have a favorable view of their “local school teachers,” but 31% hold an unfavorable view, rising to 42% among those who “always vote” Republican. Among Democrats, by contrast, the favorable/unfavorable ratio is an overwhelming 86–5. [36] Declining trust in left-wing-dominated professions and institutions appears to be a hallmark of our new era of culture-war politics.

Figure 26

Trust in Social Sciences/Humanities (SSH) Professors and Journalists: By Voter

Fear of Peers

Organizational sanctions or reputational damage are not the only sources of chilling effects. Peer judgments, linked to political discrimination, can be as or more important. The 2020 Cato study asked: “Are you worried about losing your job or missing out on job opportunities if your political opinions became known?” This question arguably taps peer judgments (including those of superiors) pertaining to missing out on job opportunities more than fears of being canceled— i.e., soft, rather than hard, discrimination. A total of 27% of respondents said that they were worried, and using the same question wording, 26% of the Qualtrics 2021 respondents had this concern. Note that this is markedly lower than the 37% of respondents who worried about being fired or reputationally ruined for past or present speech or activity. From this, I surmise that people are more worried about hard, rather than soft, discrimination outside academia.

An important subsidiary finding from the Cato study was that 60% of 55 Trump voters with advanced degrees and 40% of 167 Trump voters with degrees feared professional consequences if their views became known at work. This compared with about 23%–26% for highly educated Clinton voters and Trump voters without college degrees. This suggested that knowledge workers were more politically discriminatory, perhaps reflecting both the more liberal cast of educated workplaces and the stronger hold of progressive taboos in the large organizations where those with advanced degrees are more likely to be found.

The Qualtrics 2021 survey data find lower levels of Republican fear than Cato’s findings. A total of 37% of 96 Trump voters with postgraduate degrees reported being fearful of the consequences if their views became known, not significantly different from the 32% of 106 noncollege Trump voters or the 30% of 165 Biden voters with postgraduate degrees, especially after controls for other demographic characteristics. Whether this reflects the sample, or the timing of the two surveys (the Cato poll appeared during the Trump administration, and just before an election), is difficult to discern.

Despite this finding not being the same across the two surveys, many others were the same (Figure 27). In particular, age was the most important predictor of being worried, with younger people being more concerned. There didn’t appear to be any discontinuities by age group or generation. Next in importance was gender, with males consistently more concerned than females. The share of minorities in a person’s local neighborhood, a proxy for the diversity of one’s workplace, was significantly correlated with greater worry about one’s political beliefs becoming known at work. Importantly, the effects of nonwhite share were concentrated among white Trump supporters in both data sets. For instance, 51% of the 75 white Trump voters living in majority nonwhite areas in the Cato data were worried, compared with 21% of the 229 white Trump voters living in zip codes where the white residents exceeded 80%. In Qualtrics 2021, 44% of 34 white Trump voters in majority nonwhite areas were worried, compared with 21% of 173 white Trump voters living in zip codes that were more than 80% white.

Income, whether at the individual or zip-code level, had no significant effect. Zip codes with a higher share of postgraduate degree holders also made no difference to people’s fear levels. In the Qualtrics survey, Trump supporters and conservatives working in left- or Democratic-leaning workplaces reported a significantly higher level of concern, similar to what we saw earlier with respect to worrying about losing a job due to past or present speech. Some 38% of Trump voters in left-leaning (“liberal”) workplaces were worried, compared with 21% of Trump voters in conservative-leaning workplaces. For Biden supporters, 29% in liberal workplaces worried, versus 22% in conservative workplaces. This suggests that peer-driven pressures to stay silent come more from the left than from the right, regardless of whether one is a Biden or a Trump supporter, even as Trump voters experience the stronger silencing effects.

Figure 27

Predictors of Worrying About Political Beliefs Becoming Known at Work

A majority of Trump voters working in very liberal workplaces feared losing their jobs or reputations because of past or future speech (Figure 20). In terms of their beliefs becoming known at work, the Cato survey found that 40% of Trump voters with degrees, and 60% of those with advanced degrees, worried.

Still, overall, only a minority of Trump voters worry about being canceled (36%), discriminated against, or losing their jobs if their views become known to colleagues (27% for Qualtrics, 30% in Cato). Fifty-eight percent of Trump voters said that they would share their views at work, 24% said that they wouldn’t, and 18% were unsure. While less than the 68% of Biden voters who would freely share their views at work, these Trump numbers diverge greatly from those in academia, where just 10% of Trump voters say that they would be open about their views. [37]

A key reason that the nonacademic world is more friendly to conservative and pro-Trump viewpoints than is the academic world is political segregation in the workplace. Bill Bishop and Robert Cushing were the first to popularize the idea that Americans were becoming increasingly sorted by partisan geography. [38] As college graduates moved to metropolitan areas for knowledge work and the basis of voting shifted from income and region to education and values, urban areas became increasingly Democratic, while rural areas trended Republican. In 1992, 38% of Americans lived in “landslide counties” where one party won by more than 20%. By 2016, more than 60% did. [39]

Often people relocate not for explicitly partisan reasons but because of proxies for partisanship such as density, certain shops and amenities, or the education level of the population. [40] In the Qualtrics 2021 survey, the typical college-educated white Trump voter lives in a zip code that is 75% white and 13% college-educated, and 17% reside in rural zip codes. The same white person who voted for Biden lives in a zip code that is 69% white and 16% college-educated, with 11% in a rural area.

Since people often work relatively close to where they live (at the very least, in the same town or metro area), there is a link between the partisan composition of their neighborhood and that of their workplace. If, in addition, they choose occupations—such as professor or sergeant—or organizations (Google? Chick-fil-A?) that are politically congenial, or if these attract a certain political type, occupational segregation along political lines will be substantial.

Figure 28 finds as much, highlighting that a clear majority—56%—of both Trump and Biden voters work in workplaces dominated by their own political tribe. A somewhat larger share of Trump voters are employed in Democratic-dominated workplaces (23%) than Biden voters in Republican places (17%), but the differences aren’t large.

Figure 28

Average Party Identification at Workplace: By Vote (%)

In addition to the political tint of the workplace, Biden voters are more often found working for larger organizations than Trump voters: 47% of Biden voters in the Qualtrics 2021 survey, compared with 34% of Trump voters, work in organizations of more than 1,000 people. Some 38% of Trump voters but just 26% of Biden voters work in places with fewer than 50 employees. The type of industry also matters: those working in schools and universities rated their workplaces considerably more left-leaning than those working in factories or for the police or military.

A model of the factors that predict a more Republican workplace is presented in Figure 29. By far, the strongest determinant is a person’s own party identification on a five-point scale from strong Democratic to strong Republican. A strong Democrat, all else being equal, tends to work in a weakly Democratic place. A strong Republican, controlling for other factors, tends to work in a weakly Republican organization. Those employed in universities, schools, or government offices rated their workplace more Democratic than those working in “other” sectors, while people employed in factories said that their workplace was relatively Republican, compared with those in “other” sectors. Police/military was signed in the Republican direction but not statistically significant. The share of whites in a person’s residential zip code was correlated with working in a Republican workplace, while people living in an area with a higher degreeholding share were significantly more likely to be employed in a Democratic workplace. Race and gender had no significant effect. Rural-urban share and firm size also did not come out as significant, after controlling for other characteristics.

Figure 29

Predictors of a Republican Workplace

The takeaway from this analysis of organizational partisanship is that there is relatively high political segregation at work. While it is possible that Republican workplaces make a person more Republican, and vice versa, or that partisans have a systematically skewed view of their workplaces, the most likely explanation is that people select for occupations and organizations in a way that results in a “big sort.” Given the pattern outlined previously of liberal and/or Democratic workplaces being more censorious, discriminatory, and hostile for Trump voters, occupational segregation paradoxically serves to insulate them from hostile environments.

This may help explain why Trump voters’ worry about being canceled (35%) or politically discriminated against (27%) is not as high as conservatives working in left-leaning occupations like academia and the media might presume. Of course, the electoral resonance of an issue like cancel culture is not simply a matter of workplace experience. The issue reflects people’s views more broadly of whether they are free to express themselves in other public realms, or whether people of the same political party affiliation whom they know are affected; or whether conservative personalities in the media are affected. Those who are worried about losing their jobs or opportunities because of their politics are not likely to rank culture-war issues more highly than other voters, suggesting that their views are shaped more by their view of how cancel culture affects society than how it affects their own lives.

A number of high-profile exposés of diversity training courses inspired by CRT have come to light, many thanks to the work of Manhattan Institute Senior Fellow Christopher Rufo. The general pattern involves a combination of the following features:

• White people are identified as privileged due to their skin color, with this being viewed as a major—or even paramount—source of inequality in society.
• Racism is defined not as individual prejudice but as a system of hierarchy and oppression, in which whites occupy the role of oppressor and blacks or certain other minorities that of victims.
• “Whiteness” as a socially constructed ideology is held to be distinct from actual white people. Whiteness manifests in language (terms such as “master bedroom” are claimed to have slave connotations), ideas (“colorblindness”), practices (incarceration, deportation of illegal immigrants), and norms (standardized testing, punctuality, reason) are claimed to advantage white people while concealing the operation of a racist system beneath a veneer of nondiscrimination.
• Whites who resist the characterization of America as white supremacist or systemically racist, or who question the importance of white privilege, are accused—implicitly or explicitly— of being racist.
• As an anti-positivist doctrine, CRT’s system of racism is neither measurable nor falsifiable. The doctrine moralizes empirical questions making dissent difficult, and it does not seek to carefully refute alternative explanations: that disparities are the result of factors other than racism.

CRT as developed by certain law school professors decades ago is a claim about the basis of social conflict in Western society, no different from other “high theories,” such as Marxism or postmodernism. But when CRT is applied in the real world, it results in phenomena such as segregated classrooms, sweeping generalizations about whites as a group, and the shaming of actual white people for the past and present misdeeds of others who look like them. Despite the claim of CRT academics that race itself is a “social construct,” this theory as applied uses a highly concrete working definition based on physical appearance. [41]

Some media pundits and politicians profess to claim that CRT is limited to a few anecdotes that have been cherry-picked to whip up a right-wing moral panic. Yet the data in my surveys show that half the respondents have attended diversity training in their place of work. This is clearly not a scattered phenomenon. Given the oversample of college graduates (71%) in Qualtrics 2021, it’s important to qualify this by noting that only 36% of those who lack a college degree have attended diversity training, compared with 56% of people with a degree. [42] Given that all Fortune 500 firms and two-thirds of universities in one major survey offer diversity training, these figures seem broadly correct. [43]

The strongest predictor of whether people attended diversity training was the size of the organization they work for: 64% of those employed in companies with more than 1,000 employees had taken it, compared with 29% in organizations with fewer than 50 workers. Education is likewise important: 56% of employees with a college degree had attended training, compared with 36% with no degree. Trump voters (41%) were less likely than Biden voters (58%) to have attended, though it isn’t clear exactly why this is, because ideology, party identity, and attitudes toward immigration do not predict whether someone has attended diversity training. Those working at universities and schools were more likely to report attending, while younger people had attended more often than older people. Those working in more left-wing workplaces were somewhat more likely to have attended, but the difference is not statistically significant. Figure 30 shows the relative importance of these characteristics in predicting whether someone had attended diversity training.

Figure 30

Predictors of Having Attended Diversity Training Source: Qualtrics

Those who responded that they had attended diversity training were asked: “If you did attend diversity training, did the instructors suggest that discrimination is the main reason for race or gender pay gaps?” Forty-seven percent answered in the affirmative. Next, respondents were asked: “If you did attend diversity training, were any of the following terms used: ‘white privilege,’ ‘patriarchy,’ ‘white supremacy’?” Forty-two percent reported that one of these three critical social justice (CSJ) terms was used.

This means that nearly half the diversity training taking place is deploying empirically dubious claims based on CSJ meta-theory. Since half the respondents attended diversity training, this means that about a quarter of American employees have had direct exposure to CSJ at work.

Drilling into these questions, I find that younger respondents are most likely to have experienced CSJ-themed training. Fully 61% of those 35 and under received training that told them that group pay disparities were mainly due to racism and sexism, compared with 39% of the over-50s. Fiftyfour percent of those under 35 had training that mentioned “white privilege,” “patriarchy,” or “white supremacy,” compared with 33% of those over 50 who heard these terms. Ideology and party identity were not correlated with reporting that these radical ideas were being taught as part of the training. Younger employees are thus highly exposed to this ideology.

Subjects were then asked: “If you had to choose, what is your general view of diversity training?” Responses included “very necessary to limit bias,” “somewhat necessary to limit bias,” “unsure, mixed,” “somewhat a form of political indoctrination,” and “very much a form of political indoctrination.” The respondents were divided into 39% who said that the training was necessary, 31% who weren’t sure, and 30% who said that it was a form of indoctrination. Views on this question were heavily tied to ideology, voting, and, especially, views of political correctness. Some 54% of those opposed to political correctness (as stifling free speech) said that diversity training was indoctrination, whereas only 15% of PC supporters (as protecting minorities) agreed. Whether people had been exposed to CSJ or not in their diversity training made no difference to their view of whether they deemed it necessary or a form of indoctrination.

In Figure 9, we saw that those who had attended diversity training were more likely to support cancel culture. Even after accounting for age, party identity, voting, ideology, views of political correctness, education level, gender, and race, those who attended diversity training in Qualtrics 2021 were significantly more likely to endorse the firing of people like Brandon Eich, James Damore, or Gina Carano (Figure 31). This seems to hold across political viewpoints, suggesting that diversity training embeds cultural socialism, rolling back cultural liberalism. That is, it makes people prioritize the well-being of historically marginalized groups over people’s right to free speech and employment.

Figure 31

Should Gina Carano Have Been Canceled?

Diversity training is not only linked to being in favor of cancel culture—it also predicts a greater fear of being canceled. Forty-three percent of those who have attended diversity training say that they fear losing their jobs or reputations due to something that they have said or done. This compares with just 31% among those who have not attended diversity training. Figure 32 shows the difference across five questions. In all but one, those who have attended diversity training report more fear or loss of freedom than those who did not.

Figure 32

Self-Censorship: By Experience of Diversity Training

This finding is not an artifact of confounding variables because people who attend diversity training are more likely to work in large organizations with more educated or left-wing employees—these characteristics are not driving the findings. After controlling for demographics, party identity, and ideology, diversity training is still significantly associated with an elevated fear of being canceled or discriminated against by peers, as well as with a perceived loss of expressive freedom. Diversity training is significant at the (strong) 1% level for the three loss-of-expressive freedom questions, one of the “worried” questions, and significant on the other worried question at the 5% level. In addition, those who attended diversity training where CSJ concepts like “white privilege,” “white supremacy,” or “patriarchy” were used were somewhat more likely to worry about their jobs and reputations, though the effect was only borderline-significant. These findings add to the growing list of negative outcomes (e.g., poorer intergroup relations) associated with diversity training in its current incarnation. The policy implication of these findings, notes Columbia University sociologist Musa al-Gharbi, is that contemporary diversity training has a chilling effect on employees and should be curtailed until new approaches have been proven to limit negatives and maximize positives are developed. [44] Figure 33 shows a fuller model of fears of being canceled, with diversity training second only to being younger in predicting worries about losing one’s job or reputation because “someone misunderstands something you have said or done, takes it out of context, or posts something from your past online.”

Figure 33

Predictors of Worry About Losing Job and Reputation Due to Cancellation

Interestingly, many of the factors associated with a pro-cancel orientation are also linked with a heightened fear of being canceled. That is, one can argue that there is a consistent belief system in which a person worries about social or career loss but sees this as a price worth paying for upholding a demanding set of moral beliefs that punish offenders. In this reading, those who endorse cancellation are not “snowflakes” but rather true believers in an unforgiving version of left-modernist ideology.

In the 2020 Cato survey (after controlling for voting, ideology, and demographics), those most worried about losing their jobs or opportunities if their beliefs become known are 5–10 points more likely to support firing an executive who donated to the Trump or Biden campaign than those who aren’t at all worried. This statistically significant effect is echoed, using the same questions, in Qualtrics 2021.

To explore this, I wanted to know whether respondents agreed with a statement encapsulating these seemingly opposing positions: “My fear of losing my job or reputation due to something I said or posted online is a justified price to pay to protect historically disadvantaged groups.” Twenty-seven percent of the respondents agreed, with 43% disagreeing. Importantly, a majority— 52%—of respondents 25 and under agreed with this statement, as did 46% of those 26–30 but only 17% of those over 50.

Age is as important as ideology in predicting agreement with this statement, suggesting that young people are most steeped in the pro-cancel belief system. In addition, those more fearful of being the victim of a dismissal attempt, or of losing opportunities at work, are significantly more likely to endorse the idea that such fears are a reasonable price to pay for minority safety and equality. This echoes the 2020 Cato and Qualtrics 2021 survey findings that those most worried about losing their jobs were significantly more likely to support the firing of an executive who makes a political donation. Rather than viewing such individuals as thin-skinned or inconsistent, we should consider them to be adhering to a consistent set of ideas that places the emotional safety and advancement of minority groups above their personal freedom of expression. An individual aged 18–25 who fears cancellation has a nearly 60% chance of agreeing that this is a justifiable price to pay to protect minorities, compared with only 10% of the oldest survey respondents who fear being canceled (Figure 34).

Figure 34

Fear of Losing My Job or Reputation a Justified Price to Pay: By Age

Punishment for speech and political discrimination—what I term hard and soft authoritarianism— are both important when it comes to promoting fear and self-censorship. The impact of political discrimination, a reflection of John Stuart’s Mill’s idea that the “despotism of custom” is one of the most powerful threats to liberty, can be seen in the relationship between people’s willingness to discriminate and their willingness to fire someone for speech. In the Qualtrics 2021 survey, for instance, 32% of those who would discriminate against a Trump supporter for a job endorse the firing of Emmanuel Cafferty, the gas-company employee who was photographed cracking his knuckles in a way that looked like the “OK” sign, compared with 9% of those who would not discriminate against a Trump supporter. Sixty-one percent of those who would discriminate against a Trump supporter for a job supported firing James Damore, compared with 28% of those who would not discriminate against a Trump supporter. Even when controlling for age, ideology, and partisanship, people’s willingness to politically discriminate is a significant predictor of their willingness to support dismissing someone. Political discrimination runs a close second to partisanship and ahead of age in statistical power. I found the same results in my academic surveys (CSPI), with those on the left willing to discriminate against conservatives in hiring, refereeing, or promotion twice as likely to support a hypothetical firing campaign against a dissenter, compared with those on the left who were not willing to discriminate. [45]

There are two major trends in public opinion on cancel-culture questions. First, the balance of public opinion opposes cancel culture by a ratio of approximately 2 to 1. In the Qualtrics 2021 survey, while views of political correctness were fairly evenly divided (41% opposed, 37% in favor), the weight of opinion turns against progressive illiberalism for concrete questions about cases like that of James Damore or Brandon Eich, hypotheticals like banning an immigration restrictionist from Twitter, or preventing a debate on whether trans women are women. In the Parents Defending Education survey, respondents who had heard of cancel culture opposed it by a 63–11 margin. In the Harvard/Harris February 2021 poll, respondents opposed it 64–36.

The second major pattern is significant partisan division on many questions, with strong Democrats and the very liberal (i.e., strong leftists) standing out. This reflects the distinctive profile of the “Progressive Activist” segment of public opinion identified in More in Common’s “Hidden Tribes” report. In general, across the 20 cancel-culture questions in Figure 7, strong Democrats and strong Republicans differ by 30 to 40 points, with Trump and Biden voters between 15 and 35 points apart. In the February 2021 Harvard/Harris poll, the partisan divide on whether cancel culture is a threat to freedom is 32 points, rising to 40 points over whether employers are justified in firing employees over their online content. [46] In what follows, we’ll see that the two patterns hold across a different set of culture-war questions concerning CRT and the teaching of American history.

One set of issues in the culture wars pits cultural egalitarianism against free speech. The Qualtrics 2021 survey asked about cancellation and hypothetical scenarios involving dismissal, punishment, and no-platforming. There is, however, a second suite of tensions that involves the teaching of culturally socialist beliefs in schools and mandatory diversity training. Because this instruction is compulsory and carries penalties for noncompliance, it violates the freedom of conscience of pupils who oppose it.

Where such instruction impugns individuals’ group identities by, for example, assigning collective blame to white people for actions taken by those in the past who looked like them or smearing an entire group with the actions of a racist individual, this can be viewed as a violation of white students’ civil right to equal treatment with respect to their race, sex, or other protected characteristics. So CRT represents another theater of struggle between cultural socialism and cultural liberalism.

Nevertheless, while the conflict about CRT involves this clash of principles, the visceral motivations of those who resist CRT are about much more than liberalism and equal treatment. When the U.S. is referred to as a racist society, this offends those who hold a traditional, rather than a radical, conception of U.S. national identity. While people don’t have a right not to be offended, a perception that the nation—and, by extension, whites, who played a dominant role in U.S. political history—is being unfairly maligned is a source of discontent. The perception that history is being taught in a de-contextualized manner—e.g., ignoring the slavery and ethnic discrimination practiced around the world for centuries, atrocities committed by indigenous people or nonwhite civilizations, or concealing the variation in opinion among whites in the past—can reinforce the perception that there is a disproportionate centering on the flaws of whites compared with their achievements, and an unequal symbolic treatment of their history by comparison with other groups.

Grievances may extend beyond questions of freedom of conscience and equal treatment to encompass concerns about national identity and solidarity. If a term such as racism, one of the gravest social sins of our time, is applied to the nation in the form of a sweeping generalization and even elevated—as CRT does—to a central role in defining the country, this may be judged as disproportionate by many, stoking considerable resentment.

The connection between CRT and the education of schoolchildren, involving vulnerable and impressionable pupils, further contributes to opponents’ passion. Manhattan Institute’s Christopher Rufo, among others, has exposed some of the ways in which CRT-inspired pedagogy has entered school curricula in America. Among the examples that he cites:

• Seattle Public Schools telling teachers that the education system is guilty of “spirit murder” against black children and that white teachers must “bankrupt [their] privilege in acknowledgment of [their] thieved inheritance.”
• San Diego Public Schools accusing white teachers of being colonizers on stolen Native American land and telling them that “you are racist” and that “you are upholding racist ideas, structures, and policies.” They recommend that the teachers undergo “antiracist therapy.”
• A Cupertino, California, elementary school forcing third-graders to deconstruct their racial and sexual identities, and then rank themselves according to their “power and privilege.” They separate the eight-year-old children into oppressors and oppressed.
• The Arizona Department of Education creating an “equity” toolkit claiming that babies show the first signs of racism at three months old and that white children become full racists “strongly biased in favor of whiteness”—by age five. [47]

Rufo has also reported on similar trends in diversity training. How rare are these incidents? In Qualtrics 2021, 42% of respondents who took diversity training said that one or more of the terms “white privilege,” “white supremacy,” or “patriarchy” was used. Forty-seven percent said that the instructors suggested that discrimination is the main reason for income gaps between groups, a claim that is difficult to square with wide intra-racial income differences. Thus, CSJ (encompassing CRT) approaches to diversity training are routine rather than fringe phenomena.

Younger people are considerably more likely to have experienced these radical ideas than older people, but there are no significant differences between liberals and conservatives in their likelihood of reporting exposure to CSJ ideas. In terms of their appraisal of diversity training as a concept, the views expressed in Qualtrics 2021 were divided: 38% see it as necessary, 30% as a form of indoctrination, and 32% are unsure.

It appears that CSJ ideas have not penetrated as deeply into the education system as they have into diversity training. When the Parents Defending Education survey asked, “Is your child attending a school that is teaching social, cultural, and race-based topics in the [critical] ways we’ve discussed?” 12% said that they were, 45% that they were not, and 43% were unsure. However, 59% of parents said that over the past two years, the “emphasis on issues of race, gender, and activism” had increased, and 61% said that schools had become more political. Virtually no one said that these trends had declined.

Opinion on a wide battery of questions around curriculum content from the Parents Defending Education survey [48] reinforces the earlier two trends showing that CSJ approaches to education are generally unpopular and that there is a considerable partisan variation on many questions pertaining to the teaching of culturally socialist beliefs around race and gender (Figure 35). The partisan divide is widest over the approval of Black Lives Matter (81% of Republicans oppose vs. 8% of Democrats) and socialism (90% of Republicans oppose vs. 11% of Democrats). Bipartisanship is greatest on the question of whether schools should “Assign white students the status of ‘privileged’ and assign nonwhite students the status of ‘oppressed,’ ” a policy that only 6% of Democrats and 1% of Republicans endorse. Positive appraisals of Martin Luther King, Jr. and Abraham Lincoln were also relatively unanimous, and positive views of the Founding Fathers were held by most—though not by 23% of those who always vote Democratic. (Note that Figure 35 charts opposition only and that one cannot read a lack of opposition as support, since there is an important unsure category that tended to be used more by Democratic respondents.)

At the abstract level, CRT is very unpopular among Republicans and independents but quite popular among Democrats. A YouGov/Economist survey in January 2021 showed that 64% of people said that they were at least somewhat familiar with CRT, though only 54% of this 64% said that they had a good idea of what it is. [49] Notably, a third of Americans viewed CRT favorably, and opinion is highly divided on partisan lines. Democrats are 85% in support, but fewer than 10% of Republicans and 20% of independents back CRT. Ninety-one percent of Trump voters said that they were “very unfavorable,” while 58% of Biden voters were “very favorable” and a further 24% the more lukewarm “somewhat favorable.”

A Politico/ Morning Consult survey in June 2021 showed a similar pattern. [50] Of those who said that they knew what CRT was (N=1,408), 76% of Republicans said that they were “very unfavorable,” 11% “somewhat unfavorable,” and 13% “favorable.” Democrats were spread across response categories more, with 33% “very favorable,” 42% “somewhat favorable,” and 26% “unfavorable.” This puts the partisan divide at a yawning 73 points in the YouGov/Economist data and 63 points in the Politico/ Morning Consult survey.

Generally speaking, the more a policy involved shaming individual students, the greater the bipartisan opposition to it. However, more abstract versions of CSJ elicited more Democratic support, producing wider partisan division. For instance, rather than assigning students as privileged, the proposal in the Parents Defending Education survey, to “teach that white people are inherently privileged, while black and other people of color are inherently oppressed and victimized,” was endorsed by 43% of Democratic respondents. Sixteen percent of Democrats said that teaching that race is the most important thing about a person was important, with 51% saying that this was at least somewhat important. Twenty-nine percent of Democrats said that it was important that changes based on CRT should be instituted, with 90% saying that this was at least somewhat important. While the five-point scale (extremely important, very important, somewhat important, not that important, not at all important, plus don’t know/unsure) used in these questions, in which the middle category was “somewhat important,” may be skewing the results a bit, these items produced some of the widest partisan divides.

Damning indictments of American history and society also led to partisan division in the Parents Defending Education survey (Figure 35). Strong (“Always”) Republicans and Democrats have major partisan differences in the share opposed to schools teaching that the U.S. is founded on racism and remains racist (97–42), that it is built on stolen land (93–35), and that U.S. history classes should “focus on race and power and promote social justice political issues” (92–19). Large partisan divisions were also present over whether to teach political or social activism, promote equity, or teach about race and gender more. Notice how the dots for strong and weak partisans cluster together, with large differences between Republicans and Democrats, while independents stand in the middle. This is very different from Figure 7 (on cancel culture), where weak Democrats were closer to Republicans while strong Democrats stood out from the rest, rendering partisan divisions smaller.

Figure 35

Opposition to CRT Items: By Usual Vote

These questions often drew strong Republican disapproval while splitting the Democrats. For instance, while 81% of Republicans responded negatively to the idea of teaching about race and gender more by saying that this was unimportant, Democrats divided, with 33% saying that this was important, 29% that it was not, and 38% in the middle. While upward of 90% of oppose teaching that the U.S. is racist or built on stolen land, Democrats split 54–46 against teaching that the U.S. is racist and 57–43 in favor of teaching that it is built on stolen land. Democratic responses were evenly distributed across the four response categories (“somewhat” or “strongly” for or against) while Republicans bunched overwhelmingly in the “strongly against” category. As with the questions on cancel culture in the Qualtrics 2021 survey, overall responses leaned 2 to 1 against radical cultural egalitarianism, with near-unanimous opposition to the kind of CRT examples exposed by Rufo. This would suggest that culture-war issues represent a political opportunity for Republicans to unite their supporters while dividing Democratic voters. As with cancel culture, CRT has become a toxic brand, which suggests that it is a political liability for Democrats. A better strategy might be to devise a term such as “critical history”— conveying the importance of including the bad along with the good in the American past—and make it clear that this approach excludes the divisive beliefs, rhetoric, and practices of CRT.

The data in Figures 5 through 7 show that about two-thirds of Americans oppose cancel culture and prioritize free speech over combating hate speech. While there is a closer 41–37 split over the benefits of political correctness (perhaps interpreted as civility or politeness), this does not extend to concrete cases of dismissal or no-platforming, where opinion, as we have seen, is two-thirds opposed. The same holds for questions around the teaching of critical approaches to race and gender in schools.

The problem for the Democrats is that on a number of questions, the influential strong Democrats—and sometimes Democrats as a whole—support cancel culture. For instance, by a 64–36 margin, strong Democrats in Qualtrics 2021 backed the firing of James Damore while weak Democrats backed it 54–46. When asked, “Do you think this sets a good precedent that people should think twice about expressing views that may be seen as fostering gender stereotypes, or a bad precedent because people should not be afraid of expressing their honest views on these subjects?” Seventy-one percent of strong Democrats and 63% of Biden voters overall said that his firing set a good precedent. A 2017 poll showed closer to 50% of Democratic voters endorsing Damore’s cancellation; but either way, results show that free-speech issues divide Democrats more than Republicans. [51]

The University of Chicago’s long-running General Social Survey (GSS) has consistently shown that, since 1972, liberals (i.e., those on the left) are more supportive of free speech than conservatives. Until 2012, according to FIRE (Foundation for Individual Rights in Education), liberals were also more likely than conservatives to support the right of a racist to speak, and even in 2018, partisan differences were modest. [52] This speaks to an important point: many conservative voters are not primarily motivated by threats to free speech. Instead, perceived attacks on whites and America motivate them more, especially when directed at the family, as is the case in discussions of whether children should be taught to feel guilty for their “white privilege” or the evil deeds of past white Americans. The fact that conservatives are more sensitive to having their traditions disparaged than their expressive liberty curtailed reflects the fact that of the six main moral foundations outlined by Jonathan Haidt, conservatives differ most from liberals on sanctity, in-group loyalty, and respect for tradition. They differ less on the moral foundations of liberty, care/harm, or fairness. [53]

To see how these distinct effects play out, it is important to distinguish between cancel culture, which restricts freedom, and CRT, which targets groups and traditions to which conservatives are attached, such as America or white people. This is so even as the CRT-in-schools debate raises questions of freedom of conscience and equal treatment and thus activates people’s perceptions of fairness and care/harm.

Comparing Figure 7, on partisan divisions over cancel culture, and Figure 35, on partisan divisions over CRT and American history, reveals two things. First, strong Democrats’ support for, or weak opposition to, cancel culture marks them out as distinct. Weak Democrats are more critical of cancel culture and not so different from independents and weak Republicans. Second, when it comes to CRT, we see a different pattern: this time, strong Democrats do not differ much from weak Democrats while the partisan gap between Democrats and Republicans is far greater. For example, a little more than half of Democrats feel that radical approaches to American history, race, and gender should be taught in schools; yet an overwhelming 80%– 97% of Republicans are opposed. Republican opposition to cancel culture is not this strong.

There are significant partisan differences over whether to fire controversial individuals such as James Damore, but, as the final four bars in Figure 36 (Parents Defending Education survey) show, these partisan cancel-culture gaps are generally in the 20–30-point range rather than the 50–60-point range characteristic of the CRT-themed education questions in the first four bars. None of the free-speech questions achieves more than 80% opposition from Trump voters; and in the case of the researcher who finds that diversity reduces solidarity, only 42% of Trump supporters say that they would oppose the person being dismissed from his post. Between 20% and 30% of Trump voters support cancel culture on a binary forced-choice question. On a three-category question with a middle option for indicating uncertainty, nearly 60% of Trump voters say that they are either unsure (46%) or support (12%) the firing of an employee who finds that diversity reduces solidarity in society. In addition, 30% of Trump voters also believe that employers should be able to discriminate against a Trump supporter when hiring for a job.

For comparison, the February 2021 Harvard/Harris poll found that 29% of Republicans support Amazon banning books “based on their political viewpoint,” rising to 33% for books critical of the transgender movement, with Democrats only about 20 points more illiberal, at 55% and 51%, respectively.

One of the few other surveys to have asked about cancel culture and CRT is a March 2021 poll conducted by the American Council of Trustees and Alumni (ACTA) of 800 Illinois respondents. [54] It found that 72% of Republicans oppose “efforts to prevent speakers from expressing opinions that some members of the campus find offensive” and 17% support these efforts. Forty-nine percent of Democrats opposed these efforts, and 25% supported them, showing a partisan gap of 8–23 points on this cancel-culture question. The same poll asked the CRT-themed question of whether schools should teach that “America is founded on slavery and remains systemically racist today” or about “founding principles and documents that established the first free and democratic country in the world.” This time, Republicans went 81–10 for the latter while the Democrats broke 25–59 for the former. This represents a partisan distance of between 49 and 56 points, again showing a wider partisan gap over CRT than over cancel culture. A similar sized CRT divide could be found around teaching social justice advocacy or core skills. Asked whether white students and professors in college should acknowledge that they are inherently racist as part of their diversity training, 88% of Republicans were opposed, with 8% in support. This compared with Democrats at 44% opposed, with 33% in support, a partisan difference of 25–44 points. This paints a similar picture to Qualtrics 2021, with greater partisan division over CRT than free speech, and more bipartisanship the closer we come to separating or shaming children for their skin color. Independents were likewise close to Republicans on free-speech questions and relatively intermediate between the two parties on CRT.

These results comport with the work of Haidt on moral foundations and data from GSS that find support for liberty to be only modestly more important for conservatives, compared with progressives.

Figure 36

Opposition to CRT and Cancel Culture: By Party

A useful test of competing moral foundations of liberty and group loyalty among Trump voters is a question on the Qualtrics 2021 survey that asked whether someone who utters a racial slur against a white person at a meeting should be prosecuted for hate speech. Fifty-two percent of Trump voters answered in the affirmative, only slightly less than the 58% of Biden voters who did so. By contrast, the share of Trump voters who agreed with prosecution for a slur against a black person was 38%, versus 63% for Biden voters. Here the defense of in-group members (whites) among Trump supporters counteracts their free-speech proclivities more than the defense of a group with few Trump voters (blacks). There were only four black Trump voters in the data: they differed from other Trump voters in strongly supporting prosecution for the antiblack slur (75%), suggesting that their group loyalties are likewise an important offset against their support for freedom of expression.

Having said this, it is worth noting that more Trump voters (49%) oppose firing for speech in the case of anti-Israel/pro-Palestinian leftist academic Steven Salaita than Biden voters (38%) do—though this may result from the fact that this question was contained within a battery that mainly focused on the dismissal of conservative and libertarian dissenters. Trump voters also were significantly more likely than Biden voters, as noted above, to report feeling restricted in what they could say and in opposing political correctness. It appears that on identity issues, Biden voters are more illiberal on free-speech questions, but many Trump voters also support restrictions when they feel that their in-group is threatened by speech.

The role of in-group loyalty is revealed clearly with a question asked early in the Qualtrics 2021 survey after respondents identified their racial identity: “How important is your racial background to your sense of who you are?” The choices were “extremely important,” “very important,” “moderately important,” “slightly important,” and “not at all important.” About a third of whites said that their white identity was very or extremely important, with 14% replying “extremely important.” This is roughly in line with previous research, and answers to this question also correlate highly with attachment to ethnic background (i.e., as Irish or Italian) and to being American. [55]

What is curious about the responses to questions about white identity, however, is that they predict both support for cancel culture and hostility to CRT-themed diversity training. That is, what we might dub “high” white identifiers tend to support bans that protect identities. In Figure 37, 76% of whites who say that their racial identity is extremely important to them want to ban “forms of mandatory diversity training that tell white employees that they are upholding white supremacy and structural racism if they refuse to acknowledge their white privilege,” while only 47% of those who say that being white is not important at all agree (“low” white identifiers). However, high-identifying whites also support firing politically incorrect Professors Negy and Adamo at much higher rates than low-identifying whites, and this is true across the entire index of cancel-culture questions. Unsurprisingly, they are more likely to want to prevent a debate over the question of whether all whites are racist than low-identifying whites. These effects are generally significant at the p<0.01 level, with controls for age, education, gender, ideology, and party identification. This demonstrates that hostility to cancel culture and CRT largely stems from different value orientations, with whites who strongly identify with being ethnic (e.g., Irish, Italian), white, or American attuned to the moral foundations of group loyalty and respect for tradition. These, in turn, underpin hostility toward CRT but tend to reduce opposition to cancel culture.

Figure 37

Importance of White Identity and Attitudes Toward Cancel Culture and CRT

Academics appear to be more pro–free speech than nonacademics. Thus 42% of Trump voters actively oppose the dismissal of a dissenting diversity-solidarity researcher in the Qualtrics 2021 sample of nonacademics, but fully 82% of conservative academics in my CSPI 2021 report do (a similar share was true of the few Trump-voting academics). Among academics, even 73% of moderates and 48% of moderate leftists expressed opposition to cancellation. Having said this, the share of Trump voters who support firing the academic in question is similar between Trump-supporting academics (9%) and nonacademics (11%)—so it appears that the difference is mainly that more Trump-supporting nonacademics are ticking the “don’t know” or “neither” options than Trump-supporting academics.

It may be that research is more central to the job of academics; hence, they experience the threat of being canceled for research more keenly than conservatives outside academia, where research is not a core aspect of people’s jobs. It could also be that university faculties are more repressive work environments due to their high share of far-left staff and students, as noted earlier with the question of whether Trump supporters feel able to express their beliefs: my CSPI surveys of academics show that only 10% of Trump-supporting social sciences and humanities academics feel free to do so. Meanwhile, outside academia—even in left-leaning workplaces—a majority of Trump voters say that they would feel comfortable sharing their views at work.

The foregoing suggests that the CRT debate keys into conservatives’ moral foundations of in-group loyalty and respect for tradition, as well as their concern for harm avoidance, liberty, and fairness. While cancel culture is a concern for some Trump voters, most work in conservative workplaces or live in Republican zip codes. The share who worry about being fired for past or future speech is only about a third, no different from among Biden voters. While many Trump voters say that they feel less free today than five years ago to express themselves, so do many Biden voters. Thus, CRT arguably has more potential to mobilize the Republican base than the politics of free speech and cancel culture. Democrats, on the other hand, might be wary of permitting themselves to be portrayed as defenders of CRT.

Even if public opinion appears to present the right with a political opportunity, many people still have not heard of CRT or cancel culture or may not understand what these terms mean. Whether these concepts are a low-level concern among voters or simply adopted as the latest partisan buzzwords, they are unlikely to shift crucial swing voters.

To assess the importance of this issue, I begin by noting that key culture-war terms are gaining increasing media attention (Figure 38). Searches on Google for two key culture-war terms—“cancel culture” and “woke”—have taken off in the U.S. since mid-2019. This also describes a concrete phenomenon: the National Association of Scholars (NAS) recorded a fivefold increase in attempts to fire academics between 2019 and 2020, many of which have made the news. [56] This suite of issues is clearly rising in the public consciousness.

Figure 38

Google Searches for “Cancel Culture” and “Woke,” 2016–21 

But there is also a sense in which wokeness may be a discourse with little purchase on those who are not news junkies, highly educated, young, or online a great deal. Many have noticed that those who spend a lot of time posting political content on social media may not be representative of the population at large. Even so, the concerns of the chattering classes often percolate down the social scale, up the age range, and out to the peripheries. How important are questions of progressive speech restrictions for voters?

A majority have a negative view of cancel culture when it is explained to them. Thus in July 2020, YouGov asked a national sample, “ ‘Cancel culture’ refers to a form of boycott in which an individual (usually a celebrity) who has said something that offends some people is called out and shunned. How big a problem do you think ‘cancel culture’ is in the US today?” [57] Fifty-six percent of those polled called it a “big” or “somewhat big” problem, and just 13% said that it was “not a problem.” While 78% of prospective Trump voters agreed, so did 58% of independents, 47% of Democrats, and 41% of prospective Biden voters. Here again, we see a relatively united Republican coalition alongside a divided Democratic one.

The May 2021 survey by Parents Defending Education found that 61% of those polled said that they had a negative impression of cancel culture, compared with 9% who had a positive view. The figures for unfavorable versus favorable, net of uncertain responses, broke 80–2 among Republicans and 42–14 for Democrats, though the “very unfavorable” reply was given by 76% of Republicans and just 22% of Democrats. These results resemble those from the Qualtrics 2021 that found a majority opposing aspects of cancel culture, with the issue largely uniting Republicans while dividing Democrats. [58]

Even so, before it was explained to them, just 41% of the July 2020 YouGov sample said that they knew what the phrase “cancel culture” means. The May 2021 Parents Defending Education survey still found that nearly 30% had not heard the term. Even among those who know the term, it is not clear how important it is to them. Of course, some people may not have heard of cancel culture or wokeness but may be aware of political correctness, an older term that has a great degree of overlap with these newer concepts. Others may have heard of cancel culture but accord it a low priority compared with what they care about. What is largely missing in current polling research is a measure of what political scientists call the “salience” of this issue. I wanted to define cancel culture as broadly as possible to respondents in order to connect with the many different ways in which they might understand it.

One established tool that political scientists use to gauge salience is to ask people what the most important issue is facing the country. This has been asked in America since the early 1930s. In the Qualtrics 2021 survey, people were asked, “Which, for you, is the MOST important problem facing the country?” The options (with the percentage that chose it) included:

1. White nationalist terrorism or insurgency, far right misinformation (6%)
2. Health care, share of people without health insurance, cost of health insurance (12%)
3. Environment, global warming, man-made climate change (6%)
4. Political correctness, free speech, cancel culture, wokeness, people falsely accused of racism and sexism (10%)
5. Moral values, turning away from religion, family values (8%)
6. Foreign policy, China, Iran, Russia, overseas threats (3%)
7. Immigration, border issues, amnesty, birthright citizenship (10%)
8. Equality for historically disadvantaged race, gender, and sexuality groups (4%)
9. Covid-19 and the economy (41%)

While Covid and economic issues are often parsed out, the dominance of discussions of Covid and its economic and health impacts at the time of the survey meant that I sought to encapsulate it in one option. But after this issue, only health care, at 12%, scored as high in salience as culturewar issues such as political correctness and cancel culture (10.2%), with immigration following closely at 9.9%.

I repeated the question twice more, asking people for their second and third most important issues. The results again showed that culture-war issues are an important factor for voters, scoring just below immigration but above the environment and climate change (Figure 39). About 10% of nonvoters and independents ranked culture-war questions their top priority, and a third ranked it in their top three.

Figure 39

“Which, for You, Is the MOST Important Problem Facing the Country?” Share (%) Citing Issue as Top 3

Looking at these issues by the 2020 vote, it’s apparent that, as expected, culture-war questions are far more important among Republicans than Democrats (Figure 40). Among Republicans, nearly half ranked these issues in the top three, second only to Covid/economy and immigration, and ahead of moral values/religion. Those opposed to political correctness on the trade-off question (“stifles free speech” vs. “protects against discrimination”) were also much more likely to prioritize culture-war issues in their top three. Together, voting and views on political correctness explain a lot. A Trump voter most opposed to PC on a seven-point scale (strongly favor, favor, somewhat favor, neither favor nor against, somewhat against, against, strongly against) had a 60% chance of mentioning culture-war issues as a top-three problem, compared with just 10% for a Biden voter most in favor of PC on the seven-point scale.

Even so, nearly 20% of Biden voters mentioned culture-war issues in their top three. While only 15% of white Biden voters mentioned culture-war problems as a leading issue, 25% of minority Biden voters did. Nonwhite and young Biden voters registered significantly more concern about culture-war issues than other Biden voters, though the effects were not large. There were no significant differences by education or gender among Biden voters, however.

Figure 40

“Which, for You, Is the MOST Important Problem Facing the Country?”: By 2020 Vote

On the one hand, a random choice of issues would result in an average score of 33 for each issue. But given the overwhelming prominence of Covid/Economy in the news, the score for the other eight issues should be lower than that, perhaps in the upper 20s. With this in mind, culture-war questions can be said to be an above-average priority among Republican voters, just below immigration but more important than religion and family values.

Crucially, for electoral significance, culture-war issues are a moderate to slightly above-average concern for independents and nonvoters. Those who gave their party identity as “neither” (a measure of the electorally critical independents) ranked culture-war issues fourth of nine issues, with 33% listing it as one of their top-three concerns. It fell just below immigration, at 34%, but above moral values (30%), the environment (29%), and foreign policy (26%). Meanwhile, culture-war issues are a lower-order issue for Democrats, though minority Democrats rank the issues higher than do white Democrats.

For comparison, the Harvard/Harris poll of 1,872 respondents in late April 2021 asked respondents to select three of the “most important issues facing the country today” from a list of 24. Eleven percent chose “political correctness / cancel culture,” placing it precisely in the middle of the issues polled, similar to taxes, civil disorder, and policing. A total of 14% of Republicans named it a leading issue, compared with 7% of Democrats and 11% of independents. Seventeen percent of conservatives named it an issue, compared with 6% of liberals. [59]

The upshot of these findings is that culture-war questions are not just window dressing—a matter of adopting a party cue without passion. Rather, they are an important independent motivation for a sizable chunk of voters, with some electoral potential among swing voters. Should Covid and economic concerns begin to fade, it is reasonable to surmise that cancel culture and CRT may rise on voters’ political agendas. This is the pattern with immigration, which increased in salience in America and Europe as the 2007–08 economic crisis faded in importance, powering the Brexit referendum in England and the rise of Trump, as well as the post-2014 surge of the populist right in Europe. [60]

Bans on teaching so-called divisive concepts have either been adopted or are under consideration in a growing number of states. [61] How strong is support for such measures? In Qualtrics 2021, the share who agreed that “the government should ban forms of mandatory diversity training that tell white employees that they are upholding white supremacy and structural racism if they refuse to acknowledge their white privilege” is 53%, with 65% of Trump voters and 44% of Biden voters supporting a ban. Twenty-six percent are opposed, including 33% of Biden voters. These numbers reflect polling by The Economist that showed a majority of Americans opposed to CRT but run counter to an Iowa poll in June 2021 that found Iowans opposed the state’s CRT ban by a 56–34 margin, with 11% unsure. [62]

To better understand how bans affect support for CRT, I conducted an experiment of independent voters using the Prolific Academic online survey platform (see Appendix). The independent voters in the sample lean Democratic, with 42% of the independents having voted for Biden and 9% for Trump in 2020.

Before answering questions, people were split into three groups. One group, the control group, read nothing. The second group read the following passage:

"A middle school in Springfield, Missouri, forced teachers to locate themselves on an “oppression matrix,” claiming that white heterosexual Protestant males are inherently oppressors and must atone for their “covert white supremacy.” This kind of approach has been labeled Critical Race Theory." [63]

The third group read the passage above, but with an extra sentence:

"Due to incidents such as these, more and more Republican states are banning instruction based on CRT. Democrats oppose the bans, arguing that the CRT issue is a right-wing moral panic."

When asked if they support CRT itself, independent voters who read one of the two passages featuring the Missouri case shifted significantly in the direction of opposing CRT (Figure 41). That is, when asked about CRT, those independents who read nothing supported it by a 32%–19% margin, with 49% unsure. The groups that read about the Missouri case, either with or without mention of the Republican bans, flipped to opposing CRT by a margin of 39%–17%. The effect is significant at the highest (p<.001) level and holds with a wide range of control variables. Most of the shift involved people changing their views about CRT rather than ceasing to be unsure. This indicates that partisans think they know what CRT is but hold different ideas in their heads of what CRT means, with many Democratic-leaning independents viewing it as merely adding a focus on racial inequality to hortatory views of American history and society. When prompted by a concrete example of a CRT-linked pedagogical practice, attitudes toward it harden.

Figure 41

Support for CRT After Reading an Example: Independents

I get a similar result asking about support for CRT bans (rather than CRT itself), with Figure 42 indicating that those who did not read about the Missouri case opposed CRT bans by a 37%–19% margin, with 45% unsure. After reading about the Missouri case, with or without mention of the Republican bans, opinion shifts to a third in favor of a ban, a third opposed, and a third unsure. The effect is statistically significant in the presence of a wide range of control variables. Most of those who shift move from the unsure category toward supporting a ban, rather than changing their minds from opposition to support.

Figure 42

Support for a Ban on CRT After Reading an Example: Independents

Reading about a concrete case of CRT shifts opinion among this sample of mainly left-leaning independents decidedly against CRT and in favor of a ban. However, those reading about the Missouri case, with or without mention of the Republican ban, are not more likely to say that they will vote Republican in 2024. Among 2020 Biden voters, those who read the passage were marginally less likely to vote Democratic in 2024, but the effect was not statistically significant.

These results suggest that Republican policymakers will try to bring CRT to life through concrete examples as much as possible and that this strategy is likely to increase support for CRT bans. However, it is unclear whether the politics of CRT will be able to shift the needle of America’s finely tuned partisan landscape much. This survey suggests only a modest effect, but one passage of text is not the same as a concerted campaign. Glenn Youngkin’s success in repeatedly attacking Terry McAuliffe on the education issue suggests that such campaigns have potential. Regardless, if the Democrats are unable to distance themselves from the actions of cultural socialists, they are likely to divide their voters and may sustain electoral damage. Democratic Senator Joe Manchin’s support for Republican Senator Tom Cotton’s anti-CRT amendment is a useful example of the kind of action that centrist Democrats could take to mitigate potential downsides on this issue.

Moving from politics to policy solutions, I sought to measure the public’s policy preferences regarding policy and legal measures to contain organizations’ ability to fire or punish staff or customers.

To begin with, it is worth recalling, in somewhat more detail than Figure 7, some findings on popular support for state-sanctioned censorship of individual citizens’ speech. These are two-category forced-choice questions in Qualtrics 2021, broken down by a person’s 2020 vote. For the first two statements on slurs, the sample was split, so people did not answer both questions, as this could bias the results by making them answer the two questions similarly. Beginning with the question eliciting the strongest support for hate-speech prosecution and proceeding to the question drawing the least, the results are:

• “If someone uses a racial slur against a white person in a meeting, should he or she be prosecuted for hate speech?” (Yes: Trump 52%, Biden 58%, overall 55%)
• “If someone uses a racial slur against a black person in a meeting, should he or she be prosecuted for hate speech?” (Yes: Trump 38%, Biden 63%, overall 51%)
• “If someone makes fun of a person with a skin condition in a meeting, should he or she be prosecuted for expressing hate?” (Yes: Trump 33%, Biden 53%, overall 43%)
• “Should a public debate over whether all white people are racist be permitted, or should it be prevented because it offends white people?” (Prevent: Trump 35%, Biden 40%, overall 37%)
• “Should a public debate over whether trans women are women be permitted, or should it be prevented because it offends trans women?” (Prevent: Trump 26%, Biden 40%, overall 33%)
• “Is it legitimate for a cable firm to refuse to connect a known white nationalist to the Internet?” (Yes: Trump 20%, Biden 48%, overall 34%)
• “If banning people with anti-immigration views reduces opposition to immigration, should Twitter and Facebook ban them?” (Yes: Trump 14%, Biden 42%, overall 28%)
• “If banning people with far-left views reduces support for defunding the police, should Twitter and Facebook ban them?” (Yes: Trump 21%, Biden 29%, overall 25%)

Between a quarter and a half of respondents support prosecution—which of course goes beyond current laws—for various degrees of real or perceived hate speech, while between a half and two-thirds are opposed. Biden voters and younger voters are more likely to support protecting against hate speech rather than protecting free speech. However, 14%–52% of Trump voters, depending on the question, also support speech restrictions, so these questions do not divide people along partisan lines as cleanly as those pertaining to Black Lives Matter, CRT, or whether the U.S. is a racist country. Again, we see that debates that implicate the moral foundation of liberty do not lead to quite as vigorous a response among Trump voters as those linked to moral foundations of group loyalty and respect for tradition.

I have argued elsewhere that government intervention into institutions is warranted when institutions punish employees or customers for legal speech or engage in political discrimination that is not germane to their organization’s purpose. [64] To gauge public opinion on these issues, Qualtrics 2021 posed a series of questions, the answers to which are sorted below by the degree to which the public wants limitations to be placed on organizations to protect individuals:

• “Should the law prevent a firm from being able to discriminate against a Trump supporter in hiring?” (Yes: Trump 69%, Biden 51%, overall 59%)
• “If someone expressed legal but racist views online or at a protest, do you think it should be legal or illegal for their employer to fire them from their job for their views?” (Illegal: Trump 71%, Biden 45%, overall 58%)
• “The government should ban forms of mandatory diversity training that tell white employees that they are upholding white supremacy and structural racism if they refuse to acknowledge their white privilege.” (Yes: Trump 65%, Biden 44%, overall 53%; No: Trump 17%, Biden 33%, overall 26%).
• “The government should step in to overrule tech firms that ban or suspend users for making legal but controversial statements on social media about race, gender, or sexuality.” (Yes: Trump 51%, Biden 37%, overall 43%; No: Trump 30%, Biden 37%, overall 33%)
• “The government should step in to overrule public universities that punish university professors for making legal but controversial statements on social media about race, gender or sexuality.” (Yes: Trump 50%, Biden 39%, overall 44%; No: Trump 28%, Biden 35%, overall 31%)
• “The government should step in to overrule private organizations that punish employees for making legal but controversial statements on social media about race, gender, or sexuality.” (Yes: Trump 40%, Biden 38%, overall 38%; No: Trump 35%, Biden 39%, overall 36%)
• “Should the law prevent a firm from being able to discriminate against an Alexandria Ocasio- Cortez supporter in hiring?” (Yes: Trump 36%, Biden 54%, overall 46%)

Public support is generally in the majority when it comes to protecting employees against being fired for speech, or discriminated against because of their political views, and—excluding uncertain responses—for regulating organizations to prevent them from punishing or dismissing those who make controversial but legal statements pertaining to race, gender, or sexuality.

For comparison, the April 2021 Harvard/Harris poll of 1,872 individuals showed that 57% of the public think that the government should regulate tech firms’ ability to no-platform their users, with 43% opposed. By about a 60–40 margin, people opposed Amazon being able to ban books from its site because of a book’s political viewpoint. Yet when it came to firms firing employees for speech, just 53% said that they had “gone too far,” while 47% said that the dismissals, to date, had been “largely justified."

The partisan gap in the April 2021 Harvard/Harris poll ranged 18–40 points on these questions, with Republicans supporting the regulation of organizations more than Democrats. An insight into Republican support for companies to do as they wish came with a question asking whether laws regulating firms might be used to “silence voices of dissent, minorities, or religious groups.” This time, 68% of Republicans worried, compared with 56% of Democrats. Concern for the freedom of association of religious bakers or adoption agencies seems to be acting as a check on Republican support for laws protecting against cancel culture and political discrimination. Thus, Republicans are impelled toward regulation by their opposition to cancel culture, but this does not result in full-scale support or a massive partisan divide (as with CRT) because of their inherent suspicion of government. Democrats’ greater trust in government and moderate opposition to cancel culture also narrows the partisan divide.

Moving from cancel culture to CRT, a majority of the public in the Qualtrics 2021 survey supported a ban on mandatory CRT-based diversity training that tells white employees that they are racist if they do not acknowledge their privilege. Biden voters are less in favor of government restrictions on organizations than Trump voters, though there is, once again, a baseline of 30% of Trump voters who oppose the government regulating organizations’ ability to censor or discriminate and close to 20% who oppose government bans on mandatory CRT-themed diversity training.

Combining the three highly correlated questions on whether the government should overrule public universities, tech firms, and companies that fire or ban people into one measure, I ran a statistical model assessing what predicts support for governments overruling institutions that cancel people for speech. 

Figure 43 shows similarities with my previous analyses of support for cancel culture. Conservatives, for instance, are more likely to both oppose cancel culture and support government action to prevent institutions from engaging in it. Yet the differences are just as striking. Young people are considerably more in favor of firing those (such as James Damore and Brandon Eich) who contravene progressive shibboleths but are simultaneously more likely than older respondents to endorse government action to prevent institutions from firing or banning people. While those who worry most about losing their job for speech also tend to endorse cancel culture more than those who are not worried, those worried about being canceled are significantly more likely to back government regulations that protect them against being fired or banned. Incredibly, those who say that the risk of being canceled for online speech is a justified price to pay to protect minorities are also significantly more likely than others to endorse government limits over the ability of public universities and firms to ban or fire people for speech.

General opposition to cancel culture (as in the Damore firing), as well as feeling more constrained about what one can say compared with five years ago (not shown), predicts support for government restrictions on organizations’ ability to cancel. Those working in more left-wing workplaces tend to support government limits on institutional cancel culture more than respondents working in conservative workplaces. This suggests that the more censorious environment in left-wing workplaces inclines people to support regulatory limits on such organizations’ capacity to discipline staff for speech.

Figure 43

Support Government Overruling Organizations That Cancel Users or Fire Employees for Speech

It appears that there is a cluster of young voters who worry about being canceled but accept the need for cancel culture to protect minorities—yet are simultaneously more likely than others to support government protections against job loss or being banned for speech.

One way to understand this relationship is by examining Biden voters who worry about losing their jobs or reputations due to something that they have posted in the past or may post in the future. Figure 44 shows that Biden voters most worried about losing their job show an elevated level of support for government restrictions on tech firms’ right to ban users for “legal but controversial” speech on identity issues, with 81% of 42 Biden voters who worry about being fired supporting government limits on firms, compared with only 28% of 287 Biden voters who are not concerned about losing their jobs or reputations backing government regulation.

Figure 44

Should Government Regulate Tech Firms’ Ability to Ban or Suspend Users for Legal/Controversial Speech?

One lesson of the preceding discussion is that there is a group of older conservative voters who oppose government regulation, even if the aim is to protect conservative speech. This arguably reflects a libertarian antigovernment conservatism that sought to defend companies from economic regulation and civil rights lawsuits, or supports the right of religious organizations or individuals to discriminate in line with their beliefs. Echoes of this libertarianism can be seen in the controversy over whether the Christian owner of a bakery has the right to refuse to bake a wedding cake for gay customers, which many conservatives support. [65] This outlook prioritizes the constitutional right of an individual to free association and free exercise of religion over individuals’ right to equal treatment.

A further test of this is whether conservatives would back antidiscrimination and measures that would benefit political conservatives. We saw that about 70% of Trump voters believe that the law should prevent a firm from discriminating against a Trump voter for a job, though 30% opposed such a law. Continuing this line of questioning, I sought to gauge support for a proposal that I mooted for a mandatory equivalence in public institutions like universities between race/ gender and political diversity. [66] This time, the question reads: “Professors and journalists have shifted from being slightly liberal-leaning in the 1960s to being overwhelmingly liberal today. At the same time, they have become more representative in race and gender terms. How should these professions respond?” People were asked what institutions should emphasize more—race and gender, or political forms of “discrimination and diversity.”

Only 12% of respondents endorse the status quo—taking action on race and gender diversity but making no effort to improve political diversity (Figure 45). Twenty-one percent would make some effort on political diversity and discrimination but less than on race and gender. However, 48% of respondents want to see as much or more effort put into improving political forms of diversity as is currently made for race and gender. Some 20% want no action on either form of diversity.

Partisan differences were generally not large, suggesting bipartisan support for measures to track and improve political diversity in academia and journalism. The one exception was the response category for taking no action on any form of discrimination or diversity, which was backed by nearly 30% of Trump voters but only 12% of Biden voters. The opposition to any institutional action is especially high among Trump voters over 50, 34% of whom wanted no antidiscrimination or diversity measures of any kind. By contrast, just 12% of Trump voters 35 and under selected this option. One reason may be that older libertarian-leaning Trump voters over 50, focused on post-1980s battles for the rights of firms, oppose government intervention in society. Another is that fairness—as compared with loyalty or respect for tradition—is not a distinctly conservative moral foundation. Together, this results in no partisan difference in support for policies designed to address political discrimination and diversity in elite organizations.

These results also indicate that a broad coalition wants to dispense with the current singleminded focus of diversity, equity, and inclusion (DEI) on race and gender to pivot toward what I term “equivalent action” on improving political diversity. Equivalent action would index political/ideological diversity and race/gender diversity in such a way that institutions can opt for no diversity policies at all or pursue diversity policies—but they would no longer be able to prioritize race and gender over political diversity. Interestingly, the ACTA Illinois survey found that 69% of the public, including 64% of Democrats, agreed that “the University of Illinois should do its best to promote a balance of liberal and conservative faculty.” Only 16% of respondents were opposed.

Figure 45

How Should Professions Approach Political and Race/Gender Diversity? (%)

This report began with a discussion of cultural socialism and cultural liberalism. These two constellations of beliefs are coming into increasing conflict, with cancel culture and the teaching of CRT in schools as key battlegrounds. Strong Democrats are much more likely to support the cultural socialist side, while Republicans overwhelmingly favor cultural liberalism. Younger people, regardless of their partisanship or ideology, are more culturally socialist and less culturally liberal than older people. In fact, young people are simultaneously more in favor of punishment for speech and more worried about being punished for speech than older Americans. Yet they appear to accept the bargain as part of a package of cultural socialist beliefs: a majority of those under 25 say that their fear of losing their jobs or reputations for something they say or post online is an acceptable price to pay in order to protect disadvantaged groups, while only 17% of over-50s agree. This age profile indicates that these debates are likely to intensify in the years ahead as millennials enter the workforce and baby boomers depart.

One area of tension between cultural socialism and liberalism is over whether speech can be restricted, or speakers punished, if words can be considered psychologically harmful to minorities. Public opinion, by a 2–1 margin, tends to oppose organizations firing employees or banning users from social media platforms for legal speech, demonstrating that cultural liberalism is the majority position. But the issue divides Democrats fairly evenly, while 70%–80% of Republicans are opposed. A majority of strong Democrats favor cultural socialism while Republicans, most independents, and moderate Democrats favor cultural liberalism. At the abstract level, most people report that political correctness has gone too far, but when asked to make trade-offs between free speech and speech limits that protect minorities, a majority of Democrats and those under 30 come down on the side of political correctness.

Firing or punishing dissenters is a direct restriction on liberty. However, organizations can create chilling effects when their staff discriminates against those who hold minority political views, thereby incentivizing them not to express themselves. Political discrimination is an important problem that cuts both ways, with over a third of Democrats unwilling to hire Trump supporters for a job and only 40% comfortable sitting with them at lunch. Twenty-seven percent of Trump voters would fire a business executive who donated to Biden’s campaign, and a similar share of Biden voters would do so for an executive who donated to Trump. Among Biden and Trump voters, only 60% are comfortable having lunch with a Sanders supporter and, all else being equal, about half would choose a job applicant who supports their own party rather than remain neutral.

The greater support for cancellation among Democrats, especially strong Democrats, makes left-leaning workplaces more intolerant of speech, even for Democratic employees. Thirtyseven percent of the Qualtrics 2021 respondents worried about losing their jobs or reputations for speech, and a quarter worried for their careers if their political views became known to their workmates. In both cases, concern was significantly higher among Republicans and Democrats in the most left-wing workplaces than in conservative workplaces. Younger people were considerably more worried, as were whites in more diverse zip codes. Trump voters in very left-wing workplaces also worried more than their counterparts elsewhere, and more than Biden voters in conservative workplaces. Indeed, only 29% of Trump supporters in Democratic workplaces said that a Trump supporter would express their views, compared with 57% of Biden supporters in Republican workplaces who said that a Biden supporter would feel free to do so.

Diversity training appears to heighten chilling effects. About half the employees in the Qualtrics 2021 survey have taken diversity training, and in half these instances, respondents say that they have been exposed to CSJ ideas such as “white privilege,” “patriarchy,” and “white supremacy.” Younger and college-educated people, as well as those working in large organizations, have had significantly greater exposure to diversity training. Diversity training in CSJ ideas affects about a fifth of the workforce, rising to over a quarter of young white-collar workers.

Those who say that they have attended diversity training are more culturally socialist and less culturally liberal, regardless of their ideology or partisanship. Workers who have experienced diversity training are also more fearful of losing their jobs due to something that they have said, or might say, online. Some 43% of those who have taken diversity training are fearful, compared with 31% of those who have not.

In any given organization, only about 10% or 20% of people support firing employees who publish controversial research that greater ethnic diversity impedes social solidarity. Those working outside universities are actually more censorious than academics are. A middling group of 40%–50% support at least some cultural socialism or are unsure of where they stand, even as their cultural liberalism cross-pressures them toward being reluctant to endorse the firing of a fellow employee. Even for the 40% who steadfastly oppose dismissal for speech, many would shrink from saying so aloud.

A similar structure of opinion exists when it comes to mandatory diversity statements or racial and gender hiring quotas: there is 43%–47% support for these cultural socialist measures, with proponents also able to multiply their power because those who oppose these mandatory statements or quotas are afraid of violating social taboos against racism or sexism. This suggests that cancel culture is primarily a battle of competing ideas, pitting cultural socialists against cultural liberals. In the survey, most of those opposed to these mandates said that they would resist them, but 40% say that they would do so only anonymously because of the power of taboos. In reality, the share of those falsifying their true preferences may be considerably higher, perhaps encompassing a silent majority of 50%–60%. Nevertheless, the 40%–50% support for cultural socialist measures indicates that the principal reason such measures are being adopted in many large organizations is that there exists a considerable body of true believers in these ideas and policies. There is no silent majority too scared to speak their minds.

Despite nearly 40% of people being afraid that they might be canceled at work, personal fear does not appear to be driving voting decisions. Nevertheless, issues involving political correctness, wokeness, and cancel culture are now an important priority for voters, especially Republicans and independents, where these concerns outrank religion/moral values and climate change. Nearly 50% of Republicans surveyed rate cancel-culture questions a top-three concern, as do 32% of independents and 15% of Democrats, including nearly 25% of nonwhite Democrats.

Cancel culture can no longer be dismissed as an elite or media obsession. However, those who are worried about their jobs and reputations are not more likely to prioritize cancel-culture themes than those who are not personally fearful. Rather, these issues matter most for people who oppose political correctness or speech restrictions in general, or for the way it affects others.

In addition, workplace segregation between Republicans and Democrats—wherein almost 60% work in places dominated by members of their own party, and only 20% work in organizations dominated by the other party—helps insulate many from illiberalism. The kind of ideological homogeneity found in universities, for instance, appears to be relatively unusual, with nearly 60% of Trump voters outside academia saying that they would express their beliefs at work, compared with just 10% among Trump-voting academics. While campus trends are influencing larger, more left-leaning, organizations off-campus, many Trump voters work in smaller firms or in politically congenial environments where they feel free to share their beliefs.

This relative security combines with Republicans’ distinct moral foundations of group loyalty and respect for tradition to make CRT a stronger partisan issue for them than cancel culture. Democrats, meanwhile, are somewhat more in favor of criticizing American history and recognizing white privilege than firing a controversial dissenter in an organization, so they differ more from Republicans on questions around CRT. Bipartisan agreement on the virtues of Abraham Lincoln and Martin Luther King, Jr. and against assigning children as privileged or oppressed in class based on skin color suggests that it might be politically toxic for Democratic politicians to be associated with these practices. However, once we move away from directly shaming children for their skin color, there are wide partisan disparities on whether to teach children to be political activists and whether the country had a shameful beginning.

Trump voters are 20–30 points more opposed to cancel culture than Biden voters but are 50–60 points more opposed to children being taught that the U.S. was founded on racism and stolen land. Another way of looking at this: 70%–80% of Trump voters oppose cancel culture but 90%–100% oppose teaching children that the U.S. was founded on racism and theft. The range of issues around CRT and CSJ is, therefore, more explosive for Republican voters: it’s more likely to mobilize them than the struggle for free speech and dissent in organizations. As with cancel culture, CRT and CSJ seem to divide the Democrats more than the Republicans, even as a clear majority of Democrats support the teaching of shameful episodes in American history.

As with hostility to cancel culture, opposition to CRT seems to be belief-driven, rather than experience-driven, since the share of parents reporting their children’s exposure to CRT/CSJ practices is only 12% and is twice as high among Democratic parents as it is among Republican parents. Meanwhile, the quarter of people who have taken CRT-themed diversity training, all else being equal, do not rank cancel culture as a higher electoral priority than do other Americans. Republicans’ cultural liberalism is therefore likely to center on a civil rights–based defense of racial equality and freedom of conscience against mandatory CRT, with workplace protection policies for individuals’ right to free speech a second-tier concern. These results suggest that Democrats might distance themselves from CRT in favor of a more narrowly defined term such as “critical history,” limited to ensuring that students learn about the bad and the ugly, as well as the good, in the American past.

When it comes to policy, more people support than oppose government action (like the UK’s Academic Freedom Bill) that would overrule organizations like public universities or tech firms that fire employees or no-platform users of social media for legal speech.[67] A majority do not think that organizations should be able to fire employees for their political beliefs, and just over half think it should be illegal to discriminate against people in hiring—a stance backed by European law and legislation in several liberal jurisdictions such as Seattle, California, and New Mexico.[68] Excluding those who want no emphasis on equality or diversity of any kind, a position twice as common among Republicans than Democrats, nearly six in 10 want to see as much or more attention paid to improving political diversity in academia and journalism as is currently paid to improving race and gender diversity.

Across all policy issues, Republicans back government regulation to protect cultural liberalism more than Democrats do, but only by a 10- to 20-point margin. Older Republicans are more skeptical than younger Republicans about government regulation of organizations, indicating a possible generational shift within the GOP base toward a more interventionist and less libertarian form of conservatism. Beyond this, the breaks by party identity suggest impressive bipartisan support for government protections against threats to free speech and political discrimination in organizations, alongside measures to promote political diversity and prohibit the teaching of the more divisive aspects of CRT in schools.

Irrespective of the direction of policy, the mounting conflict between cultural liberalism and cultural socialism promises to define the politics of the current era.

See Appendix.

See Endnotes in PDF.

Eric Kaufmann is an adjunct fellow at the Manhattan Institute and a professor of politics at Birkbeck College, University of London. Kaufmann’s scholarship focuses on cultural politics, religious and national identity, and demography. He is the author of Whiteshift (2018/19), Shall the Religious Inherit the Earth (2010), The Rise and Fall of Anglo-America (2004), and The Orange Order (2007), among others. He has co-authored reports on academic freedom and the political response to demographic change, and edited books on demography and ethnicity. An editor of the journal Nations & Nationalism, he has written for the New York Times, Newsweek, Foreign Affairs, New Statesman, National Review, and Prospect. He holds a Ph.D. from the London School of Economics and Political Science and lives in London.