Academic Freedom in Crisis: Punishment, Political Discrimination, and Self-Censorship
CSPI Report No. 2
Summary
This study is the first of its kind to investigate authoritarianism and political discrimination in academia, relying on survey responses from both the perpetrators and targets of discrimination.
Across three Anglophone countries, a significant portion of academics discriminate against conservatives in hiring, promotion, grants and publications. Over 4 in 10 US and Canadian academics would not hire a Trump supporter, and 1 in 3 British academics would not hire a Brexit supporter.
Gender-critical feminist scholars appear to experience even more discrimination than conservatives. Only 28% of American and Canadian academics would feel comfortable having lunch with someone who opposes the idea of transwomen accessing women’s shelters.
Most professors do not back cancel culture in its most authoritarian forms. Only 1 in 10 academics supports firing controversial professors. Nonetheless, while most do not back cancellation, many are not opposed to it, remaining non-committal.
Right-leaning academics experience a high level of institutional authoritarianism and peer pressure. In the US, over a third of conservative academics and PhD students have been threatened with disciplinary action for their views while 70% of conservative academics report a hostile departmental climate for their beliefs.
In the social sciences and humanities, over 9 in 10 Trump-supporting academics and 8 in 10 Brexit-supporting academics say they would not feel comfortable expressing their views to a colleague. More than half of North American and British conservative academics admit self-censoring in research and teaching.
Younger academics and PhD students, especially in the United States, are significantly more willing than older academics to support dismissing controversial scholars from their posts, indicating that the problem of progressive authoritarianism is likely to get worse in the coming years.
A hostile climate plays a part in deterring conservative graduate students from pursuing careers in academia. Conservative and liberal graduate students differ far more in their perceptions of whether their politics fit academia than they do on questions related to how well academia pays, the isolating nature of the work, and other aspects of the profession.
One policy option is for government to proactively apply the law to universities, instituting sanctions for institutions that repeatedly breach individuals’ academic freedom while opening up a means for plaintiffs to appeal around their universities to a regulatory ombudsman. While this report makes no policy recommendations, this approach has been largely adopted by the British government.
For the executive summary, click here.
Introduction
This report represents the most comprehensive survey-based investigation to date of academic and graduate student opinion on political discrimination, the punishment of academics for speech, and experiences of hostility and self-censorship for political beliefs. It examines evidence from the perspective of both victims, largely concentrated among the minority of 5-10% conservative or gender-critical academics; and perpetrators, involving a substantial minority or sometimes even a majority of scholars. The report is divided into four parts.
Part I concentrates on what I term hard authoritarianism, notably the experience of being disciplined or threatened for speech – or backing the firing or disciplining of controversial academics; as well as support for policies such as mandatory reading list quotas that abridge academic freedom. Part II examines soft authoritarianism, encompassing views on political discrimination, as well as experiences of hostility and self-censorship in scholarship, teaching, and other aspects of academic life. This part replicates the findings of many existing studies but also uses new methods to go beyond them. Part II is subdivided into three sections, the first on the political leanings of academics, the second on the extent of chilling effects and self-censorship, and the third on support for political discrimination.
Part III goes off campus to ask whether the patterns of hard and soft authoritarianism we find among academic staff and graduate students are also present among college graduates working in organizations off campus. We’ll see that there are differences, but also many similarities, between the two professional realms. Part III also discusses potential policy reforms to address the problems highlighted, comparing and contrasting interventionist approaches against those that are more hands-off.
The report is meant to be read as an integrated whole but can also be accessed in a modular way by clicking on a heading in the table of contents. The reader can return to the table of contents at any time by clicking on any heading in the text. The same is true for the table of contents of tables and figures.
The Surveys
This study relies largely on survey data of academics and PhD students at universities in the United States, Britain, and Canada. It also summarizes prior work on staff and students in the United States and Europe and contextualizes findings with surveys of the general population and the wider advanced degree-holding public. The core of the study concentrates on academic attitudes, where there has been less research than on student attitudes.
In order to mitigate the effects of social desirability bias – people giving socially acceptable responses – I use a list experiment.1 This is an innovation in work on political discrimination, and gives a more accurate indication of how prevalent the willingness to discriminate on political grounds is. In the analysis, I focus more on the social sciences and humanities because political considerations are a larger aspect of these fields’ conceptual foundation, and are thus assumed to exert greater influence over the culture and practices of these disciplines.
The UK data comprise the highest quality sector-wide academic survey to date, based on a 61-76% response rate from the approximately 1100 current and retired academics that happen to be on YouGov’s 500,000 strong national British panel. Since these are respondents who are in the YouGov system answering other surveys, there is no danger that they are self-selecting into the survey due to their interest in answering questions on academic issues. US and Canadian surveys represent a mix of online surveys using convenience samples with low response rates from a large target pool, and platform-based surveys with very high response rates from smaller target populations. I also fielded a UK online mailout survey to academics in the top 100 (out of 143) universities to compare with the relatively randomly-selected YouGov UK sample and the online North American surveys.
Though response rates to the North American and British mailout surveys are only around 2-4%, the data they provide tell a very similar story to the non-selective YouGov and Prolific surveys (with 61-87% response rates), and to previous questionnaires fielded by other scholars. The result is a triangulated body of replicated knowledge on political discrimination and chilling effects that has not been contradicted and thus should no longer be seriously disputed.
In addition to replicating previous studies of soft authoritarianism (political discrimination and chilling effects), I break new ground by examining support for hard authoritarianism – in this case dismissal campaigns for controversial academics, and penalties for failing to implement mandatory curriculum diversity quotas. I also use the surveys to create quantitative models that, in a departure from existing studies, examine the relationships between the ideological skew of staff, political discrimination, self-censorship, and support for cancelling controversial scholars or employees.
Table 1 shows the methods and population sampled in each survey discussed in this report. More details of the surveys may be found in the Appendix.
Part I: Hard Authoritarianism
As we shall see, there is a small minority of academics that supports ousting those with controversial views. Administrators are often willing to use the university’s disciplinary apparatus to enforce this sentiment, and it is not clear that academics are sufficiently motivated to resist this. While it is rare for an academic to be fired, especially on ideological grounds, a number of worrying cases have recently come to light in which life has been made so uncomfortable for a person that a dissident scholar has been forced to leave.
No-Platforming and Firing
This forms part of a growing climate of political intolerance. In the United States, the Foundation for Individual Rights in Education (FIRE) maintains a database of disinvitation incidents. These are presented below in Figure 1. Incidents rose steadily in the 2000s, driven mainly by left activists, and spiked in 2015 when Nicholas Christakis was mobbed by a group of Yale students and Jonathan Haidt and Greg Lukianoff published their seminal piece on the phenomenon in the Atlantic, “The Coddling of the American Mind.” Since then, incidents have remained at an elevated level compared to the pre-2010 period.
The National Association of Scholars (NAS) maintains a database of (mainly) American academics who have experienced campaigns calling for their dismissal. The database records 4 incidents apiece in 2015 and 2016, 9 in 2017, 13 in 2018, 12 in 2019, and a striking 65 in 2020. The database relies partly on reports from users, thus it is unclear whether the surge in 2020 reflects a disconcerting new rise in intolerance or the prevalence of real-time over historic reporting since the site was established in June 2020. Regardless, 65 cancellations in one year suggests the problem cannot be dismissed as purely anecdotal.2 Meanwhile, under the radar, in just one 3-week period in the summer of 2020, John McWhorter of Columbia University received 150 messages from academics sharing their anxieties about their work climate, and he continues to receive at least one per week.3
In Britain we see a similar, if slightly delayed, pattern for deplatforming and dismissals. The trend in combined free speech incidents, as catalogued in two separate lists, with a minority of overlapping cases, is presented in Figure 2 and shows a noticeable rise from 2013, and especially from 2018. Note that 2020 data are incomplete, but already show a higher level of incidents compared to 2017.
A report by Civitas in December 2020 discovered that over half (53%) of all 137 British universities experienced demands for censoring speech around alleged “transphobic” episodes during 2017-2020. Many UK universities were also targeted by activists seeking to suppress other forms of free speech. Over half (55%) of universities had at least one open-letter campaign, 37% reported social media drives, and nearly a quarter bore the brunt of a campus pressure group drive seeking to curtail academic freedom or free speech. These forms of political intolerance thereby encompassed a majority of British universities.4
Many other cases occur beneath the radar: in a one-month period from June 13 to July 14, 2020, the UK’s Free Speech Union received 10 calls for assistance from academics or contractual staff asking for help to defend against online campaigns or university disciplinary charges brought against them for public statements. None are listed in the data compiled by Wanstall or Biggs. A call for evidence on Twitter in mid-2019 by Professor Kathleen Stock of the University of Sussex turned up 28 testimonies of restrictions, threats, and harassment from British gender-critical academics in just 7 days.5
While successful dismissals are rare in academia, there have been a number of notable cases, including those of Bo Winegard of Marietta State College and Noah Carl and Jordan Peterson at Cambridge University. In Britain, cases include Chris Hill, formerly at the University of Central Lancashire in July 2018, and Andrew Dunn, formerly a Social Policy lecturer at the University of Lincoln.6 Peterson’s case involved guilt by association with an Islamophobic fan whom Peterson did not realize to be anti-Muslim. The Hill case involves remarks some found offensive, a view upheld by the university. Hill claims that accusers were misrepresenting his views out of hostility to his conservatism. Likewise, the Dunn case involved little more than anti-conservative discrimination leading to the lecturer being forced out.
There are also an unspecified number of disciplinary actions that have not been reported by the media, many of which are subject to privacy rules, but some of which are alluded to in comments I collected in my NAS survey, such as:
“One professor at East Georgia State College was fired for ‘sexual harassment’ for disagreeing with the school’s sexual harassment policy. I have heard many stories from credible sources of [sociology] professors, students and graduate students being subjected to secret title IX trials and summarily had their lives ruined by their respective schools. Kansas State University being the worst offender.” –Politically right-wing sociologist, US
The Iceberg Model of Discrimination
We can think of the threats to academic freedom using the metaphor of an iceberg, with items that make the news – such as deplatformings and dismissals – as the visible symptoms of a much deeper problem. Figure 3 and Figure 4 present some of the main findings of our survey, from the perspectives of the victims and the perpetrators of discrimination.
As can be seen, deplatformings or dismissals affect only a tiny minority of academics. Yet when one polls academics, particularly centrists or conservatives, on their experiences, we find that there are some forms of political discrimination that are experienced by a majority of political minorities on the faculty. By and large, most right-leaning academics feel a hostile environment or self-censor in some form or another.
Surveys of the rest of academia show that conservative scholars are not imagining these things. Only a small minority of academics is willing to endorse dismissal campaigns. Nonetheless, depending on the question asked and the way that it is asked (i.e., whether a “list” method is used to get around social desirability bias) a substantial minority or an overwhelming majority engages in discrimination. Among PhD students, the results are even more extreme than they are among other academics, with over 80% of American PhDs willing to discriminate against right-leaning scholars on at least one dimension.
Disciplinary Action and Bullying
In July 2020, a high-profile statement from 150 leading liberal writers and academics, including J.K. Rowling, Noam Chomsky, and Salman Rushdie, warned of the rise of a “cancel culture” in academia, the arts, and the media. “It is now all too common to hear calls for swift and severe retribution,” they write, “in response to perceived transgressions of speech and thought.” Noting the “hasty” decisions of administrators to fire transgressors, they continue that “We are already paying the price in greater risk aversion among writers, artists, and journalists who fear for their livelihoods if they depart from the consensus, or even lack sufficient zeal in agreement.”7
However worrying the trend in no-platformings and dismissals, this is merely the most visible manifestation of a deeper and more pervasive problem in academia that rarely makes the headlines. The most authoritarian aspect of academia that flies under the radar is disciplinary action and ideological bullying, much of which stem from fellow academics acting in the role of departmental manager or peer. A study for the Universities and Colleges Union (UCU), the main union for academics in Britain, surveyed several thousand British and European academics about their experiences of being disciplined for speech.8 The authors asked, “Have you ever been subjected to informal or formal disciplinary action, or the threat of disciplinary action (up to, and including, dismissal) because of academic views expressed in [the following]?” Questions were asked about “academic views expressed” in teaching, research, within institutions, in public fora, and elsewhere. Results appear in Figure 5.
While I am unable to obtain the UCU study’s crosstabulations by scholars’ ideology, I can compare their UK and EU data to American and Canadian survey results I have collected on my online mailout survey using the same UCU question wordings. These suggest that right-leaning academics are more likely to mention that they have been the target of disciplinary action for their public statements, research, or teaching.
My US online survey results show that right-leaning academics experienced 16-100% more academic discipline than colleagues on the left on 5 of 6 measures. I did not put these questions to my UK academic samples, but according to my Prolific PhD samples (with 72% of PhDs eligible taking the survey), right-leaning US and Canadian PhD students report 3 to 6 times the level of disciplinary action as PhD students with centrist or leftist views. Thirty-eight percent of conservatives were disciplined, or threatened with discipline, for their views in at least one of the following five spheres: teaching, research, speech in private or public forums, or speech elsewhere.
The Prolific North American PhD survey and mailed out North American academic survey contain only small samples of right-leaning respondents (N= 40 PhDs and 43 academics). To rectify this, I ran a survey of the membership of the US-based National Association of Scholars (NAS). Two-thirds report a right-wing identity, and 13-28% said they experienced disciplinary action on any given dimension.9 Results show that among this group, there is a substantially elevated level of victimization, with 43% reporting being disciplined, or threatened with discipline, for their views in one of the five spheres listed above.
These results suggest that the disciplinary threat level is higher for right-leaning academics and especially doctoral students. It helps explain why I find that right-leaning UK academics are 3-4 times as likely as those of other political ideologies to report a “hostile climate” for their beliefs and over twice as likely to say that they self-censor. I estimate that right-leaning UK academics and PhD students are 50-100% more likely to experience disciplinary threats than leftist and centrist academics.
Across a broader range of disciplinary measures in Figure 6, we see the right-leaning membership of the NAS generally reporting 2-3 times as much authoritarianism as the overwhelmingly left-leaning respondents in the Karran and Mallinson EU/UCU survey.
Approximately 6-9% of NAS members experienced the most severe forms of treatment for their views, including being moved to another department or center (6%), demoted (8%), being physically attacked (8%), or having research or facilities withdrawn (9%). A quarter of NAS respondents reported being falsely charged or threatened with charges for their views, and nearly a quarter were given more, fewer, or different research, administrative, or teaching duties in response to things they said or wrote. More generally, around a quarter were disciplined or threatened for views expressed in teaching and public or university forums.
Finally, colleagues were the source of most of the problems reported by respondents in all datasets. Half of NAS respondents reported psychological pressure from colleagues for their views and 36% said they experienced bullying for expressing them. The connection between colleagues, departmental authority, and higher-level administration is important. Departmental colleagues wield considerable power, both through informal peer pressure and via their formal departmental managerial roles as well as through faculty and college-level committees that administer the university.
Hard Authoritarianism Testimonials
Some reported being targeted by the university’s disciplinary apparatus. The following are testimonials showing cases of hard authoritarianism.
“The university is right now considering disciplinary proceedings because of innocuous items I posted on my private blog. I think I’ll be alright, but conservative academics are somewhat persecuted for their views.” – Right, Geography, US
“Only recently has the attitude of the administration turned to punitive measures for people who are outspoken; thus most of us are being very careful about what we say and make sure we are covered by our collective agreement to fend off the administration.” – Centrist, Classics, Canada
“Yes, every year I am called into my boss’s office because of some controversy surrounding course curricula. At times, I have been misunderstood. Other times, I have to justify my work with students.” – Right, Education, US
“I have been dismissed as undergraduate programs chair in my department because I am a gender critical feminist. Students currently have a petition underway supporting this dismissal because my faculty association is supporting me in a grievance about it. I am also facing an interestingly timed investigation into my research ethics, supposedly ordinary procedure any time a faculty member is featured in the news but the first time I’ve been investigated.” – Very Left, Anthropology, Canada
“A professor who declined to refer to a student by a false pronoun (a male wanted to be called ‘she’) was visited by the diversity people and had all of his work and his lab scrutinized. He was placed under specific administrative guidelines for how he was to deal with this student.” – Right, Communications, US
“Professor Dennis Gouws of the Humanities Department has been subjected to years of harassment, denial of sabbatical, placement on official probation, threats of dismissal, and finally administrative decree commanding him to stay silent or be fired (which he was forced to obey). The reason was his written and spoken criticism of gender feminism.” – Right, History, US
“A friend of mine at another university got a negative tenure vote despite a strong record for having unpopular opinions (and research findings). Fortunately the provost overruled. One of his collaborators at a third school was subject to intense legal harassment (FOIA requests etc) though this was *mostly* not internal to his university.” – Right, Sociology, US
“I had heard from a peer that one of their close friends had been in the program the year prior to ours, and halfway through his department head sat him down and told him he ‘wasn’t cut out to be a teacher,’ and that he was being cut from the program. My peer believed it was because he expressed strong conservative views in class, perhaps not very diplomatically or respectfully.” – Right, Education, US
Several Leave-voting and conservative academics in my UK YouGov survey report instances of authoritarianism from administrators or colleagues, impact on their freedom to teach and research, and even threats to their jobs.
“Yes, indeed I have lost two senior jobs because I voted leave.” – Tory Leaver (Supporter of Brexit)
“I have been called in for a meeting with University marketing, my Head of Department, and an HR officer after I published an article in a peer-reviewed academic journal [redacted]. They asked why I had not explicitly condemned conservatism as immoral within this article. I explained that I did not believe it was appropriate for me to use my position as a researcher to subjectively pass judgement on modern political ideology. I was told that there are some subjects I shouldn’t remain neutral on, and that I have a moral duty to condemn those on the political right. I was told that, if I insisted on remaining impartial within my research, I was not to further research this subject and warned I may face disciplinary hearings if I did.” – Tory Leaver
“Given the derogatory views regularly expressed by my colleagues about Leave supporters, including the VC sending a University-wide email referring to us as ‘Little Englanders’, I have no doubt that if my views were known then it would negatively affect the attitude of my colleagues towards me significantly. It probably wouldn’t be career-ending, but it would reduce my influence, make it harder for me to deliver my teaching and leadership responsibilities, and quite likely force me into a position where I would have to move institutions.” – Tory Leaver
“I have to be careful about where I place my research because of two different areas of my research. Colleagues have attempted to stop me teaching.” – Tory Remainer
“I had a professor who was very left wing and who reduced the promotional chances of anyone that was centrist or slightly right of centre.” – Tory Remainer
“A previous line manager had a large photo of Jeremy Corbyn on her desk. When I failed to approve (I said nothing) she had me removed from the programme despite very positive feedback.” – Centrist Remainer
“Yes – I avoid making political statements – have also had head of department voice strong disapproval for the sort of research I do and to use the ethics approval system to prevent certain research topics being studied.” – Left Labour Leaver
These comments go beyond the chilling effect of social opprobrium from colleagues and identify the potential for abuse of power by those in the hierarchy within the university. At all levels, the institution can deploy sticks such as assigning unpopular administrative tasks or courses, limiting access to grant funding or promotion, or even forcing a staff member to leave.
The latter is especially damaging, not just for those who lose their jobs, but for the signal it sends across the sector. Academic jobs are incredibly competitive and specialized. A lecturer or professor on a permanent contract who is dismissed or forced to leave on a package is unlikely to find a permanent academic job in the same geographic area, especially if they are older, in a social science or humanities field, and have acquired a reputation as conservative. The skill sets and networks of academics may also make it difficult to make a lateral move to another profession mid-career. All of this instills a powerful sense of caution in many full-time academics, with a desire not to jeopardize a fulfilling and reasonably (though not highly) paid career. This leads them to self-censor, further constricting viewpoint diversity. This impairs the exchange of ideas that is vital for both the academic enterprise, and for beginning to intelligently negotiate between advocates on both sides of society’s major political divide.
Left-Wing Authoritarianism
Hard authoritarianism on campus is often spearheaded by a left-wing authoritarian minority. Recent work in social psychology identifies three distinct clusters within the umbrella concept of left-wing authoritarianism. The first revolves around economic orientations, with high agreement with phrases such as “The rich should be stripped of their belongings and status.” The second focuses on culture, notably “Deep down, just about all conservatives are racist, sexist, and homophobic.” A third is focused on authority, i.e., “I must line up behind strong leaders who have the will to stamp out prejudice and intolerance.” The last two are especially germane to my analysis. The underlying orientation is a preference for using power or violence to upend the current moral order and replace it with a new regime that compels people to conform to new principles. This orientation is characterized by moral absolutism, intolerance of dissent, and a preference for censorship, exhibiting considerable overlap on many psychological measures with right-wing authoritarianism.10
Testing Support for “Cancel Culture” Among Academics
How does left-wing authoritarianism manifest itself within the professoriate? Having surveyed the hardest forms of internal discipline in academia, I next move to examine the level of backing for “cancel”-style dismissal measures. To do so, I probe responses to controversial research findings adapted from actual cases, but worded in an abstract enough way that individuals are unlikely to make a connection to a particular high-profile incident. While most people do not endorse these research statements, and expect that most academics feel likewise, I sought to use them as a litmus test, to empirically measure the extent to which scholars may disagree with the substance of findings but agree that those undertaking unpopular research should not be driven from their posts. The questions I selected are as follows:
If a staff member in your institution did research showing that greater ethnic diversity leads to increased societal tension and poorer social outcomes, would you support or oppose efforts by students/the administration to let the staff member know that they should find work elsewhere? [Support, oppose, neither support nor oppose, don’t know]
If a staff member in your institution did research showing that the British empire did more good than harm, would you support or oppose efforts by students/the administration to let the staff member know that they should find work elsewhere? [Support, oppose, neither support nor oppose, don’t know]
If a staff member in your institution did research showing that children do better when brought up by two biological parents than by single or adoptive parents, would you support or oppose efforts by students/the administration to let the staff member know that they should find work elsewhere? [Support, oppose, neither support nor oppose, don’t know]
Please imagine a member of your organization has done work showing that having a higher share of women and ethnic minorities in organizations correlates with reduced organizational performance. Several thousand professionals, some from your organization, have signed an open letter calling for the staff member to be fired in order to protect disadvantaged groups from a hostile learning environment. A small group have started a counter-petition defending the staff member on grounds of academic freedom. Would you: [a) Sign the open letter, which called for the staff member to be fired, b) Support the views expressed in the open letter, but choose not to sign it, c) Not support nor sign either letter, d) Support the counter-petition, but choose not to sign it, e) Sign the counter-petition, f) Don’t know.]
Results are displayed in Figure 7. These show American faculty’s willingness to support a “cancel” campaign to dismiss an academic on the one hand, or, on the other hand, to oppose it publicly or privately. Findings reveal an important reservoir of support for academic freedom among staff at US universities, with just 7-18% of lecturers and professors, depending on the issue, willing to back campaigns to fire academics who dissent from norms on hot-button issues.
As a comparison, I include a question on whether academics would support or oppose firing a member of staff who wants immigration to be reduced. Eight percent of US faculty would favor the dismissal of a restrictionist academic, but 78% would oppose this, with 14% uncertain. Most of those surveyed would undoubtedly take a progressive position on these five questions, but most also recoil from campaigns to remove academics who adopt a dissenting view on them in their research.
The flip side of this largely positive portrait, of course, is that there is a 7-18% minority of American faculty who would support dismissal campaigns that directly violate academic freedom. In addition, the share of academics who oppose a given dismissal campaign is not higher than 52% for any issue other than the immigration question, and falls to just 31% in the example of a study finding that a higher share of women and minorities lowers organizational performance. Many are noncommittal, including about half of those on the question just cited – unwilling to cancel but also unwilling to oppose those who would seek to do so.
British YouGov data, in Figure 8, show very similar findings. Just 6-13% support dismissal across the four campaigns, but no more than 51% would oppose a dismissal campaign. Eighty-three percent of British academics would oppose firing a member of staff who favored reducing immigration, slightly higher than in the US data. For the British mailed survey, the numbers are similar: 4-18% favoring dismissal, but with a maximum of only 54% who would oppose these actions.
As in America, the softest opposition to dismissal was in the “women and minorities lower performance” case, with just 27% willing to oppose a dismissal campaign and 6 in 10 unsure. The British mailout online survey showed results nearly identical to those in the YouGov survey.
These findings replicate with PhD students: in the UK, just 25% would oppose dismissal in the “diversity lowers performance” case, but 65% would in the immigration restrictionist case and 31-34% would for the diversity, empire, and parenthood questions. For North American PhDs the analogous numbers are similar: a mere 19% opposing cancellation in the diversity-organizational performance case, 50% opposing dismissal for an academic who backs immigration restriction, and 29-50% against firing on the diversity, empire, and parenthood hypotheticals.
PhD students on both sides of the Atlantic are noticeably more likely to endorse dismissal than academics: in the UK, 9-10% of PhD candidates back dismissal for immigration restrictionists and those doing research on the parenthood, diversity, and empire questions, rising to 27% for the diversity-organizational question. In North America, fully 25% of PhD students would seek to force an immigration restrictionist academic from their job. On three other scenarios, between 11 and 17% back dismissal, and for the diversity-organizational case, fully 41% back cancellation. These figures are worrying inasmuch as they concern the generation entering academia. It seems that acquiring a post generates a new appreciation for job security, countering these impulses somewhat, but, as we shall see, younger academics are more politically intolerant than their elders, portending a rising illiberalism problem.
The general pattern, therefore, is that the majority of academics is foursquare against cancellation only in the case of an immigration restrictionist academic. For the hypothetical case of a scholar finding that race and gender diversity reduces organizational performance, only 19-31% of academics or PhDs across all five surveys said they would oppose, publicly or privately, an open-letter campaign to get the academic fired even as few actively endorse a cancel campaign.
Who Backs Dismissal?
Compiling responses to the first four binary questions (all except “restrict immigration”) in Figures 7 and 8 above into an index allows us to ask about the characteristics of people who support illiberalism when it advances the cause of perceived racial and gender justice, broadly construed. Combining the four questions, each with answers as 0 for anti-dismissal or 1 for pro-dismissal, into one measure (also scored as 0 or 1) allows me to develop an approximate “likelihood to expel” score for each person in the surveys.11
Using this measure, an average of 20% of current UK faculty, rising to 25% in the social sciences and humanities, back at least one dismissal campaign. On the other hand, just 5% of current British academics in the social sciences and humanities back all four illiberal measures.
Comparing across the US, Britain, and Canada in Figure 9, including academics and PhD students, shows that a majority of both staff and PhD students do not support dismissal campaigns on hot-button issues around race, gender, and sexuality. Indeed, only 2% of American academics and PhD students support all four dismissal campaigns.
However, fully half of North American social sciences and humanities (SSH) PhD students backed at least one illiberal measure, as did 36% of UK PhDs and 20-25% of academics. PhDs are more authoritarian than academic staff. Is this because they don’t have a job to worry about losing? Perhaps. But, as we shall see, another (less powerful) reason is because they are younger than professors. Age plays an important role among both professors and PhD students in predicting their willingness to cancel an academic for transgressing progressive values.
Figure 9 above shows that 24% of American SSH academics, 20% of Canadian SSH academics and 25% of current British SSH staff endorse at least one of four dismissal actions – even though only 1-2% back all of them. As a rough guide, therefore, around 1 in 10 SSH academics back any single dismissal campaign and close to a quarter support at least one of the four hypothetical cases offered. Against this, between a quarter and a half oppose a given dismissal campaign.
Yet in almost all cases there is a large undecided group. Around half are unsure in 3 of 5 cases – they neither support dismissal nor oppose it. By contrast, only 1 in 5 are unsure on the immigration restrictionist case while more than half are unsure on the diversity and organizational performance case. The unsure group is often the silent majority among staff and PhDs when it comes to the more controversial cases. I would argue that this lack of certainty reflects a cross-pressuring between a defense of academic freedom and job tenure on the one hand, and a desire to protect disadvantaged identity groups on the other. For instance, as we shall see, most British academics diverge substantially from the wider UK public by overwhelmingly (76%) supporting the notion that the virtues of political correctness in protecting minorities outweigh its threat to free speech.
Far Left Activists more likely to Back Dismissal
Far-left and activist scholars are a substantial minority of SSH academics in North America and Britain. Figure 10 shows that far-left scholars who agree, to a greater or lesser extent, with the statement “I would consider myself as an activist” make up between 8 and 18% of SSH staff across the US, Britain, and Canada, with far leftists at 16-28% and self-described activists at 26-38%. The British YouGov sample, which is arguably the most representative, shows a lower share of far-left activists than the UK mailout survey. On the other hand, the PhD surveys, which captured 62-85% of PhD students on Prolific Academic, found that 35% of 124 North American SSH PhDs identified as far left and 44% as activists. In Britain, the 77 SSH PhDs were less likely to self-identify in these ways, with 22% far left and 19% activist.
Ideology and Intolerance
Illiberalism is much higher among academics who identify as far left, especially if activist. Far-left activists are often in the forefront of “social justice” activism within the university, and many challenges to academic freedom stem from this group.
Thirty-seven percent of US academics, 40% of Canadians, and 40% of British faculty who identify as far left favor at least one firing campaign. Among far-left activists, 40% of US, 51% of Canadian, and 41% of British academics back at least one dismissal campaign.
Are Young Scholars Less Tolerant?
Even among far-left activists, it is noteworthy that a bare majority would not support forcing dissenting voices from the academy in any given case. The UK data in Figure 11 show how support for dismissal rises as we move from all current and retired staff surveyed, to current staff only, to current SSH staff, and then to far-left and far-left activist faculty.
Is age still important when we take a person’s ideology into account? Maybe young people are just more leftist, so the two factors are confounded? To answer this question, I construct a model, with willingness to dismiss (0 for no, 1 for yes) on any of the four questions as my outcome measure. The results for Britain are shown in Figure 12.
Ideology and age are the main predictors of a scholar’s support for dismissal and have separate and important effects. Left-right ideology is the most important predictor of viewpoint intolerance on the four aforementioned items in Britain, but only “very left” (when contrasted to centrist) is statistically significant in picking out the intolerant academics in the British YouGov sample, with those on the moderate left not more likely to favor hard authoritarianism than those in the center.12 The average likelihood of an academic endorsing a dismissal campaign rises 50 points as we move from a “very right” academic to someone on the far left.
But age also matters. Younger staff, regardless of whether they are far left or moderate, tend to be more illiberal than older staff. This is concerning, as it points to the possibility that the potential for authoritarian activism is likely to rise in the years ahead. Much turns on whether this reflects the fact that younger academics prioritize social justice over academic freedom more than senior staff; or whether this is a maturity issue, with younger staff likely to outgrow their relative illiberalism as they age.
Among the other characteristics that one might assume are associated with favoring speech restrictions in the name of sensitivity are being female and nonwhite. However, in the model in Figure 12 above, neither reached conventional statistical significance. Ethnic minorities were slightly more favorable to removing academics that perform dissenting research on racially-sensitive issue topics when I disaggregated the outcome to focus only on the race-related issues. Being female was not a statistically significant predictor of opinion on any campaign. SSH academics were not more illiberal once I controlled for ideology. This is because SSH disciplines are more far left than STEM, and when this is screened out, the effect of SSH falls away. British academics who share their opinions on Twitter are also somewhat more likely to back expulsion, though this is only significant in some models.
The situation in North America in Figure 13 shows a very similar pattern, albeit with some shifting of the predictors and a stronger overall model. Far leftism is again a powerful predictor of illiberalism, along with age. Younger staff are consistently more likely to favor dismissal – indeed this is the most powerful predictor in the North American model. Interestingly, for North America, SSH academics are again no more illiberal than their STEM colleagues when you control for the fact that SSH has more far-leftist members of staff.
Activist and moderate-left academics are also more supportive of dismissal in North America than centrist staff, whereas in Britain this was not significant. These findings reflect the fact that centrist academics in North America express sentiments (such as feeling censored) that are more similar to conservatives whereas in Britain centrists report low self-censorship and seem closer to moderate leftists in their appraisal of questions of academic freedom.
In addition, women and minorities are more illiberal than men and whites in North America, whereas these characteristics did not predict higher levels of support for dismissal in Britain. The US findings for minorities are accounted for by a somewhat higher effect among African-American respondents (N=36), although black academics were not more illiberal than others in Canada or Britain, granted with very small numbers being polled. Finally, there was no significant difference between American and Canadian academics in their support for dismissal campaigns.13
Concentrating on the main drivers of illiberalism in the British YouGov data results in the chart in Figure 14. This shows that being far left increases the chance of favoring a dismissal campaign by about 25 points. Age matters as well: between age 30 and 70, the chance of backing the ouster of an academic for politically incorrect research decreases by around 15 points. Together, they exert a powerful effect. Thus a 30-year-old far-left British academic has a 1 in 2 chance of endorsing one of the dismissal campaigns, while a 70-year-old who is not far left has barely a 1 in 10 chance of doing so.
Among both staff and PhD students, there is a significant age gradation, with younger staff more intolerant. Are PhD students more pro-cancellation than academics? Figure 15 compares American PhD students and academics in the 23-33 age range, which contains 8 in 10 PhD students in my sample. There is also an important number of academics in this age bracket in my academic survey, resulting in a combined dataset for this age group consisting of 60% PhD students and 40% academics.
Results of a model that controls for gender, race, and age (PhD students contain more women and minorities and are younger) show that PhD students are more intolerant than academics, even controlling for age. However, the difference is concentrated among STEM PhDs. Among scholars in the social sciences and humanities, there is no detectable difference between PhDs and academics (this is confirmed in a model that excludes STEM participants). In addition, combining American, Canadian, and British respondents weakens the difference between academics and PhD students in Figure 15, leaving it statistically insignificant.
To examine the raw effect of studying for a PhD as compared to holding an academic position, where a person may feel the need to protect against risk to job loss, I run a model in Figure 16 looking at all age groups using a combined sample of 2 North American and 3 British surveys, combining PhD students and academics. Age is the strongest predictor of support for dismissal campaigns, followed closely by left-right ideology. PhD status is also strongly significant, along with being female rather than male. Being North American rather than British, or in STEM rather than SSH, does not predict pro-cancel sentiment.
There is an important wrinkle, however, that makes the transatlantic dimension important. Namely that American, and to a lesser degree Canadian, PhD students are noticeably more pro-cancellation than either British PhD students or American and Canadian academics. This is evident in the model in Figure 17. This shows that American PhD students 35 and under (Millennials) have a 56% chance of backing at least one of the dismissal campaigns and Canadian PhDs the same age have a 44% chance of doing so.
This compares to a 41% chance of supporting at least one dismissal campaign among academics 35 and under in Canada, a 38% chance among young American academics, a 36% likelihood for British PhDs, and a 32% chance for British academics. While there is only have sufficient data points for PhDs 35 and under in North America, it is clear they are more intolerant than academics the same age, especially in the US where Millennial academics are 18 points (38% vs. 56%) more tolerant than PhDs of the same cohort.
Millennial academics are twice as intolerant as those 55-64 in Britain, four times as intolerant in Canada, and close to three times as intolerant in the US. While younger academics are more pro-cancellation, there also appears to be a discontinuity, with the Millennial cohort substantially more intolerant than those in the next 36-45 cohort in three of four cases where I have comparable data.
Age matters more than any other factor for political toleration. In addition, being a Millennial is also important, with those 35 and under especially likely to support dismissal campaigns even when accounting for their age (continuously measured in years). While PhD students are significantly less tolerant than academics, this mainly concerns STEM PhD students in North America being significantly more pro-dismissal than STEM academics.
Support for academic freedom may wane as older academics retire, and as a less tolerant cohort of Millennial PhD students and academics replaces the more free speech-oriented Boomer and Xer generations. It is encouraging that Millennial academics appear more freedom-oriented than younger graduate students, which could be due to workplace socialization, role change, or the desire to be protected from the risk of dismissal. Nevertheless, young academics remain only half as tolerant as older academics. Intervening to familiarize undergraduate, master’s, and PhD students with the history and importance of academic freedom would seem an urgent task. Later we shall see, however, that by the time a student enters a PhD program, their views on these questions may be difficult to shift.
The emerging problem of Millennial illiberalism may stem from a rising culture of intolerance among younger cohorts rooted in a feelings-based ethic that some have termed “therapeutic totalitarianism.”14 Among American undergraduates, for instance, support for free speech appears to be very limited, with a reflexive intolerance for most forms of controversial speech. The Foundation for Individual Rights in Education (FIRE)’s 2020 survey asked 20,000 undergraduate students enrolled in four-year programs at 55 leading US universities the following:
Would you support or oppose your school ALLOWING a speaker on campus who promotes the following idea:
• Abortion should be completely illegal? 72% opposed.
• Black Lives Matter is a hate group? 75% opposed.
• Censoring the news media is necessary? 64% opposed.
• Some racial groups are less intelligent than others? 85% opposed.
• The US should support Israeli military policy? 50% opposed.
• All white people are racist? 74% opposed.
• Transgender people have a mental disorder? 72% opposed.
• Christianity has a negative influence on society? 56% opposed.
Only the small minority of conservative students consistently backed free speech across all questions.15 Figure 18 shows that just 7-18% of US undergraduates show “strong support” for free speech, with strong opposition at 40% or higher across five questions.
These findings resemble those of other surveys. A 2018 Knight Foundation survey found that students, by a 53-46% margin, favored the aim of a “diverse and inclusive society” over protecting free speech. Thirty-seven percent said it was acceptable to shout down a speaker, and 44% agreed that “people who don’t respect others don’t deserve the right of free speech.”16 Academics are more tolerant than PhD students, who might be more tolerant than undergraduates. However, it is unclear whether the rising cohort of new academics will import the intolerant ethos of the current generation of students from elite universities into the professoriate.
Conformity or True Belief?
According to Timur Kuran, many people may appear to support an authoritarian regime 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.17 Sunstein terms this “pluralistic ignorance,” arguing that this characterized mass publics during the Nazi and Soviet regimes, and does so in organizational life today.18 Is this the nature of contemporary academia with respect to support for cancelling academics on “social justice” grounds? That is, do most professors oppose cancel culture but remain silent? My fourth illiberalism question helps shed light on the problem. Recall the earlier question on how individuals would respond to a finding that more women and minorities in organizations correlates with worse performance.
Among social science and humanities academics currently in post in Britain, just 7% would sign the open letter denouncing the professor. Another 11% would support the views expressed in it. Sixty percent wouldn’t support either side or said they didn’t know. Nine percent would support but not sign the counter-petition while 13% would sign a counter-petition. These figures suggest that while only 18% of serving British SSH academics take an illiberal stance, just 13% would publicly stand for academic freedom against a “social justice” challenge. Another 9% fit Kuran’s preference falsification pattern of privately backing the counter-petition while staying silent.
Looking across the range of five surveys in North America and Britain in Figure 19 shows limited evidence for preference falsification: in almost all surveys, those who opposed dismissal were more likely to say they would publicly sign a counter-petition than remain silently opposed.
Of course, it may be that when the rubber hits the road those who say they would sign the counter-petition against dismissal get cold feet. Arif Ahmed, a Reader (Associate Professor) in Philosophy at the University of Cambridge reports that it was very difficult for him to acquire the 25 signatures needed to get his free speech motion on the ballot at Cambridge, but it passed with 80% faculty support.19
Even so, the notion that there is a silent majority motivated to oppose cancellation but afraid to do so does not receive backing from the data presented here. If this were the case, I would expect to see a clear majority opposed to dismissal across all questions, but with most saying they would only do so privately.
In fact, there is a large group that is normatively undecided. Even if there was no fear of speaking out, there might not be a majority opposing a dismissal campaign. As Figure 19 shows, the pro- and anti-free speech positions have broadly similar backing, but there is a large group of undecided or neutral respondents, comprising 30-60% of the faculty, depending on the question. I found a similar result with students in my previous report.20
Moreover, as Figure 19 – which is limited to those on the pro-freedom side – shows, most of those on the pro-freedom side are willing to nail their colors to the mast. Indeed, slightly more are likely to say they would sign publicly than are supporters of dismissal. The only exception was among North American (88% American, 12% Canadian) PhD students, where there is more evidence for a pattern of preference falsification in which 14 of 22 (about 70%) of people would support but not sign their name to an anti-dismissal counter-petition.
Cross-Pressured Between Value Commitments
Earlier, we saw that more than 6 in 10 British academics and 5 in 10 American academics are noncommittal on whether to defend the hypothetical academic who found that diversity worsens organizational performance. Many in the academy are torn between the claims of self-esteem egalitarianism and those of freedom and reason. The upshot of this is that the problem of academic freedom in academia is predominantly a battle for hearts and minds, rather than just a collective action problem like Kuran’s “private truths, public lies” that we see among those who conceal their private views in authoritarian regimes. Among North American and British academics, there are few active authoritarians, but the share who care enough about academic freedom to even privately support (i.e., without saying anything) an academic against a campaign to fire them is generally not higher than 50%.
Even among right-leaning professors and lecturers, just half in Britain and 60% in North America express private or public support for the academic freedom counter-signatories in the diversity-organizational example. This indicates that many academics have not made up their mind on questions where academic freedom and “social justice” collide. This reticence likely also informs their belief that opposing such measures may out them as a heartless conservative. It does not support the idea that a clear “silent” majority favors academic freedom but is too scared to speak up: the “emperor’s new clothes” scenario in which it takes just one person to prick the balloon.
The noncommittal stance of most staff may also have to do with the fact that free speech activism on behalf of politically incorrect researchers may be implicitly associated with the right. In addition, many on the left may implicitly accept the group affirmation that extremists provide because, as Clark notes in a review of the psychology literature on political tribalism, “Extreme, rigid, dogmatic defenders of our political ingroups demonstrate tribal loyalty that can be appealing to those of us who care about our political coalition’s success, even if we are more moderate or have more nuanced beliefs and policy preferences.”21
Focusing only on centrist and conservative academics who do not endorse dismissal – options c) through e) above, and controlling for ideology, I find that women and younger lecturers are especially likely to opt to remain neutral, with significantly greater support for actively promoting academic freedom coming from those on the right compared to those in the political center.
Gender seems to match left-right ideology in explanatory power. The model in Figure 20 shows, controlling for ideology, that there is a 54% chance of a 70-year-old non-leftist British male academic publicly or privately backing a counter-petition, but this falls to 17% among a 30-year-old female non-leftist academic. Even among conservatives and centrists, women and the young appear to be more cross-pressured between progressive and academic freedom commitments.
Combining British, American, and Canadian respondents, we see a similar pattern. What is vital for the report, however, is to appreciate the connection between viewpoint diversity and resistance to intolerance. 35% of 700 centrist and 214 right-wing academics and PhD students say they would publicly oppose dismissal compared to 21% of 1,719 respondents on the left. Among those under 40, who represent the future of academia, public opposition is even more concentrated among conservative academics (34%) compared to centrists (20%) or leftists (11%). Thus the problem of hard authoritarianism does not simply arise because the larger post-1990s pool of far-left activists in the professoriate produces more agitation for disciplinary action and, as a result, a chilling effect. It also involves a weakening of resistance to authoritarianism due to the paucity of centrists and, especially, conservatives. This limits the ranks of those who would actively resist authoritarian measures, clearing the path for institutionalized illiberalism.
A Lopsided Trade-Off: Free Speech and Political Correctness in Academia
The cross-pressuring among academics on the question of dismissal does not mean that most faculty resemble the median voter. In reality, the typical SSH academic trades free speech and emotional safety off considerably further in the direction of the latter.
Noting the way that processes of social closure operate, some argue that academia is becoming a moral community, with a set of norms that constrain the questions scholars may ask. What is termed “social closure” within particular occupations like the arts or academia involves a boundary being drawn around a community, so enabling a group-based identity to be constructed.22 At the extreme, some consider these norms to be sacred values, with the campus a “safe space” that must be kept pure, free of heretics who would profane values of cultural equity and diversity. For instance, sociologist Christian Smith speaks of a teleological “liberal progress narrative” that has come to dominate the field of sociology in America. He summarizes it as follows:
Once upon a time, the vast majority of human persons suffered in societies and social institutions that were unjust, unhealthy, repressive, and oppressive. These traditional societies were reprehensible because of their deep-rooted inequality, exploitation, and irrational traditionalism ...there is much work to be done to dismantle the powerful vestiges of inequality, exploitation, and repression.23
On Smith’s account of the liberal progress narrative, a set of “social justice”-based normative commitments are prioritized above even the search for truth.
The question of political correctness is a useful barometer for assessing how far outside the mainstream academia lies. More in Common’s Hidden Tribes report segments US and British society into values “tribes” based on their survey responses to ideological questions. The “Progressive Activist” segment of society makes up 8% of the American adult population and 13% in Britain. It is largely white, urban, and professional, leaning consistently left across all issues, and bulks larger in academia than elsewhere. In Britain, this group is 6 times more likely than average to post political content on social media, and in America, 3 times more likely. When it comes to political correctness, Progressive Activists stand apart from all other groups in support. For instance, 80% of Americans say “political correctness has gone too far” but just 30% of Progressive Activists agree. In Britain, the comparable numbers are 72% and 28%.24
Younger Americans are more pro-PC. Within the general population, the 2016 American National Election Study (ANES) pilot survey asked:
There’s been a lot of talk lately about “political correctness.” Some people think that the way people talk needs to change with the times to be more sensitive to people from different backgrounds. Others think that this has already gone too far and many people are just too easily offended. Which is closer to your opinion?
Among whites, and controlling for education level, Figure 21 shows that the youngest respondents (born in 2000) are, on average, 25 points more likely to favor the politically correct position than those born in 1940. Though ideology and party identity are far stronger predictors, age is still associated with support for political correctness among American whites when other variables are held at their mean values. That is, even when screening out ideology and partisanship, a 16-year-old remains 15 points more politically correct than a 76-year-old.
Most Britons and Americans oppose PC. However, most also oppose hate speech, which is often used to justify PC. When the question is phrased as a trade-off between speech restrictions and discrimination, a clearer picture of people’s priorities emerges. I use UK data due to the enormous sample size of YouGov’s UK panel, from which my academic survey is drawn. Consider a question asked across YouGov’s British panel:
Thinking about political correctness, are you generally in favor of it (it protects against discrimination), or against it (it stifles freedom of speech)?
Figure 22 shows that when anti-discrimination and free speech are set against each other in this way, 47% of 164,000 members of the British public oppose political correctness and 37% support it. Against this, the balance among the 603 academics in my YouGov sample (includes both STEM and SSH, active and retired) who also answered the PC question as part of separate polling that is included in the Profiles database, is 64% in favor and 31% against. Among the 176 currently employed academics in the social sciences and humanities, however, I find 76% backing PC and a mere 20% against. For comparison, even among the sample of 56,000 university graduates in the sample, the pro-PC tilt is just 48-41, much narrower than in academia. In a related vein, Samuel Abrams finds that in the US, 78% of Democratic, but only 39% of Republican, academics support safe spaces.25
Within the sample of 603 British academics who answered the political correctness question in Figure 22, the main predictors of their political correctness are ideology, politics, and activism, as shown in Figure 23. This suggests that the political skew of academia renders it less friendly to freedom of speech than other occupations when the competing value of emotional harm-avoidance comes into play.
In academia, two other characteristics are associated with increased support for political correctness: being a professor rather than a lecturer, and being female. Given the influence of professors in running universities, the former finding is especially concerning for those who fear for academic liberty. No significant difference was found by age, or by whether an academic is retired or active, however, once other factors had been taken into account.
Notice that age does not significantly predict attitudes to PC among British academics in Figure 23 above. This is in marked contrast to the picture among the general public in the US data in Figure 21. Figure 24 presents a sample of roughly 1,200 mainly non-academic PhD holders drawn from the UK YouGov panel. Here there is a big age gap: PhD holders under age 40 are over 30 points more inclined to support PC than PhD holders over 60.
But this age gradation does not exist among academics, especially SSH academics. According to Figure 25, there is an important divergence between British PhD holders who remain in academia and those who leave for other fields. The difference is especially noteworthy among academics currently teaching in the social sciences and humanities, where being older does not appear to correlate with diminishing enthusiasm for political correctness.
Therefore, while PhD holders under 40 inside and outside academia are fairly similar in their high support for PC – a finding that we’ll see also holds (on related measures) for current PhD students of all ages – there is a wide divergence between academic and non-academic PhDs over age 60. This points to possible academic socialization, especially within SSH fields, as a stimulant to retaining support for political correctness.
But there is another possibility, with wider implications. Namely, that the change is generational, and PhDs outside academia will henceforth remain PC as they age, altering the ideological composition of non-academic professions. On this reading, the high support for PC among older academics, especially in SSH fields, is because they are a left-modernist ideological vanguard in society that is leading where others will follow. If the latter interpretation is correct, this means that non-academic professional spheres are likely to become more PC in the years to come.
However, the age profile of left-wing frequent social media users is flatter than it is for PhD holders, indicating that ideological fervor counteracts the conservatism that is associated with being older (whether for life cycle or generational reasons). Among British non-academics who are frequent social media users, identify with the Labour Party, and are far leftist, there is no difference between young and old. Indeed, the over-60s in this group are as or more PC than their academic counterparts. This evidence suggests, in line with the model in Figure 23, that leftist consciousness, social media use, and gender are central, with age per se a smaller factor. Education level on its own appears to have only a modest impact, if any, on support for PC.26
Support for political correctness is grounded in empathy for historically disadvantaged groups, especially those based on race, gender, and sexuality. Yet, as Paul Bloom warns, empathy can be a flawed guide to morality.27 An effect of empathy for a particular person or group can be that others who deserve sympathy disappear from view. Worse, those who are viewed as harming the group with which one empathizes may cause one to become angrier than would be justified by a more wide-angle, evidence-based view.
This is illustrated by the fact that empathy for some identified victims can be used to rouse people to commit atrocities. As Bloom remarks, empathy towards white women was used to fan anti-black violence in the Jim Crow South, and feelings for the plight of the poor helped to mobilize anti-bourgeois genocide in communist regimes. In today’s context, empathy for minorities can lead to hostility towards whites or conservatives. Rather than see the whole and trust in empirical regularities to guide one’s anger and calibrate one’s response, powerful attachments and narratives take over, leading to flawed moral reasoning. Moreover, when empathy for a particular subset of disadvantaged groups and hostility to their purported oppressors becomes a leading priority, values other than cultural egalitarianism – notably freedom and reason – which universities have long proclaimed, may become endangered.
Decolonizing the Curriculum?
Returning to my academic surveys, I sought to probe faculty views on a less dramatic test of academic freedom than support for firing heretics:
Please imagine there was a new initiative in the Humanities, Arts and Social Sciences at your institution, stipulating that on each reading list, at least 30% of readings must come from women and 20% from authors of color.
Response categories were the same as in the diversity-organizational performance dismissal scenario, permitting us to gauge not only people’s views, but the extent to which academics censor their private beliefs:
a) Publicly express opposition (e.g. at a department meeting)
b) Privately oppose the initiative, but not say anything publicly
c) Neither support or oppose the initiative
d) Privately support the initiative, but not say anything publicly
e) Publicly express support (e.g. at a department meeting)
f) Don’t know
This kind of reading list campaign has occurred in a number of jurisdictions, with one Swedish academic, Erik Ringmar, complaining that he had to cancel an entire course on fascism due to his inability to find sufficient women writers on the subject to satisfy a quota of 40% females per reading list mandated by the University of Lund’s political science department.28
Figure 26 summarizes the results across all surveys, focusing only on SSH subjects. It highlights that this question led more people to jump off the fence, with the share of neutral and don’t know answers, 16-26%, much lower than in the dismissal scenario. This is 2 to 3 times less uncertainty than for the question of whether to dismiss an academic who argues that diversity impairs organizational performance.
The second major aspect to note with this question is that a near majority of SSH academics support the decolonization agenda across all jurisdictions: between 44 and 48% of academics in the first four surveys back the idea of racial and gender reading list quotas. In all surveys, this comfortably exceeds the share who oppose quotas though there is significant opposition: between 29 and 34% of academics, depending on the survey.
Social science and humanities PhD students in both Britain and North America are especially enthusiastic about quotas, with 61% of British SSH PhDs and 70% of their North American counterparts in support. This tallies with the views of the 1,216 PhD holders under 35 in the YouGov non-academic sample, who leaned 67-20% in favor of political correctness. While these figures might be expected to dip somewhat when PhD students enter academia, have to make up their syllabi, and find their intellectual wings clipped, it is a worrying sign of how far the balance between social justice and academic freedom has tipped in the direction of the former among the youngest cohorts entering the academy.
Respondents who backed or opposed the reading list quotas tended again to say they would do so publicly rather than keep their beliefs to themselves. Naturally, a portion of those who claim they would stand publicly may fail to do so in the breach, but Figure 26 does not support a Kuran-Sunstein picture of mass conformity to social justice claims concealing majority support for academic freedom.
I also have data from my survey of National Association of Scholars members, where 91% of the 227 respondents opposed the decolonization initiative and just 3% backed it. Among the opponents, 61% – two-thirds – said they would publicly sign a counter-petition while 30% said they would keep their views private.
However, this doesn’t mean there is no concealment. Compared to the dismissal scenario, where there is less of a clear pattern of self-censorship among opponents, here there is more evidence of a chilling effect on those opposing the decolonization agenda. Figure 27 indicates that supporters of reading list quotas are far more likely than their academic freedom opponents to publicly state their views rather than keep them private. Among academics, supporters are between 4 (315%) and 8 (693%) times more likely to voice their opinion publicly than privately. By contrast, opponents of decolonization are much less willing to go public, being just 26-174% more likely to speak up than stay silent.
This tips over into a preponderance of silence when we shift from academics to PhD students, where opponents are between 33 and 58% more likely to stay silent than speak up – reflecting the heavy dominance of pro-decolonization sentiment among graduate students.
In particular, it is noticeable that among the 14-20% minority of PhD students who oppose decolonization quotas, there is more reticence to come out publicly than is the case among professors who oppose quotas. Among PhDs, a majority of opponents in Figure 28 say they would choose to keep their beliefs to themselves rather than voice opposition publicly. This could be related to the greater support for hard authoritarianism among PhDs compared to academics. This comports with a 2020 FIRE survey that finds that conservative students tend to self-censor more at relatively liberal universities.29
The final point to note is that in this question, across almost all surveys, opinion among academics under 40 generally contained 5-10 points fewer neutral responses than among the over-40s, possibly indicating that there are more settled views of this question among younger staff, in the direction of both diversity quotas and, to a lesser degree, academic freedom. This age-pattern was not apparent in the previous question on organizational performance, however, so it may not hold more generally.
To probe the malleability of this group of undecided scholars, I asked respondents:
If you had to choose, which of these do you think is more important:
a) that course content should be inclusive, representing the racial and gender makeup of the students.
b) that course content should feature the most intellectually foundational books and articles in the field.
c) Don’t know.
Results are plotted in Figure 29. These show that academics back content over quotas in both the UK and North America. North American faculty and, especially, graduate students are relatively more favorable to quotas over foundational content compared to British faculty, who clearly privilege content over quotas. Both surveys of UK social science and humanities academics show that they prioritize intellectual merit over quotas by 18-25 points. British PhD students break 50-50 between the two options. On the other hand, North American PhD students incline 15 points more toward quotas over foundational texts.
Note the difference between these numbers and those in the similar question in Figure 26 above. This time, I introduce a more explicit trade-off between quotas and intellectual merit (“foundational texts”) than was presented in the previous question asking people to merely support or oppose an initiative “stipulating that…at least 30% of readings must come from women and 20% from authors of color.” Among North American academics, support for quotas has slipped 6 points in the second version (45% supported quotas in the first question, falling to 39% support in the second).
In Britain, mention of intellectual rigor and the key value of education seems to have swayed academics even more than in North America, lowering support from 44% (YouGov) and 51% (mailout) in the case of support for the diversity initiative to 32% (YouGov) and 34% (mailout) on the question of quotas versus foundational texts. So too among PhD students: UK PhDs reduce support for the quota option from 61% in Figure 26 to 46% in Figure 29 while North American PhD students lower their support from 70% to 55%.
Diversity Statement Experiment
In order to examine the firmness of these beliefs, I conducted a survey experiment in each of the five surveys. Before answering the question, half the sample read an equality and diversity statement modeled on that of the University of California:30
Please imagine an organization is committed to the full realization of its historic promise to recognise and nurture merit, talent, and achievement by supporting diversity and equal opportunity in its education, services, and administration, as well as research and creative activity. This organization particularly acknowledges the acute need to remove barriers to the recruitment, retention, and advancement of talented students, faculty, and staff from historically excluded populations who are currently underrepresented. Many personally identify with underrepresented groups in the curricula (that is, the stories, histories, findings and research of women and BME members, among others) and the organization is determined to serve them through an education aligned with social justice and inclusion.
In the British YouGov sample, among those who didn’t read a statement, just 25% supported the view that course content should represent the racial and gender makeup of students instead of the intellectually foundational texts in a field. This jumped to 39% among those who read the UC diversity statement. The effect seemed to register across most academics, but not among Brexit supporters, as Figure 30 shows. Clearly many British academics have values that render them open to a Social Justice appeal.
The typical academic who voted to remain in the European Union in 2016 (the vast majority of academics) shifted 16 points after reading the diversity statement, from 27% to 43% support. By contrast, the minority who voted to Leave the EU barely budged, from 13% among those reading nothing to just 15% among those reading the diversity statement, showing no statistically significant difference.
Another important factor affecting a person’s view on course content is support for political correctness. PC supporters were, as one would expect, far more supportive of quotas than opponents of PC. In addition, as Figure 31 shows, the greater an academic’s support for PC, the more susceptible they are to being convinced of quotas after reading the diversity statement. Here it seems that institutional initiatives around equity and diversity, like diversity statements, strike fertile soil among the pro-PC majority of academics: recall that three-quarters of current social sciences and humanities academics favor the concept to help combat discrimination, even if it stifles free speech.
Looking across the wider array of results in Figure 32 unearths two patterns. First, culturally conservative groups, notably Brexiteers (known as “Leavers”) and those favoring immigration restriction, have low support for quotas and are not affected by reading the diversity statement. Second, groups that already have high support for quotas – young, women, the far left – shift less. It seems that a broad receptive group in the middle – center-left, Remain, men, pro-PC – move most toward backing quotas.
The first point to note is that 40-45% of younger academics (under 40) oppose racial and gender quotas for reading lists. While younger academics are more supportive of the “decolonization” agenda, a near majority are not, and their lack of movement after reading the diversity statement suggests there is an important group of young faculty who are aware of diversity claims, but reject them when they conflict with academic freedom.
That said, the survey experiment shows that there is a flexible component of academic opinion in Britain, just as there is with the student opinion I analyzed in a previous co-authored report, which I reprise in Figure 34. Yet these results only show that academic opinion can be swayed by a social justice/diversity appeal, which is less institutionalized in Britain than North America. Might an appeal to free speech shift this group away from the pro-decolonization position?
Academic Freedom Statement Experiment
The diversity statement had a profound effect on British academic opinion, suggesting that opinion in the academic center is malleable on this question. Might it be shifted in the other direction, away from quotas? The answer, it appears, is no.
In order to test whether academic freedom arguments might work, I introduced a new statement group, in addition to the diversity statement group and the control group who read nothing:
Throughout the ages, those who have expressed unorthodox opinions – on religion, on the government, on homosexuality, socialism or capitalism – have been subject to persecution or censorship. Britain [America] has an especially important history of resisting threats to freedom of expression from the authorities. Sometimes these threats take the form of being killed or jailed, while at other times they involve being turned into a social pariah, losing one’s livelihood or being subject to reputational shaming. Many have died for our right to freely speak our minds and exchange ideas, our precious inheritance. Liberalism is not easy because it is about tolerating ideas we don’t like, which is why it is rare in history and in much of the world. The role of the university is to permit a wide variety of opinions to be expressed and debated, even if they challenge social convention or offend people’s sensibilities. The quest to teach and research the highest truth must prevail over other priorities. Tolerating only views we agree with is contrary to the spirit of free speech.
Figure 33 summarizes the impact of the two treatment statements on my mailout sample of UK SSH academics. The results reveal that opposition to decolonization is similar among both those who read nothing (57%) and those who read the free speech passage above (60%). This shows that reading the free speech vignette has no impact on academics’ value priorities when academic freedom and diversity are in tension. However, those who read the diversity statement were nearly 20 points more likely to favor quotas – replicating findings in Figure 32 from the YouGov data. The “don’t know” responses are similar, indicating that many changed their view from the “foundational texts” position to the quotas position after reading the diversity statement.
For the North American academic and PhD surveys, and for British PhDs, neither the diversity nor free speech experimental passages made any significant difference to their responses to the quotas-versus-merit question. This suggests that diversity-based quota logic is already well-established among North American scholars of all ages and younger cohorts in British academia, with views fixed in both directions. Thus new messaging on academic freedom or free speech to PhD students and faculty may have little impact on their trade-off function between social justice and free speech.
The lack of responsiveness to the free speech message across all academic and PhD surveys stands in contrast to that of British undergraduates, who are more biddable to the free speech side. In previous work, I found that the free speech passage above had a significant impact on undergraduate student opinion on the question:
When in doubt, which policy should your university support?
a) prioritize free speech, even if this makes people upset;
b) prioritize emotional safety, even if this limits free speech;
c) don’t know.
Figure 34 reproduces that work, showing that 63% of undergraduate students who read the free speech passage backed the free speech option but only 49% of those who read nothing did so, and just 41% backed free speech among those who read a paragraph about the need for emotional safety for disadvantaged groups. Thus the free speech treatment moved opinion 14 points (49 to 63%). The emotional safety passage had a similar impact the other way, shifting views in the direction of emotional safety by 14 points (34 to 48%). Female undergraduates shifted more than male undergraduates in response to the emotional safety passage while both genders were equally moved by the free speech appeal.
It appears that British academics and PhD students have relatively fixed views. British academics are amenable only to a diversity-based appeal for restrictions on academic freedom, not to free speech appeals. North American academics and PhDs have settled views on the matter so are resistant to either message. Advocating for free speech is therefore more likely to bear fruit if directed at undergraduates. By the time students have been socialized into the diversity and emotional safety-oriented academic culture at the PhD level, it may be too late.
Age, Gender and Support for Academic Freedom
We have seen that views on the decolonization versus academic freedom question have crystallized in North America and among British PhDs, and are relatively impervious to competing perspectives. This suggests stronger divisions on this question could prevail going forward. The question this then begs is whether opinion is shifting away from merit toward quotas due to generational turnover. In order to explore what may be coming, I parse the views of faculty and doctoral candidates by age.
In general, younger faculty lean more toward quotas than older faculty. Figure 35 shows how academics under 40 are, depending on the survey sample, 4-23 points more likely to back the reading list quota initiative than those over 40, while PhDs under 30 are 10-29 points more likely to do so than PhDs over 30. The age differences are somewhat larger among North American than British academics and doctoral students.
When the question explicitly juxtaposes “intellectually foundational texts” against reading list quotas (see Figure 29), support for quotas falls across the board, but as Figure 36 demonstrates, the age gap remains – though only in the North American samples. This may point to the greater penetration of curriculum quota activism in North America, which has roots in the 1960s and with the multicultural education movement of the late 80s and early 90s. Only recently has there been a major push on this front in Britain.31
Gender is also a consistent predictor of attitudes to diversity/safety versus freedom questions. Figure 37 charts a wide gender divide that ranges from 12 to 38 points but is greater than 23 points in all but one survey. This echoes findings in student surveys, which also reveal consistent gender disparities on free speech versus emotional safety questions.32 Having said this, an important group of between 24% and 34% of female academics and graduate students favor intellectually-foundational texts as the basis for course reading lists rather than diversity quotas.
As new generations of PhD students enter the academy and, if the share of women rises, we should expect the balance of internal opinion to move in the direction of emotional safety over academic freedom. In addition to limiting the freedom of academics to set the course texts they believe to be most pertinent for teaching, further moves in the direction of emotional safety are likely to mean that research on controversial topics around race, gender, and sexuality may become increasingly off limits. This circumscribes and distorts knowledge produced by the university, reducing its value.33
Attitudes toward Nonconformity
If a new initiative mandates that people meet diversity quotas for reading lists, what is the punishment for those who refuse to yield their academic freedom to these rules? While one may see diversity as a positive goal, are those in favor willing to own the authoritarian implications of such a policy for those who refuse to comply? I put the question to my sample of over a thousand American and Canadian academics. Table 2 lists, among those in favor of quotas, their preferred sanctions, from least to most punitive, for academics who refuse to comply with the quota requirement:
The appetite to punish severely is not as strong as the desire for quotas. Only 3% recommended firing dissenters, though 19% supported formal disincentives from dismissal to being allocated less favorable teaching, research, and administrative positions. Another 27% favored social pressure while a further 28% backed mandatory implicit bias awareness training. These punishments are the mirror image of the complaints outlined in Figure 6, which found that 38% of right-wing PhD students in North America and 43% of the mainly right-leaning National Association of Scholars (NAS) members experienced formal or informal sanctions for expressing their views in research, public fora, and teaching. These showed that the most prominent form of punishment experienced by nonconformists was social bullying, followed by adverse teaching or administrative roles or reduced research funding.
The strongest proponents of severe punishment were academics under 40, with 11% of them advocating forcing dissenters to cancel their course and teach a quota-compliant module. By contrast only 2% of academics over 50 selected this option. Those who identified as activists and “very left” were similar in their inclination to punish, and being female had an effect only slightly weaker than age, activism, and ideology.
Yet much of this is an artifact of these groups’ stronger support for quotas over intellectual merit. When limiting the analysis only to those in favor of quotas, the relative authoritarianism of the under-30s, and lack of it among the over-65s, is what jumps out of the models. Ideology and being female no longer predict authoritarianism, though self-described activists are somewhat more inclined to punish than others (this effect was just outside the boundary of conventional statistical significance).
What this question does not ask is whether people would be willing to publicly object if the university decided to dismiss those who resist, or to punish them severely. We saw that for controversial research, about half of academics opposed dismissal campaigns, and for less controversial views – such as favoring reduced immigration – around 8 in 10 did. A little over half the opponents said they would be willing to publicly express their opposition. Assuming that a recalcitrant academic who does not wish to comply with mandatory reading list quotas attracts responses intermediate between these positions, this leads me to estimate that close to half of academics would oppose a dissenter being terminated, but that little more than a quarter would say so publicly. This would possibly clear a path for administrators to adopt a hard authoritarian position against those who refused to comply with a reading list quota directive.
Academic Freedom or Social Justice?
A related but distinct question is whether the tendency to favor quotas over intellectually foundational texts, or to back new diversity quota initiatives, reflects a general preference for social justice over the freedom to seek the truth, wherever it lies. Here the question is:
When it comes to tension between the freedom for academics to publish research on their interests and concerns over social justice for disadvantaged groups, which comes closest to your view?
a) I prioritise social justice, and have strong beliefs in this area
b) I prioritise social justice, but don’t have strong beliefs in this area
c) I have thought about it, but don’t have a strong view either way
d) I haven’t thought about it much, and don't have a strong view either way
e) I prioritise academic freedom, but don’t have strong beliefs in this area
f) I prioritise academic freedom, and have strong beliefs in this area
When phrased in these more intellectualized and abstract terms, the balance of answers shifts away from the social justice position towards academic freedom. Whereas American and Canadian academics favored foundational texts over quotas 47-39, Figure 38 reveals that they prioritize academic freedom over social justice by a whopping 56-27 margin (58-26 in the US). Even among North American PhDs, a 15-point lead for the quota position has eroded to just a 6-point lead for the social justice view in this formulation. UK PhDs move less: from an even position between foundational texts and quotas to +4 for academic freedom. The difference can likely be attributed to both the intellectual appeal of the term “academic freedom” and the more abstract nature of the question. Elsewhere I find that more concrete trade-offs involving actual people, policies, or events tend to be resolved more in favor of the social justice position.34
It is also noteworthy that North American PhD students favor social justice over academic freedom by 6 points (9 points among SSH PhDs), and that social justice is nearly on par with academic freedom among British PhDs. This indicates that support for academic freedom may be less robust among the emerging generation of faculty.
The correlates of these positions are similar to those in the quota versus merit models, with younger scholars, those on the far left, activists, and women more inclined toward the social justice position. Being a racial minority is not statistically significant, bearing out a pattern whereby race is a less important correlate of views on academic freedom questions than age, gender, or ideology.
Figure 39 shows how gender and ideology interact to condition attitudes to academic freedom or social justice, with women increasingly diverging from men toward social justice as we move left on the spectrum, but with more limited gender disparities on the right. Those of “other” gender (29 in dataset) also tend to favor the social justice option over academic freedom. Crucially, men – even far-left men – tend to favor academic freedom over social justice.
In terms of statistical effect sizes, identifying as an activist had the strongest effect on prioritizing social justice, followed closely by left-right ideology, gender, and age. American academics were marginally more likely to favor academic freedom than Canadian scholars.
Not Against Academic Freedom, But Not For It, Either
The upshot of our analysis of the academic freedom vs. social justice questions is that most academics are cross-pressured between their commitments to cultural progressivism and their attachment to academic liberty and reason. On a positive note, just 20-25% of SSH professors and lecturers, rising to 36-45% of SSH PhD students in North America and Britain, would support at least one of four authoritarian measures against a dissenting conservative academic. This still leaves three-quarters of SSH academics as consistent opponents of drives to remove professors from their posts for politically incorrect research. Far leftists are considerably more likely to back illiberalism, but even here, the far-left faculty divides fairly evenly between supporters and opponents of dismissal even when using my expansive measure of support for firing (backing any one of four hypothetical campaigns). In any given campaign, the majority of far-left academics will oppose dismissal.
Fear plays a part for an important minority of pro-free speech academics, inclining them to remain silent rather than advocate publicly for freedom, but sincere belief and not fear is the main obstacle to tackling today’s illiberal campus climate. Activists and administrators are the instigators of authoritarian policies, but in the absence of the climate of opinion prevailing within the faculty, where half are hesitant about protecting academic freedom when it collides with progressive aims, their efforts would meet more resistance.
The fact that many academics do not publicly nail their colors to the academic freedom mast due to their beliefs is not a good sign. This is especially so if younger academics do not become more inclined to defend academic freedom as they ascend the academic ladder. The rising share of women in academia may also affect active support for academic freedom due to their somewhat higher predilection for diversity arguments – even though women are not more likely to endorse authoritarian punishments than men who share the same belief system.
Most academics and PhD students in the United States, Canada, and Britain don’t actively oppose academic freedom, but they don’t actively support it either. Most are unlikely to speak up for colleagues who dissent from social justice-inspired restrictions on their academic freedom. Thus progressive authoritarianism succeeds through a kind of sin of omission. This indicates that intervention would need to occur at the institutional level, to ensure activist-driven proposals cannot become university policy. At present, college authorities are generally silent on academic freedom or subordinate it to the countervailing imperatives of equity, protection from harassment, diversity, and college reputation.
Currently, the momentum inside the university lies largely with those who would abridge academic freedom and free speech in response to equity and diversity claims on behalf of disadvantaged groups. This is well-established in the American and Canadian cases.35 But there is also a rising volume of diversity initiatives at British universities, spurred on by the Race Equality Charter (REC) of Advance HE and ATHENA SWAN, influential charities that get universities to sign up for their monitoring and compliance pledges, building these into their research assessment scores.36 Both Universities UK (UUK), the industry body for universities, and the major sector union, the UCU, are strongly promoting this agenda. This gives internal diversity proponents within the professoriate a metrics-based rationale to move in the direction of mandatory targets or even quotas. In North America, such initiatives are even more entrenched as a combination of staff and student activists and better-funded (compared to Britain) equity and diversity administrators enforce compliance. These have truly become what Jonathan Haidt and John Ellis term “social justice” rather than “truth” universities.37
While Cambridge University’s governing body famously rejected, by an 80-20 margin, a proposed policy shift that would have mandated “respect” for identities over toleration of difference, the vote involved just 1 in 8 academics at the university, and included retired academics and some non-academic staff. While my survey results intimate that the vote would still have prevailed had all staff been polled, they also suggest that the final vote would have been closer.38
These findings also indicate that rebalancing internal messaging from equity-diversity toward academic freedom would have a significant impact on attitudes among undergraduate students, and – to the extent that messaging around emotional safety for protected groups is reduced – also among UK professors and lecturers. Where views have largely crystallized, as among North American academics and PhDs, and among UK PhDs, messaging is less likely to persuade.
Government policy can play an important role in conveying the will of the democratic majority and the letter of the law (at least among institutions in receipt of public funds), while issuing guidance to ensure that university administrators (though not academic staff) act in a politically-neutral manner. This intervention could also help set the tone at universities based on what society and the law – as well as most academics – expect. As Cass Sunstein notes, law often signals the public mood and can dispel misperceptions about the approved values of an organization.39
Part II: Soft Authoritarianism
Most academics reject hard authoritarianism. But the verdict is less clear for soft authoritarianism. As noted in Part I, faculty matter because they play a cardinal role in setting the culture and environment of departments on a day-to-day basis. Whether permanent, or on time-limited contracts as researchers and/or teachers, they do so by shaping how research and teaching are conducted, and by being responsible for the variety of decisions in which, for instance, papers are accepted for publication; grants are awarded or not; job offers and promotions are received; and students are admitted. More generally, faculty set the tone.
Taken individually, very few actions by individual faculty members are of decisive significance. What is of interest and significance are larger patterns, which help shape the culture of the university and ultimately influence the public contribution it makes. To what extent are academics free to pursue research or public engagement that may have a political interpretation, without paying a social or professional penalty?
This part of the report examines political discrimination against conservative and gender-critical academics and its corollary, the chilling effects produced within these target groups. In line with the “iceberg” models presented earlier, it pursues the question through a focus on both the perpetrators and victims of discrimination. The former involves half or more of staff and students at the university while the latter is narrowly concentrated on conservative and gender-critical academics and students.
Targeted scholars either experience direct victimization or feel chilling effects by anticipating that if their views were to become known to colleagues or the wider scholarly community, it would damage their careers. This is especially pointed in view of the scarcity of academic positions, with hundreds of applications for permanent posts. Hyper-specialization results in limited mobility – especially if a researcher wants to live in a given geographic area or work at a high-status university. Meanwhile, reputations travel quickly through the “invisible colleges” that affect hiring choices, making it very difficult for academics to move jobs. This produces even greater risk. As a result, dissenters prudently adapt to this manifestation of John Stuart Mill’s “despotism of custom” through self-censorship, limiting their academic freedom, and constraining the truth-seeking mission of the university.
Prior Work
In recent years, a growing body of work has begun to examine the possibility of discrimination on ideological or political lines. “Political” or “ideological” discrimination involves one’s professional judgment about another person or their work being affected by the extent to which they agree with the political or ideological orientation of the person.
This becomes even more urgent in light of the growing political polarization of the 2000s, initially in the US, but increasingly in Canada and the UK as well.40 Indeed, political polarization is connected with political discrimination. For instance, in 2014, Shanto Iyengar and Sean Westwood found that affective polarization – hostile feelings towards opposing political partisans – was far more pronounced than negative feelings based on race. Further, affective polarization impacts discriminatory behavior to a significantly greater degree than racial prejudice.41
While political discrimination could occur in any sector of society, it is of particular interest within academia, where political viewpoint may directly shape or affect the content of someone’s work. Accordingly, some studies focus on political discrimination within academia specifically. Key contributions to this literature include studies by George Yancey,42 Yoel Inbar and Joris Lammers,43 Nathan Honeycutt and Laura Freberg,44 and Uwe Peters and co-authors.45
Yancey found that 30% of his sample would discriminate against a Republican while 50-60% of anthropologists and English professors said they would be less likely to hire an evangelical Christian. The latter three contributions adopted a similar question wording so can be readily compared. Each used an online mailout survey to gauge the attitudes and opinions of a given sample. Inbar and Lammers sought to evaluate the attitudes of psychologists in the US, and focused on the attitudes of and towards political conservatives (n = 508). Honeycutt and Freberg extended that study, using a sample of academics in California across disciplines, repeating these questions for political liberals in addition to conservatives (n = 618). Peters et al. used an international sample of philosophers, and asked about a more variegated set of political orientations and viewpoints, rather than a simple left-right division (n = 794). These studies confirmed, for their respective samples, the same central finding as that of Sam Abrams (see Figure 40 below), namely that the American professoriate has an overwhelmingly left-liberal orientation and a paucity of political conservatives relative to the general population. For instance, in the Peters study, 75% of philosophers leaned left, 11% were moderate, and 14% leaned right.
On the willingness to discriminate politically, the findings were dramatic. In Inbar and Lammers’ study, over 1 in 3 of those who identified as liberal would discriminate against conservatives in hiring decisions, while 1 in 4 would discriminate against them in reviewing their grant applications.46 Honeycutt and Freberg replicated this result nearly exactly, but showed that the willingness to discriminate was symmetrical on both political sides. In their sample, 33% of liberals were willing to discriminate against conservatives in hiring decisions (answers varying from “somewhat” to “very much”), which was equally matched by conservatives’ willingness to discriminate against liberals, at 32%.47 This study also found that the more ideologically committed a respondent was, the more likely they were to be willing to discriminate.48 From the Peters study, the highest levels of willingness to discriminate were again reported for hiring decisions. For a right-leaning hire, over 55% of left-leaning philosophers reported being willing to discriminate “occasionally” to “all the time.” For a left-leaning hire, the equivalent proportion of right-leaning philosophers willing to discriminate was over 45%.49
A larger sample was collected in 2017 by Terrence Karran and Lucy Mallinson.50 These researchers surveyed 2340 academic members of the University and College Union (UCU), the main union for academic staff in Britain and an even larger complement of European scholars. The UCU survey focused on threats to academic freedom, germane to my study here, and principally on internal threats to academic freedom arising from managerial favoritism, unspecified forms of discrimination such as methodological tribalism, and imperatives flowing from research and teaching assessment exercises.
Qualitative comments gave insight into a series of “invisible” threats to academic freedom that are rarely identified. The most widespread was psychological bullying from colleagues, reported by 27% of respondents. There were also reports of disciplinary actions from department heads or managers. Such individuals, often senior staff in rotating appointed positions at department or School/Faculty level, hold considerable power over individual careers. Survey respondents mentioned cases of threatened or actual dismissal from employment, increasing an academic’s teaching or administrative load, removing them from a cherished teaching or administrative position, berating them during performance reviews, and withdrawing grant funding. Other instances I am aware of include departmental or higher-level decisions compelling conservative speakers to be balanced by progressives (a requirement not imposed on left-only panels), or repeatedly delaying an event without cancelling it.
These studies have been vital for highlighting dynamics within universities in terms of both hard and soft authoritarianism. Of course, each of these studies depended on participants opting in to take the survey. This creates a potential risk of self-selection bias. Among the surveys presented in this report, that risk also applies to the North American academic survey, the NAS academic survey, and the UK mailout academic survey, but not to the UK YouGov academic survey or the Prolific surveys of American, Canadian, and British PhD students. By comparing these samples, we are able to show that the online mailout samples do not differ systematically from the accidental, non-selective surveys. For more information on survey data and methods, see the Appendix.
In addition, previous surveys rely on participants’ willingness to report their attitudes in a way that is susceptible to social desirability bias. This report is based on a series of studies that address this through the use of a concealed list technique that can reveal the true willingness of academics to engage in political discrimination.
Do academics feel comfortable expressing their views on politically-salient or controversial issues, and have they decided not to express their views in research or teaching because of the risk of penalties? Reported chilling effects are not, in themselves, direct evidence of a problem. Someone’s fear of penalties may be irrational, and should thus be handled through personal resilience rather than external protection.
To address this, I consider a suite of attitudes that affect the social and professional context in which academics work. A key aspect of this is the degree to which the perceived political import of their work may affect their career. To assess such political discrimination, I focus on attitudes towards grant applications, papers submitted for publication, and applications for promotion. Another is the softer but arguably no less consequential issue of workplace social climate. Are academics comfortable interacting in an informal context with colleagues who hold politically-opposing or controversial views?
From Hard to Soft Authoritarianism
Hard authoritarianism, being fired or threatened for one’s views, is a more serious violation of academic freedom than soft authoritarianism, not being hired, promoted, awarded a grant, or published in a journal. But both matter for academic freedom. Active social bullying is more punishing than social ostracism, which is in turn worse than socially avoiding someone or not including them in one’s social circle. Even so, there is a sliding scale from hard forms of authoritarianism to softer “everyday” exclusions that can affect a large number of staff. All forms of political sanction matter, however, and ideally would be kept to a minimum given the importance of collegiality to the academic enterprise.
Part I examined both perpetrators’ willingness to endorse hard authoritarianism and victims’ experiences at the sharp end of such measures. Part II considers soft authoritarianism: political discrimination and its impact, in the form of chilling effects, on nonconforming academics. Here Mill’s “despotism of custom” rather than activist bullying or university disciplinary power is what ensures compliance.
This doesn’t mean people should not be allowed to discriminate in their social interactions. There is an important distinction to be drawn between the negative duty not to exclude, i.e., “I won’t associate with conservatives,” which should be unacceptable, and the positive desire to socialize more often with those who share common beliefs or identity, i.e., “I like to associate with neo-Marxists.” Limiting discrimination is not coterminous with positively mandating patterns of socializing, which is contrary to the spirit of free association that is central to liberalism. Something similar holds for inviting speakers to campus, where staff or student interests may organically result in invitations skewed to the left. This is quite different from actively rejecting conservative speakers by stretching the definition of racism to encompass their views and thereby justify excluding them from campus.
In addressing the question of academic freedom, I begin from the standpoint of the victims who experience discrimination or anticipate that they will be penalized for their views. Academics’ decisions to restrict what they teach or research, on the basis of concern about colleagues’ potential discriminatory reactions, is what is often termed a “chilling effect.” This is at the heart of soft authoritarianism, as opposed to hard authoritarianism, which relies on threats of disciplinary action or dismissal.
Is there a rational basis for scholars – notably conservatives or gender-critical feminists – to feel a chilling effect? To answer this question, there is a need to focus on perpetrator attitudes, namely those of the leftist and (to a lesser extent) centrist academic majority toward the conservative minority. If these views are indeed penalizing, then perceptions of a chilling effect are based in reality, and self-censorship becomes a reasonable response to the threat of genuine antipathy and unequal treatment.
The degree of faculty support for dismissing academics who take an unpopular position on a sensitive issue is one barometer of whether those who perceive a chilling effect are correct to do so. This connects hard authoritarianism to the soft authoritarianism of public opinion. The perpetrator test here is whether academics back the idea that someone should lose their job or post, or be subject to an online campaign, for research that is deemed to have contravened a moral norm.
Political Discrimination
Soft authoritarianism arises from political discrimination. That is, the willingness of perpetrators to penalize someone for their perceived politics. Short of dismissal, this may involve discrimination in hiring, promotion, and refereeing; or social ostracism in collegial spaces. The surveys below test for these forms of discrimination. Though short of dismissal, these less dramatic forms of discrimination may nevertheless prove extremely important in shaping the perceptions of those who might otherwise have chosen to pursue their genuine research interests. In this manner, speech is subject to soft authoritarianism.
Second, I consider the problem from the other end: the chilling effects reported by the victim side. Here I measure how academics perceive their scholarly environment. This sets out what the perceptions are, and whether they correspond to actual risks.
This furnishes two foci for the work on soft authoritarianism that follows. First, the willingness to engage in political discrimination; and second, the experiences of those who report being subject to (or fear being subject to) viewpoint-based penalties. Both are connected, as we shall see, to the low ideological diversity of the professoriate.
Section IIa: The Ideological Evolution of Academia
The ideological composition of the professoriate is germane to this conversation. For, as we shall see, when two sides discriminate against each other, there may be parity in discriminatory intent, but a structural imbalance in discriminatory effects because there are more perpetrators on one side than the other. This is very much the dynamic in academia, where there is a pronounced ideological imbalance among staff.
Consistent with prior research, I find that a significant majority of academics incline left in terms of political outlook. Academics’ attitudes on political subjects such as Trump, Brexit, abortion, or immigration are very distant, in aggregate, from those of the average voter. This is not simply because the “ivory tower” is somehow an outlier. Those with advanced degrees in all sectors hold views intermediate between professors and the general public. Interestingly, as we shall see in the non-academic part of this study, academics as individuals seem to discriminate on political grounds at similar rates to non-academic professionals with analogous levels of education.
Yet academia is different for two reasons. First, the scale of the left-liberal majority is more exaggerated than in most other professional settings. Second, colleagues’ political views are more visible in their work – at least among social science and humanities scholars – in universities. The structural effect of having so few conservatives and a progressive majority, in combination with the difficulty in hiding one’s views, results in a high level of net discrimination against non-progressives.
It is vital to emphasize that, on my evidence, a narrow majority of lecturers and professors in the United States, Britain, and Canada would not discriminate on political grounds in any way. This includes a slight majority of those on the left. In each country surveyed, for instance, over half of left-wing academics would not discriminate against a Trump or Brexit voter in a job application.
The Political Views of Academics
There have been several studies of the political leanings of academics in the United States and Britain since the 1960s. In 1969, a Carnegie Commission survey found that 28% of American university faculty were conservative, 27% centrist, and 45% liberal or left. Thus the left:right ratio was below 2:1. While Sociology leaned left by a 4:1 margin, the ratio in the Humanities and Law was 2:1 and in the Sciences 1:1.51 In 1984, the Carnegie data showed that just 39% of faculty were left or liberal, and the conservative share had risen to 34%, a balance approaching 1:1. The left share among social science and humanities academics was high but appeared to have declined from 66% in 1969 to 56% by 1984. This finding led Hamilton and Hargens to claim that “the incidence of leftism [among faculty] has been considerably exaggerated.”52 In 1989 and 1997, the Carnegie surveys showed a leftward shift to a 2:1 ratio compared to 1984. This was confirmed by 1989 and 2001 HERI studies.53 By the mid-2000s, however, this skepticism was difficult to sustain. A study by Gross and Simmons (2007) showed that Democratic voters outnumbered Republicans in the professoriate 3.5 to 1, with an 8:1 ratio in the social sciences and humanities, the same tilt uncovered by two other studies at the time.54
Likewise, among American social psychologists in 1960, the ratio of liberals to conservatives was about 2 to 1. But this began to change in the 1980s, and by 2014, the ratio of liberals to conservatives in social psychology had reached 15 to 1.55 Thus the initial decline in the number of conservatives seems to have taken place between the early 1980s and early 2000s, a period when the share of conservatives to liberals in the population remained relatively unchanged.
More recent trends, between the mid-90s and early 2010s, point to a continuing decline in the share of conservatives among college faculty combined with a new decline of the center and rise of the left. Using triennial surveys of tens of thousands of professors from the Higher Education Research Institute (HERI) at UCLA, Sam Abrams found that the share identifying as left-wing increased from around 40% in the mid-1990s to 60% by 2010-11.56 This occurred during a period when the political leanings of the US public remained roughly steady at around 25-30% left, thus wider societal shifts cannot account for the change among the professoriate. Figure 40 shows the trend.
More recently, groundbreaking work by Mitchell Langbert used a comprehensive representative sample of tenure-track professors from leading research universities (over 7,000) and liberal arts colleges (over 9,000). Tracking the political registrations of professors, he found that, excluding military and religious colleges, the ratio of Democrats to Republicans was 11.5:1 in the SSH departments of major research universities and 13:1 in a sample of 66 liberal arts colleges. In terms of academic disciplines, Langbert uncovered a staggering ratio in a number of social science and humanities fields (see Figure 41). In liberal arts colleges, in Anthropology, Communications, and the “Studies” fields (race, gender, sexuality) he found a perfect monoculture, with no registered Republicans at all.57 Looking across over 7,000 permanent faculty in Economics, Journalism, History, Law, and Psychology at major research universities, Langbert and colleagues found just 4.3% to be Republican, compared to 50% Democrats, 16% unaffiliated, and 29% unregistered. Moreover, other studies find that while academic Democrats are solidly left-leaning across all policy issues, Republican professors tend to be more ideologically diverse.58
Developments in Britain broadly parallel those in the United States. In 1964, 35% of UK academics voted for the Conservatives and 47% for the left-wing Labour Party, a modest difference. Support for the moderate Liberal Democrats or SDP rose substantially, reaching 35% by 1989, during Thatcher’s tenure in office, and squeezing support for the Tories among the professoriate to just 18%.59 The Liberal Democrat rise had less impact on left-wing Labour and the Greens, whose combined share was 44%.
By 2015, however, Chris Hanretty, using randomly-sampled Understanding Society data, found that closeness to left-wing parties (Labour and Green) had reached 68%, with closeness to the Conservatives down to 15%. Figure 42 seems to show – even as I cannot definitely prove this given shifts in question wording and methodology – that the same shift from centrism to the left that occurred in America between the early 1990s and 2000s also took place in the UK.60 As in the US, we see a first phase of conservative decline followed by a second phase in which the left expanded at the expense of the center.
To triangulate more recent data, I provide a compilation of recent surveys of British academic opinion in Figure 43. This includes self-selected Times Higher (THES) data, where readers offer their views, randomly-sampled data from Understanding Society, and my UK YouGov survey, where a passive sample is approached by a survey firm, ensuring less bias toward those who like to venture their opinion and may therefore not represent the median academic. Unweighted results from the YouGov UK 2020 survey show academics’ 2019 and 2017 (recalled) vote.
My results are broadly in line with Understanding Society’s smaller-sized random sample, as well as the Times Higher’s opt-in surveys, increasing confidence in the results. These show that about half of academics vote Labour and between 10 and 20% favor the Conservatives or UKIP/Brexit Party. Restricting to current academics and using survey weights, the figures show that 16% of my sample voted Tory or Brexit Party in 2019.
Using my data, and comparing across the US, UK, and Canada, focusing only on currently active social science and humanities faculty, produces the picture in Figure 44. The most representative data come from Britain, where there is a 6:1 ratio of support between left or liberal and conservative parties among SSH faculty. In the US, the ratio in my survey is 12:1 and in Canada 10:1. In a UK online mailout sample of over 200 UK SSH academics, using the same method as the North American survey, and achieving a similar response rate, I found a 12:1 ratio rather than the 6:1 from the YouGov survey.
The PhD surveys, which are not convenience samples, show a 9:1 ratio of left to right in both Britain and North American academic figures may overstate the degree of left-liberalism across all US colleges, mainly because they do not include low-ranking institutions. Those scoring higher on the psychological traits of agreeableness and trust, who tend to lean left, may also be more likely to complete voluntary surveys.61 Against this, those who are interested in filling out a survey concerned with academic freedom may lean more centrist or conservative. All told, the 12:1 Democrat-to-Republican (D:R) ratio in my data matches Langbert et al’s (2016) representative voter registration-based findings for the top 40 US Economics, Law, Journalism, History and Psychology departments. It also matches his 12:1 D:R ratio for social science academics in 51 leading liberal arts colleges using the same method, though it is well below his 32:1 ratio in the humanities for these institutions. It also seems in line with previous such studies and thus a good representation of the ideology of the faculty of leading North American institutions.
Finally, the US sample is limited to the top 100 institutions and in Canada to the top 40. Thus the US data are most exclusive, followed at a very considerable distance by Canada, with Britain offering an unselective cross-section across all ranks of university. Since academics at elite institutions are, from what we know from previous research, more left-leaning, this may mean that North American academics as a whole are somewhat less left-leaning than my surveys suggest, and may not differ as much from their British counterparts as appears to be the case here.
Another way to examine ideology is to ask people to place themselves on a 5-point or 10-point left-right scale. In a review of five recent studies, Shields and Dunn find the share of conservatives in the social sciences to be 5-15%, and in the humanities, 4-8%.62 In the European Union, data from the high-quality European Social Survey, which is a random stratified rather than convenience sample and obtains a nearly 80% response rate, find professors to be substantially more left-wing than other professions, especially on immigration and attitudes to the European Union.63 For Canada, Nakhaie and Adam (2008) showed that 11-13% of Canadian SSH and STEM academics voted Conservative or Reform in the 1993-2000 period, with the left outnumbering the right 5:1.64 They suggested Canadian academics were thus more centrist than their American counterparts. Our data does not support this contention, though ours is a more elite, SSH-leaning and considerably smaller sample.
Incorporating my data and that from a large-scale survey of European academics and members of the 100,000 strong UCU, which represents about 80% of Britain’s academics (Mallinson et al. 2017), the data in Figure 45 tell a fairly consistent story. The left-right ratio is lowest in continental Europe, at 5:1, with various Anglosphere surveys picking up ratios of between 6 and 15 to 1. Right-leaning academics form around 5% of the total in SSH-leaning samples, similar to the figures recorded by Langbert et al. (2016) using voter registration data.
Regardless of differences in sampling technique, the broad trends are similar, pointing to a left-right ratio of between 6 and 12 to 1 among SSH academics in North America and Britain. STEM academics are consistently 12-16 points less left-leaning across all datasets, though the share of conservatives is only modestly greater than in SSH. The biggest gap between SSH and STEM is with respect to the share of far leftists, which is much lower among STEM faculty. Note that the British YouGov data include 40% retired academics and is half STEM, while the British mailout survey is limited to currently active SSH academics.
As with the van der Werfhorst study on Europe, my data, presented in Figure 46, show that academics’ voting and attitudes on leading cultural issues such as Brexit and immigration differ dramatically from those of the wider population. For instance, 52% of the public voted to leave the European Union, compared to only 17% of academics.
Data for the US, in Figure 47, and Canada, in Figure 48, show an even more skewed profile, with the share of American academics favoring limits on immigration nearly 9 times lower than that of the population, and nearly 7 times lower in Canada. While this is a greater gap than the threefold distance between British academics and the public, my UK online mailout sample mirrors the US and Canadian online mailout results, showing just 7% favoring an immigration decrease, eight times lower than the public. We may surmise that a representative sample of North American academics would likely lean more toward the three- or fourfold level of difference from the public seen in the British YouGov sample in Figure 46.
Much of the difference on cultural-populist questions is accounted for by education level. Around 2 in 3 Britons with a master’s or PhD voted to Remain in the EU, with little more than a quarter opting for Leave.65 A closer examination of YouGov’s UK Profiles data, with a sample size of several hundred thousand, shows that degree-holding professors and lecturers in my sample (most of whose participants can be matched in Profiles) are more left-wing than secondary school teachers, though, as Figure 49 shows, differences are on the order of just ten points more than senior levels of the teaching profession. Part of the difference is accounted for by the fact that academics nearly all have advanced degrees whereas many teachers do not. Among advanced degree-holding teachers, differences are more modest between academics and teachers, though still persist. The share of right-leaning academics stands out as especially low.
A similar difference holds for the EU referendum vote, as Figure 50 shows. Advanced degree-holding teachers (not shown) are more similar to university professors and lecturers in their Brexit vote, and in their views on immigration.
Finally, Figure 51 shows that degree-holders in the education profession as a whole (universities and schools) lean modestly more toward Remain compared to other degree-holding professionals. Education is not the most Remain-leaning profession, however, and degree-holders in other sectors all lean toward Remain. Brexit voting among academics (80% Remain, 16% Leave) locates them as even more Remain-leaning than graduates working in media, marketing, PR, and advertising.
Data are more difficult to come by for North America or continental Europe, but work by van der Werfhorst (2019) showed that in the European Union, only artists were more left-wing than academics. Teachers, while left on the economic dimension, placed well to the right of academics on cultural-nationalist issues like immigration or the European Union. In the United States, political donation data from the FEC find academics rank 7 of 62 professions in their propensity to donate to Democrats over Republicans. As Figure 52 illustrates, 90% of academic donations went to the Democrats, with near-total Democratic unanimity among professors of English, History, and Law.
Much of academia’s left-liberal tendency is connected to the fact that those who score higher on the big-5 personality trait of openness tend to self-select into higher education.66 Once there, they come to be influenced by like-minded people, especially in the social sciences and humanities. Those with the most advanced degrees are then disproportionately employed in academia. Moreover, the academic sector is, like the school system, largely publicly funded, and thus leans economically left. Finally, occupational typing, akin to the stereotype of bus drivers as men and nurses as women, may add a feedback loop to the process, locking in a process that discourages political diversity.67 I consider further evidence for this later in the report.
Ideological Age Profile of Academics
The ideological changes that have been charted in academia since the 1960s should have left an imprint in today’s academic age structures. In the US, Langbert et al. (2016) looked at five SSH fields in the top 40 US universities using voter registration data. They reported that among those under 35, Democrats outnumbered Republicans 23:1, falling to 16:1 among academics aged 36-45, 12:1 for those 46-55, 10.5:1 for those 56-65 and 10:1 for those over 65.68 In a second study selecting the top 2 public and 2 private institutions across 30 states, including a mix of SSH and STEM departments, they show that assistant professors, typically the youngest faculty, have the highest Democrat:Republican ratio (10.5:1) compared to associate (8.7:1) and full professors, the oldest group (8.2:1).69
My data for the United States is much less of a complete sample than Langbert and colleagues’ findings, though it has the offsetting benefit of including unregistered and non-identifiable staff. Like Langbert et al., I find that younger SSH academics in Figure 53 are clearly more left leaning (41% far left) than their elders. This is not true for the share of right-wing academics, however, which is higher in the youngest group (8%), perhaps indicating greater polarization at younger ages.
In Britain, Figure 54 shows there is a slightly higher share of far leftists in the youngest cohort (22%) than older cohorts of SSH academics, but no clear linear pattern, with those over 70 nearly as far left (18%) as those under 40.
Figure 55 shows the broad age pattern for four main ideological categories across four academic surveys for SSH academics. The American trends fit the classic story of younger SSH staff being more leftist than their elders. This looks like a pattern of cohort change that tallies with surveys over the past few decades, with each new generation more left-wing than the previous one. But this is much less clear in the UK and Canada, as well as among PhD students in the US, UK, and Canada, where there is no consistent pattern. It is therefore difficult to be sure that there is a general pattern across the three countries.
Looking at the other end of the scale in Figure 56, there is no evidence that right-wing academics are going extinct. If anything, there is a modest trend of younger SSH staff being slightly more right-leaning than their elders – though this is not a consistent pattern. However, note that the right-wing share consists of very small numbers of academics: below 10% of SSH faculty. All of which makes it difficult to screen noise from signal – indeed, there is no statistical significance to these age trends in my models.
It could be that the interpretation of right and left differs between generations. Yet, on the acid test issue of modern populist conservatism, immigration, there is also no clear age trend across studies, as Figures 57 and 58 show. Roughly 50-70% across surveys favor an increase and 5-15% seek a reduction.
Taking a more systematic approach which controls for confounding variables, Figure 59 demonstrates that age is not a significant predictor of left-right ideology on a 5-point scale except, marginally, in the case of North American PhD students. The effect of being older is pointing in the direction of being less left-wing and more right-wing, but does not reach conventional (p < .05) statistical significance in any survey. On immigration, age is only related to more restrictionist attitudes among North American and British PhD students, but even here in opposing directions, confounding clear conclusions.
Recall from our earlier discussion of evidence from previous studies that the leftward shift of the professoriate occurred in two waves, one between the 1960s and the late 1980s, when the share of conservatives to liberals declined, and the second between the late 80s and the present, when the share of far leftists increased while centrists declined along with a continued modest decline of conservatives. To the extent that these shifts concern the passing of the Silent Generation (born before 1949), there will be no record of this in my data because they will be mostly, or wholly, retired by around 2015-20. I thus cannot comment on pre-Baby Boom shifts, which appear to have been decisive.70 We are therefore comparing Boomers with Generation X and Millennials, and here there appears to be no significant ideological difference between the generations.
While it is hazardous to read cohort change directly off age effects because people may change their views as they age across the life cycle, the existing literature suggests the direction of change would be expected to run in a conservative direction over the life course.71 The absence of statistically significant age effects in Figure 59 work against an explanation based on more leftist cohorts replacing conservative cohorts within the same demographic (i.e., younger left-wing men replacing older conservative men). The clear age gradation we see in the US survey, with the young more left-wing, fails to turn up in the form of significant age effects in all models.
What may be more consequential is gender, with women significantly more left-wing among both North American academics and PhDs, and among British academics in the YouGov survey. Even in the two surveys where women were not more left-wing, the effect was in the predicted direction. Nonwhite academics were only (statistically) significantly more left-wing in one of the four surveys where race was measured, so increasing racial diversity does not seem as likely to have altered the ideological composition of the SSH faculty as the greater proportion of women.
Even so, the data don’t tell a clear story of growing leftism arising as a consequence of a larger share of female recruits at younger ages. Plots of ideology by age within male and female subsamples do not show, as we might expect from the gender-driven leftism thesis, flat age lines (with the young no more left than the old) for both male and female subsamples; with only the plot of all respondents showing a sloping pattern of younger people being more left than the old (due to the compositional effect of more females at younger ages). It could be that women leave the profession in larger numbers than men, offsetting what might be an even larger shift to the left.
Finally, being in an SSH discipline as opposed to a STEM field is associated with being more left-wing in three of the four surveys (no STEM were sampled in the UK mailout). This aligns well with existing research.72 This said, STEM respondents were only significantly more right-wing than SSH respondents in 1 of 4 surveys. By contrast, STEM scholars were less likely than STEM scholars to identify as far left in 3 of 4 surveys.
The mystery is also not solved by examining changes in the wider political environment. The shift of those with degrees, especially advanced degrees, to the Democrats is more recent. Figure 60 shows there was a modest shift among advanced degree-holders to the Democrats between 1966 and 1992, some return to the Republicans during the Clinton years, followed by a shift back to the Democrats in the post-George W. Bush period. Note that even in 2016, about 35% of advanced degree-holders identified as Republican, far above the approximately 5% share among American SSH academics recorded in my survey.
When it comes to ideology, Figure 61 shows that a statistically significant trend toward liberalism among American advanced degree-holders is mainly a post-2012 development linked to Trump. There is also a steady decline of conservatism and modest rise in far leftism between 2004 and 2012 that doesn’t reach statistical significance. While this cannot explain the American academic macro trend of declining conservatism prior to the mid-1990s, it may be connected to the continued fading of conservatism from academia after 2004. Even so, the observer is struck by how little of the longitudinal trend charted by Sam Abrams and others can be tracked in the wider national opinion data. More educated Americans have been becoming more liberal over the last few decades, but the ideological transformation of academia has been much more extreme.
Having said this, it is also noteworthy that advanced degree-holders have been consistently more Democratic-leaning than undergraduate degree-holders in most years, a statistically significant phenomenon (see Figure 62). There is an education effect, but it doesn’t seem to explain the chronological shifts to the left picked up in the academic data since the late 80s. Changes in the meaning of left and right due to political events or ideological sorting do not therefore account for change in the partisan composition of the professoriate. The emergence of more populist Republican leaders such as George W. Bush and Donald Trump cannot explain why Republicans were largely absent from the SSH academy by 1999.
Age and Selection Effects Inside the Academy
Something distinctive is occurring in academia, but if this involves a narrowing of the intake, we should see more of an age gradient, and we don’t, unlike older studies that did. Lipset and Ladd, for instance, using 1972 data, found faculty under 30 to be 35 points more liberal than those over 50 and nearly 50 points more supportive of black activism.73 By 2006, however, Gross and Simmons reported that faculty under age 36 were less likely to be on the left than those aged 50-64, though the youngest cohort contained no more conservatives than the 1960s Baby Boom generation. Those over 65 (representing the Silent Generation) were more conservative than the Boomers though the left still outnumbered the right in this group by over 3:1. The authors reasoned that the Baby Boom generation had introduced the greatest ideological change. Indeed, Rothman and his colleagues, analyzing Carnegie 1984 and 1999 survey data, reported that the left:right ratio had shifted from 39:34 to 72:15 in 15 years. While their sample may have exaggerated the extent of the change, this would be in keeping with the thesis that the 1960s generation was remaking the academy.74
My UK YouGov sample consists of 40% retired academics with a median age of 70. While a purely generational theory would presume that pre-Boomer retired academics (born before 1949, thus over age 70) would be more right-wing than Boomer retirees (aged under 70), I found no significant relationship.
Does this mean that the leftward shifts charted by Abrams and others for the US and Halsey for Britain are an artifact of sample bias? Not quite. One reason why age data may not align with cohort shifts over time is selective exit. Selection effects are important for explaining academia’s ideological makeup. As early as 1959-61, a study of Berkeley students found that socialists were four times more likely than conservatives to “realistically consider” an academic career.75 The same may hold for leaving the profession. The self-selection of non-leftists out of academia would explain why older academics are not more conservative than young scholars even as academia has been shifting left over the generations.
Indeed, Figure 63 shows that it is younger retirees 65 and under who are more conservative than academics of the same age who remain in the profession (in my British YouGov sample of current and retired academics).76 Conservatives or centrists are more likely than leftists to select out of academia before age 65.
The share of ex-academics may be considerably larger because only those who listed their profession as university professor or lecturer with YouGov were counted as such. Those who switch professions and state a new one are not captured by this data, yet the YouGov Profiles data on PhDs in Britain visited earlier shows that older PhD holders (mainly working outside academia) are considerably more conservative than PhD holders working as academics. The ideological climate may thus operate not only to discourage conservatives and centrists from entering, but to disproportionately nudge the conservatives who start their careers in academia out the door.77
Left-Modernist Hegemony?
It may be the case that the predominantly left-wing ideological profile of sociology professors recorded by Lipset and Ladd in 1972 has come to spread across the entire academy and even down to an entire generation of students. Figure 64 summarizes Prolific demographics for various student categories, drawing on a 100% sample of those taking surveys on the site. US data show that PhD students on the platform are the most left-leaning (10.8:1), with master’s and undergraduate students somewhat less so (between 8 and 9 to 1), and non-students between 18-24 markedly less so (3.6:1). In Britain, there is a steady education gradient, with each higher level of education more left-leaning than the previous level. The general pattern is one in which further education correlates with greater leftism. The Prolific data are heavily skewed to the left compared to the more representative ANES data we visited earlier, where 35% of American advanced degree-holders voted Republican in 2016, with no significant variation by age.78
The best benchmark is the UK YouGov Profiles data, which is more representative of the wider population. Figure 65 shows that in Britain, there is a 2:1 left-right ratio among 18-24-year-olds who are not in education, and among undergraduate and master’s students alike. This jumps to nearly 5:1 among PhD students, however, indicating an important step-change between master’s and PhD study. This in turn compares to a 1:1 ratio in the population at large. On immigration attitudes, young non-students back restriction over increasing immigration 2.6 to 1, just below the population average. This slips to around 1.5:1 among undergraduate and master’s students, and 1:1 among PhD students. Here PhD students stand out on the left:right measure, though immigration attitudes move along more of a steady gradient.
These data speak to powerful political selection effects as students ascend the academic ladder. Do those with inquiring minds who pursue knowledge score higher in the Big 5 personality trait of openness to new experience and thereby tend to embrace left-modernism? My Prolific PhD surveys show that STEM PhDs are 16 points less likely to be on the left than SSH PhDs in North America and 13 points less likely to be so in the UK. A similar gap shows up in the data on academics.
We also know that there is an important political gap between advanced degree-holders who work in academia and those who work outside it, which widens with age. This points to the content and ethos of SSH academia as an important factor. How much of this is legitimate (social problems naturally attract the left-inclined) or illegitimate (universities and especially SSH departments are hostile environments for conservatives, and thus fail to ask important social questions due to bias)? This is the question I turn to next.
Research into why conservatives select out of academia by Woessner and Kelly-Woessner (2009) finds that conservative undergraduates are only half as likely to major in SSH subjects as left-wing students. Conservatives tend to prioritize practical over expressive goals more than leftists. Even so, between their freshman and senior years, only left-leaning students become more interested in the idea of pursuing graduate work while conservative interest remains static. Conservatives also report substantially more dissatisfaction with their SSH elective classes than their STEM electives, in contrast to left-wing students. In political science classes, the authors note, “we found clear evidence that students who believed they were at odds with their professor’s politics were generally more critical of the professor, the course, and the subject matter.” Conservatives are also less likely to report being mentored by (mainly liberal) academic staff. This suggests that while self-selection is the strongest reason for academia’s skewed political profile, the academic environment is probably a contributing factor.79
The authors mention the possibility that a hostile environment may be shaping preferences, though their data do not ask this directly. It is also worth inquiring into the extent to which selection pressures operate further down the academic pipeline at the master’s level. To examine these questions further, in a December 2020 Prolific survey of 843 master’s and PhD students, I asked:
How interested are you in pursuing a career as a university academic (i.e. Lecturer, Professor)?
Answers were provided on a 5-point scale from “not at all interested” to “extremely interested.” For analysis, I collapsed the “very” and “extremely” answers into one category to signify strong interest in an academic career. This was followed by, “How much do the following considerations affect your decision?” and a battery of six questions, each on a 7-point scale from strongly agree to strongly disagree:
· “Academic jobs are too hard to get, especially where you want them.”
· “I can earn more money in a different job.”
· “Academia is too isolating, I prefer more social interaction at work”
· “There is too much form-filling and hoop-jumping in academia”
· “My political views wouldn’t fit, which could make my life difficult.”
· “I’m not the academic type.”
The first point to note in Figure 66 is how clearly ideology structures the answers to the “my political views wouldn’t fit” question, with conservatives much more likely to believe their views could “make my life difficult” if they pursued an academic career:
The argument that conservatives are repelled by academia’s low earning potential, at least among those who made it to graduate school, finds no support in the data. Figure 67 presents ordered logistic regression coefficients for models predicting responses to all six questions, with controls for age, gender, income, SSH v. STEM, master’s or PhD, and country. The chart shows that left- and right-leaning graduate students only differ in their answers on two of the six academic career considerations. Namely political views, which is by far the strongest effect, and in their view that it would be too difficult to find an academic position where they want one. The latter may indicate that conservatives are more oriented toward geographic stability than liberals. This may also point to Woessner and Kelly-Woessner’s observation that conservatives prioritize starting a family more than liberals.80 In any event, within the self-selected group of graduate students taking the survey, the prospect of being able to earn more money elsewhere does not seem to deter conservatives more than liberals, echoing findings in previous work.81 Moreover, those on the left are, if anything, somewhat more likely to say they are “not the academic type.”
The foregoing doesn’t mean that the political atmosphere in academia is a deal-breaker for all conservatives. According to Figure 68, among graduate students, concerns about political fit rank fourth, and then only for right-leaning graduate students. For those on the left, not fitting in even seems mildly correlated with interest in an academic career. Being “not the academic type” and saying that academia is too isolating are the stronger predictors of who is interested in being a professor.
An interesting dynamic concerns the difference between master’s and PhD SSH students. Right-wing SSH master’s students who sense they won’t fit politically are significantly less likely than leftist and centrist SSH master’s students who say they won’t fit politically to be interested in an academic career. Figure 69 shows that just 9% of right-leaning SSH master’s students who say their views would make a poor fit with academia have a strong interest in pursuing an academic career compared to 29% of leftist or centrist SSH master’s students who say their politics would fit poorly.
On the other hand, 42% of right-leaning SSH master’s students who say that political compatibility is not a problem say they are interested in an academic career. While there is just a 4-point difference between leftist/centrist master’s students who think their politics is, or is not, a fit with academia, the gap reaches 33 points on the right. And, as we have seen, right-wing graduate students are already much less likely to say their views would make a good fit with an academic career than are left-wing or centrist counterparts.
Sensitivity to a hostile political atmosphere thereby acts to deter conservatives from pursuing an academic career when they are studying at the master’s level. The data show that graduate students who say they are “very right” are about 30 points more likely to say their politics doesn’t fit academia than those who are “fairly right.” We might also surmise that the conservative master’s students who say their views would make a difficult fit with academia are more culturally conservative whereas those who say fit is not a problem may be more economically conservative. On the left, the meaning of saying one’s politics don’t fit may be different, perhaps reflecting a more general feeling of nonconformity to institutional dictates.
Importantly, the few conservatives who commit to studying for the PhD are less likely to be put off by academia’s political atmosphere. Conservatives unwilling to mute their views or put up with hostility from the left have arguably selected out by this stage. Importantly, the share of “very right” compared to “fairly right” graduate students does not vary from master’s to PhD level, or between those with an interest in academia and those who have no interest. This holds out hope that a small remnant of conservatives will remain in SSH academia into the foreseeable future.
What, then, accounts for self-selection effects being stronger at master’s than PhD level? To begin with, the share of right-wing SSH PhD students who say their politics are incompatible with academia is 33%, lower than for SSH master’s students on the right, where 49% say their politics are incompatible with academia.82
Second, right-wing PhD students who see their politics as a difficult fit with academia are more interested in pursuing an academic career than their politically-incompatible master’s counterparts. Whereas there is a 32-point gap among right-wing master’s students between those who view their politics as fitting versus not fitting an academic career, right-wing SSH PhD students who see their politics as incompatible are actually 10 points more likely to express interest in an academic career – though this is not statistically significant as there are only 20 right-leaning SSH PhD students in the sample.
Running the master’s portion of the above as a statistical model, with strong interest in becoming an academic as the outcome measure, and controlling for age, gender, income, country, university rank, and the five other academic career questions, results in Figure 70. This illustrates that it is among right-wing master’s students that perceived political hostility makes an impact. Resilience, an ability to conceal one’s views, or a greater comfort navigating the rules of academia probably explain why right-wing PhD students are better adapted than right-wing master’s students for academia. The effects are driven by the tendency of right-wing master’s students in SSH fields to view themselves as politically incompatible. Those on the left or in the center who see themselves as politically incompatible, whether in STEM or SSH, are actually more likely to express an interest in academia, suggesting that the meaning of political incompatibility may vary by ideology.
Finally, it should be said that political considerations are not a significant predictor of interest in an academic career among STEM graduate students at any level. Figure 71 shows that political compatibility is only consistently related to interest in an academic career among SSH graduate students – and even here, it is significantly correlated only among right-wing SSH master’s students. Notice as well that the transition from master’s to PhD seems to coincide with reduced interest among STEM students and increased interest among SSH students, which likely has to do with more plentiful career opportunities for STEM PhDs outside academia. Social isolation rather than loss of earnings is the dominant factor in deterring graduate students overall from pursuing an academic career.
These findings largely comport with those of previous researchers. As Gross and Fosse summarize, “Woessner and Kelly-Woessner (2009) find that twice as many liberal as conservative college students aspire to complete a doctorate.” Other work (Gross and Simmons 2006) shows that being conservative, Republican, or an evangelical is associated with lower confidence in universities and viewing professor as a lower-prestige occupation. The authors themselves argue that academia is “typed” as a liberal occupation the way the military is typed as conservative, and that this occurred in part due to a growing post-1960s identification, in the minds of many, of academia with the left.
Though the authors are liberals who stress the importance of self-selection, they also allow that conservatives may face hostility for their views. Moreover, they acknowledge that entire disciplines and elite universities construct identities (“images of intellectual personhood”) that include being left-wing. Thus, they continue, hiring committees at elite universities may be biased against conservative applicants because right-wing candidates don’t fit their idealized image of what a leading academic should look like.83
Section IIb: Chilling Effects and Self-Censorship
A Hostile Climate for Conservatives in Academia?
One reason for conservatives to stay away from, or leave, academia is a perceived hostile work environment. Academia may lean left, but is this infringing on the liberty of academics to teach and research? Not if the composition of the academy is the result of self-selection in an atmosphere supportive of all points of view. If, however, it results from deliberately exclusive or hostile actions from institutions or academics, then this does represent a constraint on academic freedom and the mission of universities to pursue knowledge.
Before proceeding, it is important to address a popular theory on the right: that universities are indoctrinating students into left-wing beliefs. Repeated studies, including my previous co-authored report Academic Freedom in the UKshow little to no effect of classroom exposure on students’ political views.84 Thus, rather than impacting students’ opinions, the more pressing effect of political discrimination – if this is indeed taking place – is on the academic experience of staff and students, the quality of research, and the principles of expressive freedom and the pursuit of truth.
My first aim therefore is to take the temperature of the academic workplace, a climate predominantly established by colleagues, students, and university administrators. How strong is the pressure to conform? Are there fears that nonconformity will be punished? This perception, if widespread, would restrict the expressive freedom of a large number of faculty, and present a far greater challenge to academic freedom than no-platforming.
I began by asking about subjects’ own political ideology on a 5-point left-right scale, then asked where they thought the average member of their department was on the same scale. UK YouGov survey results are summarized in Figure 72. In terms of respondents themselves, 53% identified as left, 35% as centrist, and 9% right. For currently employed academics in the social sciences and humanities, the figures were 61% left, 30% centrist, and 7% right.
What is especially noteworthy is the significantly higher presence of those describing themselves as “far left” and as “activists” within social science and humanities disciplines as compared to STEM fields. Indeed, among currently serving academics, 16% of SSH academics are on the far left compared to 6% for STEM. Twenty-six percent of current SSH academics consider themselves activists compared to 15% of STEM staff. These findings mirror those from US scholars who similarly found a big difference in self-identified activists, radicals, and Marxists between SSH and STEM.85 The proportion of centrists in SSH fields, at 32%, is also markedly lower than the 43% for STEM subjects.86 Gross and Simmons also found that younger professors were less Marxist, radical, and activist. In my data, academics under 40 are indeed less likely to be activists (16%) than those over 65 (27%) but the under-40s are 5 points more likely to be far left (15.6%) than those over 65 (10.5%).
Figure 73 shows an even larger gap between STEM and SSH subjects in the United States than in Britain. The British online mailout SSH sample shows 30% activists, similar to the 28% for the YouGov sample, and well below the 38% activist component recorded for US SSH, and 35% for Canadian SSH, academics. The UK online mailout results also show 22% “very left” compared to 15% for the YouGov UK sample, 26% of US SSH sample, and 23% for the Canadian SSH sample.
As Figure 74 shows, SSH academics in Britain are significantly more left-wing than STEM academics. UK women are also more left-leaning than men, though the gender gap is twice as wide in STEM, whereas SSH fields tend to be uniformly left-leaning across gender.
The data for the US (as for Canada), presented in Figure 75, also find social science and humanities academics to be more left-wing, and women are again more left-wing than men – as they are in the wider population.
British academics’ perceptions of their departments reflect the reality that SSH fields are more strongly left leaning. In Figures 76 and 77, I compare self-identified ideology with academics’ perceptions of the average political leaning in their department. Though the two questions are not directly comparable, the answers suggest that most academics have a fairly good sense of the climate of opinion in their workplace.
This said, while British SSH academics are more left-wing than their STEM counterparts, Figure 77 shows that British SSH staff view their departments as 8 points more left-wing than themselves. The degree of this perception contrasts with STEM staff. While this could be an accident of my sample, it may also be due to the clearer cues about political values provided by the content of teaching and research in SSH fields.
In particular, right-leaning SSH academics tend to perceive a greater left skew in their departments compared to both left-leaning SSH academics and those in STEM subjects. Figure 78, for instance, shows that “fairly right” British SSH academics see their departments as half a scale point more left-wing than their “fairly right” STEM counterparts. Among “very right” SSH academics in Britain, the gap is a full point, which means that “very right” SSH academics view their departments as “very left” while “very right” STEM staff see theirs as “fairly left.” Results for “very right” academics should be taken with caution, however, as only 8 individuals in the data (4 STEM and 4 SSH) identify as such. On the other hand, leftist and centrist SSH staff view their departments as only slightly more left-wing than STEM staff. This shows that right-leaning staff in SSH subjects may perceive their work environments as more politically distant than they really are, as compared to their conservative STEM counterparts. The combination of SSH being more left-wing in reality, and it being perceived to be even more so than it really is (arguably due to the more transparently political content of these fields), creates a marked sense of dissonance among right-leaning social science and humanities academics.
The situation in the United States and Canada is relatively similar to Britain, with perceived and sample average (“actual”) ideological composition fairly closely aligned. However, Figure 79 illustrates that while the size of the far left is underestimated relative to the sample in Britain, it is overestimated by American survey respondents compared to their own makeup. In combination with the considerably higher perceived far-left share in America (26% compared to 7% in Britain), this indicates that academia in the US – at least in the top 100 departments – is likely to be more left-leaning than in Britain.
Comparing SSH and STEM disciplines, the pattern in the US resembles Britain insofar as more right-wing American SSH academics perceive their departments as being somewhat more left-wing than leftist SSH academics do. Thus the most right-leaning SSH academics in America view their departments as somewhere between “very left” and “fairly left,” whereas left-leaning SSH academics see their departments as “fairly left.” This said, there is little difference in the ideological skew in perception between STEM and SSH academics in the United States, as the two lines in Figure 80 run parallel with each other.
In Canada, the data look more similar to Britain, with a little over 60% of respondents perceiving their department’s median academic as being on the left.
Chilling Effects
In the absence of strong collegial norms against political discrimination, an ideological skew translates fairly readily into a hostile climate for conservative scholars. The combination of leaning right and being in a social science or humanities department, for instance, conditions an academic’s answer to the question of whether “there is a supportive or hostile climate towards people with your political beliefs in your department.” Figure 81 shows that 44% of “fairly right” and 63% of “very right” British academics in our YouGov sample perceive a hostile climate in their department. Notice as well that “fairly left” academics are most comfortable, with “very left” academics feeling slightly more hostility, though this is not statistically significant. The UK mailout survey shows, similarly, that 17% of far-left, 5% of center-left, 26% of centrist, and 60% of right-wing academics report a hostile climate. For UK PhDs, the numbers are 16-4-9-71, a similar profile, albeit with an even stronger sense of exclusion among those on the right.
In my American sample, centrists and those on the right perceive a considerably higher level of hostility for their beliefs than is the case in Britain (see Figure 82). The fact that nearly 4 in 10 centrist American academics report a hostile climate dovetails with earlier data suggesting US universities may tilt further left than in Britain.
However, our US sample is considerably more exclusive than the British one, in that it is drawn from top 100 schools and is more SSH than STEM in composition compared to the UK YouGov sample (but not the British online mailout sample). Canadian data are also heavily SSH, but come from a relatively wider selection of universities. Comparing “apples to apples” across the three countries, restricting to SSH disciplines, and also including two high-response rate Prolific samples of PhD students from Britain and North America, reveals the pattern in Figure 83.
Here the responses look relatively uniform across the board, regardless of country, with between 60 and 80% of right-wing academics and PhDs reporting a hostile climate for their beliefs. Generally, both centrists and “very left” respondents report higher hostility than the “fairly left”, who appear to be the modal category in academia. One sample that was not included in Figure 83 is a survey of 227 of the members of the National Association for Scholars, a pro-free speech academic association. NAS members in the sample break 10% left, 24% centrist and 64% right-wing, with 58% voting for Donald Trump in 2016. Sixty-seven percent of the NAS sample reported a hostile environment, including 72% of those on the right – mirroring the results in Figure 83.
In a statistical model of UK YouGov data in Figure 84, controlling for age, gender, rank, income, and whether someone is retired or active, I find a sharper ideological hostility gradient among SSH than STEM staff. Left academics perceive their departmental political climates as supportive, while academics on the right, especially SSH staff, experience a hostile climate.
Focusing only on UK academics who perceive a “somewhat hostile” or “hostile” climate for people with their political beliefs, I find that a quarter of centrists, half of “fairly right” academics, and all “very right” academics in SSH disciplines experience this. Among STEM academics, the share of right-leaning staff who feel hostility is lower, at between a quarter and a half. Once again, Figure 85 shows a pattern of greater alienation among right-leaning staff in SSH fields.
A somewhat similar story holds with respect to identifying with the culture of one’s department. Here I collapse “fairly” and “very” right-wing academics into one category because of small numbers. As Figure 86 shows, a majority of those on the left identify with their departmental cultures, but among centrists and those on the right, fewer than half do. In addition, among right-leaning academics, the share identifying against the culture of their department (42%) is nearly twice as large as that identifying with it (22%). There is little difference in perception between the “fairly” and “very” left-wing professoriate.
American data in Figure 87 show a sharper ideological gradient than in Britain, with 61% of right-leaning and 33% of centrist academics identifying against the culture of their department. Identification with one’s department is also higher among left-wing US academics than their British leftist counterparts, hinting at a more closed departmental atmosphere. Again, however, we must bear in mind that the US sample is more elite and more SSH in composition than is true of the British data. Finally, data from the National Association of Scholars sample show that 71% of the 150 right-wing NAS academics in the survey identified against their department, comporting with other results.
With controls for status and demographic factors in Figure 88, we see that the main difference between British STEM and SSH academics is on the left of the spectrum. Here, the SSH professoriate identify significantly more strongly with their departmental culture than do their peers in the STEM professoriate. Again, we see a stronger ideological gradient in the social sciences and humanities than in STEM fields.
Running a slightly different version of the model for North America in Figure 89 shows that ideology is again a statistically significant predictor of identifying with one’s department, with the effect more pronounced in the social sciences and humanities than STEM.
Among the 77 SSH PhD students in my British sample, 60% of those identifying as right-wing, but just 17% of centrists and 7% of leftists, said they identified against their department’s culture. Among the 124 North American SSH PhDs, the equivalent split is 55-31-7. These numbers are fairly similar, but indicate a slightly higher share of US and Canadian centrists identify against their departmental culture compared to Britain.
Expressing Political Views to Colleagues
In the overall British YouGov sample, 78% of academics voted Remain and 16% Leave. In my previous co-authored report on Academic Freedom I found that fewer than 4 in 10 Leave-supporting students would feel comfortable expressing their Brexit view in class compared to nearly 9 in 10 Remain-supporting students.87 The results in Figure 90 show a similar disparity.
While 85% of all academics in the sample said Remain voters would feel comfortable expressing their Brexit view to colleagues, just 37% of Leave voting academics said Leave supporters would feel comfortable doing so. Narrowing the focus to exclude STEM and retired academics, I find that just 18% of current Leave-supporting SSH academics say a Leaver would feel comfortable sharing his views with colleagues.
The US sample tells an even more extreme story with respect to Trump supporters in the 2016 election. In Figure 91, just 14% of Clinton voters and a mere 3% (1 individual) of the 33 Trump-voting academics in the sample agree that a Trump supporter would feel comfortable expressing this view to a colleague. In contrast, over 90% say a Biden supporter would feel comfortable expressing their views to a fellow lecturer or professor. All told, this is a staggering disparity, and, whatever Trump’s flaws, offers powerful evidence for the operation of Mill’s “despotism of custom” and the absence of what Lukianoff terms a “free speech culture” among the professoriate.88 This limits the range of potential interchange.
If we just focus on current Trump- or Leave-supporting academics and PhD students in the social sciences and humanities in Figure 92, we see a consistent pattern of silence, with just 0-33% of those polled saying that someone on their side would feel comfortable expressing their view to a colleague. By contrast 83-100% felt that Remain or Biden supporters could express their views openly. Though there are few Trump/Leave supporters in the surveys (sample sizes range from 6 to 26), the data tell the same story across countries and survey methods. Among the 126 Trump-voting academics in the NAS academics survey, 11% said Trump supporters would be comfortable expressing their views, 82% said they wouldn’t be, and 7% were unsure. This compares with 87% of NAS Trump voters who said a Sanders supporter would be comfortable doing so. The portrait this paints is of a profound chilling effect and the shutting down of important opportunities for the exchange of perspectives and ideas across central societal divides.
There is important variation according to whether someone is retired and whether they are in STEM or the social sciences and humanities. Figure 93 shows how comfort among British academics varies between SSH and STEM staff, and by Brexit vote, controlling for academic rank and demographics. Among currently employed social science and humanities faculty, just 18% of Leave supporters indicate that a Leaver would feel comfortable expressing their view. Widening this to include retired SSH faculty, 28% of Leavers say a Leaver would feel comfortable expressing this opinion to colleagues. And among Remain supporters or non-voters, just 31% of SSH staff think a Leaver would feel comfortable expressing their view, suggesting this is not just the perception of Leavers. This paints a stark picture, in which Leave supporters feel they need to keep their views to themselves. This chilling effect is especially pronounced in the social sciences and humanities, precisely the places where an open exchange of social and political views is vital to the mission of research and teaching.
A more serious level of self-censorship is for academics to actively silence themselves in their research or teaching. Thus I asked respondents, “Have you refrained from airing views in teaching or academic discussions, or avoided publishing research, because of possible consequences to your career from doing so? Please describe your experiences (if any), and how frequently you have faced them.”
Before delving into the substance of the qualitative feedback I received, I coded responses as 1, where people had refrained from airing views for reasons of politics or ideology, or 0, where a negative response was received or complaints had to do with non-political content. Results for the UK YouGov academic sample in Figure 94 show that self-censorship is over twice as high among right-leaning academics as others.
Among two groups of academics, namely right-wing academics who currently teach, and those in the social sciences and humanities, over 40% said they self-censored. When the overlap of these two groups is considered, namely right-leaning staff currently teaching in the social sciences and humanities, the share who have self-censored reaches 50%.
The model in Figure 95 shows that SSH academics self-censor more than STEM faculty, though the disciplinary effect doesn’t widen among right-leaners. The model confirms that, even controlling for status and demographic factors, right-leaning academics self-censor at much higher rates than leftists or centrists. There is slightly higher self-censorship among very left compared to fairly left faculty, but this is not statistically significant.
In North America, self-censorship appears to be greater than in Britain. Figure 96 shows that 70% of right-leaning American social science and humanities academics, and 42% of centrists, say they self-censor in their teaching or research.
The pattern in Canada appears closer to the United States than Britain, with 47% of centrists and 56% of right-leaning academics self-censoring. Figure 97 shows that the red line for SSH is again higher than for STEM when controlling for age, race, and gender. As in Britain, SSH academics self-censor significantly more than STEM academics.
A full comparative chart, taking in all SSH datasets, appears in Figure 98. Though there is variation, the broad pattern shows a substantially higher level of self-censorship on the right, with some also taking place among centrists and far leftists. Centrists in North America seem to have an elevated fear compared to their UK centrist counterparts. One exception is the National Association of Scholars survey, where around 60% of 226 respondents said they self-censored, with little variation by ideology. An important group of NAS scholars on the right said they spoke out despite considerable blowback, suggesting that legal-moral support groups like the NAS may have a constructive role to play in upholding academic freedom, or that they select for more resilient free speech activists. There was no clear age pattern for self-censorship, which occurs among those of all age brackets.
British Testimonials
A range of comments from our British YouGov survey underscore that social pressure from colleagues is the primary deterrent to expressing views. The following pages outline this qualitative data in detail. Those who wish to bypass this portion of the report can move ahead to the following section. Testimonials have been lightly edited for spelling and grammar.
While many simply indicated a “yes” or “no” when asked whether they have “refrained from airing views in teaching or academic discussions, or avoided publishing research, because of possible consequences to your career from doing so,” others offered an insight into the climate of self-censorship that prevails for many academics on campus.
“Totally, as I am not massively left-wing, and most of my colleagues were. In the interests of harmony and a comfortable working environment, I used to just keep my views to myself or make grunting noncommittal noises during discussions which turned political. This would happen in staff meetings, or often during subject seminars for lecturers. It went on for years.” – Centrist, Remainer.
“Yes. Avoided expressing views due to bullying in workplace” – Centrist, immigration restrictionist.
“Yes. I have to be careful about expressing my views, because the bigoted politically correct, woke, thought police would object to my conservative (with a small ‘c’) views.” – Right Leaver
“I voted leave but was scared to reveal this as my colleagues were so aggressive in their attitude.” – Tory, Leaver
“While enmity toward even centrists, let alone conservatives, is not overt, the atmosphere is such that homogeneity of opinion and sociopolitical views is implied and expected. For all the protestations of diversity and inclusion from the ‘woke’ squad, the one thing they most singularly do not want is diversity of opinion.” – Tory, Remainer
“Yes, I tend to remain quiet. Letting my leave views out in the open would definitely harm my reputation in the uni.” – Tory, Leaver
“Yes. I am a liberal conservative and the vast vast majority of my colleagues at university are very Left wing. I support individual responsibility, thrift, independence, hard work, low taxation and small government. My colleagues opposed all of these. I have long been a euro sceptic and support leave- there were some colleagues who shared this view but were very loath to say so publicly. The hypocrisy of the champagne socialist colleagues was breathtaking. As was their open hostility to anyone who held views different to their own.” – Labour, Leaver
“Yes. No specific instances but generally I ‘guard my tongue’ except with people I fully trust.” – Tory, Leaver
“1) I hinted that Leave was not so bad as it seems to 2 colleagues who then accused me of being a leaver and were very angry and abusive. 2) I told someone I had voted Leave and they called me a racist.” – Centrist, Leaver
“Yes, refrained from talking about leave vote because the atmosphere is that this is an inappropriate view to hold. This being the case since the referendum.” – Centrist, Tory, Leaver
“As a Conservative voter I would not share my political views within the workplace. I am certain I would have been regarded with hostility from a number of fellow employees. I have seen this happen with other employees. The ability to discuss and explore political beliefs seemed to have disappeared.” – Centrist, Tory, Leaver
“Been told leavers are fascists” – Centrist, Leaver
“I have had research ideas that I have not pursued as I think they would have negatively impacted my career.” – Centrist, Tory, Leaver
“Certain areas are off limits in academia and you risk your job and reputation if you go against the herd. The days of academic freedom are long gone and it is suicide to express views that differ from those of the liberal thought police.” – Centrist, Remainer
“I have ambivalent feelings about the Hijab, however, I would not voice them in my department for fear of being considered Islamaphobic, rather than it being an issue of patriarchy, which should be fought against.” – Centrist, Labour, Remainer
“I don’t think my views on trans issues are entirely in line with the current orthodoxy in academia.” – Centrist, Labour, Remainer
“Yes. It’s hard to put a number on it in relation to frequency but there is definitely a strong view within social work academia that some views (left-wing) are acceptable and some views (right-wing) are not. Social work as a discipline sees itself I think not really as an academy in which debate and free-thought is the aim but as an advocacy group for marginalized groups (which is a laudable aim, don’t get me wrong - but it’s very different from traditional views of academia in my opinion).” – Leftist, Labour, Remainer
“Agreement is often assumed by ardent Remainers, and I usually avoid expressing my views unless asked directly. Agreement is also usually assumed by management for neoliberal employment policies (incl. redundancies): I spoke publicly against these and was ‘frozen out’ as a result (i.e. moved to another, non-leadership role). Agreement is also often assumed with various fashionable ‘right-on’ ideas about, e.g., sustainability, decolonizing the curriculum, etc. I generally avoid such discussions. I am aware of a colleague who has been complained about (anonymously) for her so-called ‘TERF’ views – I support her fully. She only discovered my similar views because I do not self- censor what I publish (though I do tend to self-censor in fact-to-face interactions at work unless asked a direct question about my views).” – Leftist, Labour, Remainer
North American Testimonials
American and Canadian respondents told a similar story, in which considerations of social pressure from colleagues, and to a much lesser degree students, was the primary motivation. However, this was not unconnected to career opportunities as colleagues often control departmental resources and opportunities, something especially important for younger staff. Moreover, in the US and Canada, concerns about political correctness were prominent not just on the right, but among centrists and even around half of those identifying as “fairly left” (among those who reported self-censoring).
The few conservatives in the sample had the highest incidence of feeling censored. Most cited social pressure from colleagues:
“Yes though not for career reasons so much as social reasons, not trying to stir the pot. There is no room in the academic setting as I know it to express even casually anything less than the most extreme left liberal views. I do not bother to try to have a conversation. I just smile and nod, every day, and say nothing. I avoid confrontation, and err on the side of being as polite as possible…I don’t want to hurt anyone, or be hurt, or create conflict that does not lead to new knowledge or growth, and so I say nothing about any of this, while hearing far left views very, very regularly, as though others didn’t exist at all, even as a possibility.” – Right, English, US
“I frequently bite my tongue around the lunch table with my colleagues, and I have refrained from publishing articles at outlets like Quillette or ArcDigital on controversial topics, because of fear of cancellation/public shaming. It’s just not worth it.” – Right, Political Science, Canada
“The few conservative professors I know have told me to keep my head down. There were even active attempts to cancel two gay professors in my department who publicly resisted the pronoun issue. I tend to speak fairly openly in seminars, but I’m good at building relationships and I typically speak my mind only after establishing a reputation as a thoughtful and considerate colleague. However, there is a crushing weight of assumptions in the air around me. Trying to make any argument about my position usually requires deconstructing so many assumptions that my interlocutors no longer even notices as assumptions. As for damage to my career, yes, I have frequently wanted to make written (rather than verbal) contributions but I don’t because I don’t want to leave a written record right now. In my teaching I am trying to co-opt the ‘rhetoric of cunning’ that comes from Critical Pedagogy in order to smuggle my viewpoints into my curriculum.” – Right, English, US
“I definitely keep my views to myself and adopt a posture of non-conflict with my colleagues and students at all times.” – Right, Psychology, US
“I generally try to avoid any discussion of personal politics, as I don’t think it’s productive” – Right, Political Science, US
“I generally wouldn’t share my views with colleagues, as I think I’m probably more conservative than most/all of them.” – Right, Political Science, US
“I am a closeted Conservative; I feel I can’t express my views because of the strongly leftist environment in universities.” – Right, Modern Languages, Canada
“Daily experience. Discrimination against conservatives is very real.” – Right, Music, US
Centrist responses were very similar in tone to those of conservatives:
“The legal academy in general, and my faculty in particular, is very Left. Teaching and publishing are not major issues for me because I believe in separating my personal beliefs from my teaching and my analysis anyway. Open meetings, whether faculty meetings or academic gatherings, are where centrists like myself, and certainly conservatives, will not bother speaking up to offer different opinions because they will be ostracized socially and professionally” – Centrist, Law, US
“Refraining from airing views generally occurs on Twitter. I would not, for example, criticize the book ‘White Fragility’ or suggest that defunding the police is a silly idea on Twitter. These topics are less likely to come up, for me, at work. However, post George Floyd, there will be a reading group where one of the suggested books is Kendi’s ‘How to be an anti-racist.’ I am uncomfortable with this book, but I would be reticent to raise objections at work. We’ll see what happens (will I object?) as the reading group materializes.” – Centrist, Psychology, Canada
“When colleagues invoke vague notions of equity to justify their contentious positions, I’ve refrained from asking for more information or questioning the studies that they contend make their points indisputable. I don’t want to seem hostile to equity as a force in departmental discussion. Also, on issues such as organized labour and Indigenous rights I’m a leftist by broader Canadian standards but a right-winger in my department, and so avoid saying much that would give my position away. To be clear, I often avoid asking follow-up questions when colleagues make sweeping and contentious statements on these subjects as if their views were obvious and avoid stating my own.” – Centrist, History, Canada
“I frequently (maybe once every other week per class) do not say things that are true for fear of career assassination from colleagues who are much further left than I am.” – Centrist, Sociology, US
“This spring my department released a statement in support of George Floyd that attributed all racism and oppression to Euro-American imperialism during the past 500 years. I am familiar with the evidence from ancient civilizations around the world. That evidence shows oppression, domination, and othering are characteristics of humanity and civilization no matter where and when. Rather than engage in heated Zoom discussions, I skipped the department meetings and then just didn’t vote when the statement came up for a vote.” – Centrist, Sociology, US
“I disagreed with a statement that the university was systematically racist, but was the only person in the department and feel this has hurt my reputation and standing in the department even though we had no discussion on the what the statement meant and whether it was true. The university personnel committee received at least one training session each year about implicit bias, but it would plainly have been unspeakable to wonder whether affirmative action and diversity hires might balance such biases out. Our leftist chair wanted to give a session for our first-year graduate students on supporting BLM. I questioned whether a conservative student would feel able to voice their contrary opinions at the start of their graduate career and obviously caused the chair irritation. These sorts of incidents happen only a couple of times a year, but they poison the atmosphere for me.” – Centrist, Classics, US
“Since the explosion of ‘antiracism’ in the past few months, I’ve heard a lot of antiracist ideas that are just...bad. I would be uncomfortable offering my candid assessment because ‘you’re either…antiracist or racist’ according to the new dogma. I’ve been spending more time on the phone with my brother, because we can have open discussions about these things.” – Centrist, Psychology, US
“I never air any view that is not strongly left/progressive out of fear of retaliation/‘prosecution’; this happens at nearly every faculty meaning when matters of teaching and student support are discussed, or when personnel matters are discussed.” – Centrist, Linguistics, US
Career-Related Concerns
In a large minority of instances where subjects elaborated on their experiences, concerns bled over from trepidation about peer pressure to fears that their words could have career consequences. This affected their choices about what to research, whether they should comment in extramural outlets like Twitter, and what to teach:
“Absolutely I have refrained. I am by no means an ‘in your face’ type political person and I am someone who is very aware of Trump’s bad character; however, I see more importance in policies being carried out and this is why I overall support him right now. I definitely have to be careful who I tell, because it is automatically assumed I am a horrible person due to the way I view the presidency right now even though I am someone more than willing to discuss differences and be friends with someone who believes differently. All in all, very frequently I worry and have to refrain from convos with other professors.” – Right, Neuroscience PhD, US
“I served on a hiring committee, and others on the committee quite frequently cited a candidate’s gender (female) or race (underrepresented minority) as a reason to hire them. I believe the level of bias shown is morally wrong, and probably illegal. I avoided expressing this view. More generally, I avoid discussing political topics with colleagues because there is no upside to doing so. I do not yet have tenure, and you never know you you’re going to upset. Regretfully, I do not have faith that people in academia are capable of separating political views from their evaluation of a person’s work.” – Centrist, Economics, US
In some cases, respondents mentioned that this pressure narrowed the range of their academic inquiry:
“My department is very liberal, and often mixes social causes into how we run the department. I don’t know that I necessarily oppose some of these actions, but my inclination is to question and investigate if they are necessary. But I hold my tongue because everyone else seems so sure, and I am not certain that my less than gung-ho approach would be welcome. I have also avoided undertaking at least one research project because of the potential fallout-I think the project is very reasonable, but from what I’ve read I doubt my administration or department would support pursuing research that challenges (or even has the potential to challenge) their moral/political views. I hope I’m wrong, but at present I’m too unsure of my position to try.” – Centrist, Sociology, Canada
“It is harder to publish research if the results could be considered ‘harmful’ to marginalized groups (for example, studies that do not find evidence of bias/unfair treatment, but instead find differences in risks/behaviors). Similarly, it is always a risk to talk about such information in class.” – Centrist, Sociology, US
“Frankly, even the publication of my survey responses would be enough to seriously damage my career. I’m not even that – I’ve never voted Republican in anything but a local election. But I have major disagreements with the prevailing thinking that comes from critical race theory, or deconstructions of gender, and saying so would be enough to tank my career.” – Centrist, Music, US
“Waited until after promotion to full professor to publish some of my research findings.” – Centrist, Sociology, US
“Currently in my department, Marxist+ further left is the king; therefore a solid centrist like me is considered conservative. I am up against all sorts of political, social, and economic barriers with my research because I work in business anthropology.” – Centrist, Anthropology, US
Staff who had not acquired tenure were especially skittish about revealing views that did not conform to progressive norms:
“I have refrained from airing my views because I know that I will be bullied, mocked, and abused on social media and in person. Additionally, I don’t have tenure, so being honest about my conservative views is dangerous to my ability to keep my job.” – Right, English, US
“Do not speak of politics, because I am a member of the precariat. Would like to keep my job.” – Fairly left, Psychology, Canada
“Because I fear repercussions and know for sure that my colleagues discriminate against people who do not share their views, on issues where I differ, I keep quiet. This is true particularly on matters of race, immigration, and the Middle East (everyone is pro-Palestine and I am pro-Israel). Not in my research nor in my teaching (I am not in a contemporary field) but in meetings and policy discussions, I have 100% remained silent because I am certain of retaliation. The biggest issue is that I believe ALL LIVES MATTER and that it is racist to say otherwise and single out some races as more valuable than others. But if I were to say this out loud I would be completely excoriated and banned and there would be serious consequences. As a Humanist, I am not allowed to say anything because what I actually think would be censured severely. So I keep my mouth shut and only vote when it is a secret ballot. My department issued a statement after the BLM protests that I absolutely disagree with but I have to see it up on our website as if it represents me. It is infuriating and they will not listen to reason. I am no racist and in fact I think THEY are the racists. They make race a judgement point in any decision; I think fairness demands that we treat ALL people fairly as MLK said, according to ‘the content of their character, not the color of their skin.’ I hate how racist my department is but I cannot say a word or I’d be run out, totally persona non grata, and my students would face bad consequences, and my research would lose funding.” – Centrist, English, US
“I am in the field of Middle Eastern Studies which, alas, is highly politicized. For many years, I did not voice political views in academic settings.” – Centrist, History, US
“At first I did not, but once I began to be politely asked to remove myself from conversations because my views are heterodox, I began to just stay quiet more often to avoid the hassle with people. There is A LOT of research that is not published for political reasons and A LOT of research that is published for political reasons. This is completely wrong, and while I have not suffered from this yet, I know that this is an ongoing discussion especially in my field of political science where people are often investigating politically controversial topics.” – Centrist, Political Science, Canada
“I am very wary about how (and with whom) I talk about trans issues. I know people who have lost their jobs over this, simply because they talked about ‘Bruce Jenner’ (‘deadnaming’) and challenged the ideas behind trans rights to enter sex-segregated spaces, change birth certificates, etc.” – Fairly left, History, US
“Yes, I have refrained from airing my views because I know that I will be bullied, mocked, and abused on social media and in person. Additionally, I don’t have tenure, so being honest about my conservative views is dangerous to my ability to keep my job.” – Right, English, US
“Yes, often because I am contingent (aka NTE) faculty who can be hired and fired at whim.” – Fairly left, Music, US
“I also cannot express my objection to far left movements and university policies. It will certainly get me fired.” – Right, Economics, US
“Yes, because of the tenure track system people are afraid to express their views and risk their tenure.” – Fairly left, History, US
To a greater extent than their British counterparts, North American left-wing academics voiced concerns over having their words misconstrued by their own ideological group, which might make those of their own political stripe mislabel them. A significant number of center-left academics voiced worries about intra-left pressures:
“I have avoided discussion in teaching so as not to be mis-categorized as reactionary.” – Fairly left, Natural Resources Science, US
“The big issue for us these days is racism versus other kinds of discriminatory conduct – disability, anti-Semitism, etc are not taken as seriously (if acknowledged at all). It’s a matter of just how progressive you are, with intolerance for nuance or diversity of views. I’m more senior now so braver.” – Fairly left, Law, Canada
“Yes. I used to teach a Gender Differences course in the late 90s. We were really able to talk openly about male/female differences. Now, it’s impossible to not say the wrong thing. I have definitely made what I say more ‘vanilla’ than I used to in the past. We say that education should be about exploring difficult topics but when people misconstrue things that are said it can turn one apathetic.” – Fairly left, Communications, US
“Yes, I am frequently appalled at the entitlement I hear from my colleagues but keep this to myself. They frequently sound as immature as the students they teach on political issues and engage in ridiculous shouting matches over these views.” – Fairly left, History, US
“Yes. As pre-tenure, there have been discussions in which it was clear that my opinions could affect my relations with colleagues.” – Fairly left, Musicology, Canada
“While overall, my department seems welcoming of dissenting opinions, including political views, I have the impression that certain perspectives are taboo. Some involve criticism of left wing perspectives, e.g. any critique of affirmative action or anything that is not 110% supportive of minority empowerment. But our field also punishes certain forms of left-wing advocacy, especially related to unionization, advocacy for equitable pay and tenure rights, and open critiques of older faculty with biased views (e.g. sexist views).” –Very left, Music, US
“I was asked to modify a class exercise on personal safety in low income neighborhoods because students viewed the association as racist.” – Fairly left, Geography, US
“Yes. My network is very left. I consider myself also very left, but I disagree with them on many issues. I do not feel like I belong in my left network and I do not think they are left as they claim to be. This has gotten significantly worse during the COVID crisis. I find the polarization occurring to be very troubling, and I do not agree with the left on their reaction to this current situation.” – Fairly left, Sociology, US
“Yes. A few times – more since this spring. I have not raised points where I disagree with ‘cancel culture’ and aspects of ‘post-colonialism.’” – Fairly left, Rhetoric, US
“Yes. Less a matter of refraining from asserting opinions than of being picked on for any particular formulation of those ideas – that is, using language that someone might not approve of.” – Fairly left, English, US
“Yes, my workplace is toxic, and it seems difficult to speak up about without facing retribution in terms of resources.” – Fairly left, Environmental Science, US
“I shifted my scholarly focus away from Native American writers/history because of hostile comments at conferences.” –Fairly left, English, US
“I would be concerned to speak against even the most extreme left wing ideals.” –Fairly left, Media and Communications, US
“I am sometimes thinking twice before saying things when I’m teaching or writing things in a paper that I believe can be construed as politically incorrect.” – Fairly left, Economics, US
“Avoided sharing my opinion at our weekly diversity and inclusion training meetings.” – Fairly left, Psychology PhD, US
“We were encouraged (strongly) to attend certain ‘optional’ workshops to help the BLM movements with political actions and while I agree 100% with the cause, I do not like my workplace pressuring me to take political action.” – Fairly left, Psychology PhD, US
“Absolutely it would be career suicide to say anything besides a far-left opinion out loud in front of other faculty members in our college.” – Very left, Sociology, US
“Yes – any opinions that could be seen as right of centre (especially wanting controlled immigration) would damage career.” – Very left, Journalism PhD, UK
“I feel I will be judged if fellow faculty members know my partner’s views.” – Fairly left, UK, Sociology
Transgender Issues
A second major area of concern on the left emerged from the worries of the significant number of gender-critical scholars about running afoul of approved pro-trans orthodoxy. Indeed, 6 UK YouGov respondents, 5 on the left and 1 centrist, mentioned self-censoring their views on the trans issue, representing 10% of complaints from leftist respondents. In the North American data, 11 respondents, all on the left (5 “fairly left” and 6 “very left”) mentioned self-censoring on transgender questions. This likewise represents around 10% of the stated reasons for self-censorship provided by leftist academics, among the most common complaints provided by leftists who self-censored. Here again we see evidence from a wider sample of the powerful forces arrayed against gender-critical perspectives in the academy:
“I avoid talking about trans and queer issues with colleagues.” – Very left, US
“I am fairly left wing but have avoided doing research or talking about gender and trans issues because I anticipate censure and exclusion from the left. (it would be bad for my professional relationships with colleagues even to raise certain questions.)” – Very left, Philosophy, US
“Yes, academic discussions on trans women have become dangerous at my university and one faces serious career consequences for airing them. Dr. Kathleen Lowrey was fired from her role as Associate Chair for saying women do not have penises and being gender critical. There is no room for debate on this subject at my university. If you even have a question about trans women or reject gender reassignment surgeries for children (e.g. phalloplasty, mastectomy of children under the age of 18) etc then you are called transphobic. I have been warned not to air any such opinions or my career will be ruined. Lists are apparently kept of women who air such opinions and they are prevented from advancing in their career. I was also very frightened to see Dr. Lowrey received death threats, rape threats on social media which in some cases seemed to be encouraged by faculty.” – Fairly left, Anthropology, Canada
Leftist Complaints Against Conservatives
Those who self-censored due to right-wing pressure worried mainly about administrators or students at religious and conservative universities, or about threats from the off-campus right. Some cited Middle East/Israeli politics as a third rail. Student evaluations were noted as a particular concern for some:
“As I teach at a Catholic university, I am careful in what I say about abortion.” – Fairly left, Political Science, US
“As a teacher (PhD student) I try not to say anything in front of my students that could be perceived as hostile to their beliefs, unless it is something the uni already takes a stand on (e.g. transphobic comments would be shut down).” – Fairly left, English, Canada
“I avoid political discussions with younger students because I know the university is mostly conservative, though my department is not.” – Fairly left, Psychology PhD, US
“I used to work at a very conservative Christian college in the USA, and was regularly told not to publish or discuss certain topics (e.g., abortion, evolution, gay rights, etc.)” – Fairly left, Philosophy, Canada
“I worry that my students may be more conservative than I assume and what I present as fact from my expertise may be interpreted as lefty ‘political opinions.’ I worry that speaking against Trump policies when they come up in the context of class discussion could cause complaints from students/parents.” – Very left, Art History, US
“Yes, not frequently but I avoid extremely far left opinions in case they have repercussions but also so as not to alienate students.” – Very left, English, US
“I am careful not to discuss my political beliefs in undergraduate teaching because I don’t want to alienate right-wing students.” – Very left, Communications, US
“Many of my students are more conservative than I am, so I need to be careful about what types of theories and which theorists I introduce in which order, or else students might take action against me.” – Very left, English, US
“I teach at a religiously affiliated university that tends to have a more conservative student body, therefore I try to keep my political views out of the conversation, and instead focus on research and what the science says.” – Fairly left, Psychology PhD, US
“Previously worked at a private Baptist college, teaching gender, sexuality, race & ethnicity, social problems, and cultural diversity courses – many students were openly combative to even broaching topics that challenged their beliefs and would penalize me in evaluations for teaching what is considered canon in my discipline as inherently political or heretical. Experienced at least with two or three students in each course taught (4/4 load over 2 years), if not more.” – Very left, Sociology, US
“I avoid airing my (left leaning) political views in teaching and academic discussions. I work at a public university and the state that funds it has a very conservative government. When left wing faculty air their views in class, they typically cause a backlash that is an annoyance to everyone. Their conservative undergraduate students tell their parents, who tell rich alumni donors, who tell their legislators, who then bother the administration, which can do little because of legal protections for academic freedom. For many reasons I try to separate my activism from my work. One further reason is that I don’t tend to like the ‘academic left’; they typically talk a lot and ‘cancel’ each other, but rarely do anything of political significance.” – Fairly left, Linguistics, US
Some leftist professors perceived threats from outside the campus to academic freedom. These include comments expressing fear of the activities of Turning Point UK in monitoring left-wing lecturers in Britain:
“I recently had an incident where a student expressed his concerns about the lack of right-wing views expressed by lecturers and this person also criticized my department as being left-wing. I am not afraid of what the university might do but rather I might find myself the subject of a report to Turning Point, or outed on twitter for what I’m teaching.” – Labour, Remainer
“Not at the moment, but I am concerned about Turning Point UK.” – Labour, Remainer
“With organizations like Campus Watch, and the Proud Boys, and other White Supremacist, misogynist groups, I am always very careful to keep my own activism out of the classroom. Of course, I am also careful to create a space where students who are vulnerable because of the their immigration status, gender, race, class, disability, etc. are protected from attacks.” – Very left, Sociology, US
“I once published a pro-life piece that suggested government transfers reduce abortions. I was pilloried on blogs and in private emails, including threats, by right-wing activists.” – Fairly left, Political Science, US
Middle East Politics, Israel, and UK’s Prevent duty to report on radicalization were cited by some on the left as reasons to hold back:
“Prevent makes me very uncomfortable, particularly when discussing urban uprisings and armed resistance during one of my modules.” – Labour, Remainer
“No but as a supporter of rights for Palestinians and an arranger of placements in the West Bank for students over the summer vacation (linked with their course/career choices) am mindful that it causes considerable angst amongst some others.” – Labour, Remainer
“I avoided doing research on antisemitism and islamophobia pre-tenure because these are ‘third rail’ topics in my subfield of Middle East politics.” – Fairly left, Political Science, US
“I have changed my syllabus to avoid discussions of Israel, because I expect conservative colleagues to attack me.” – Fairly left, Philosophy, US
“I am careful not to express my support for Palestinian rights, as that can lead to false accusations of antisemitism.” – Fairly left, Political Science, US
These leftist comments raise important issues about the boundaries of the Prevent duty in the UK, which can impinge on criticism of British or American Middle East policy, or of Israel. Here it is important to allow maximum academic freedom within the law. If Turning Point were to instigate Twitter mobs against particular lecturers, or aim to get them dismissed on the basis of their comments in class, this represents a clear extra-campus threat to academic freedom. Universities should not act upon student or outside actors’ complaints about lecture content unless the lecturer is blatantly quashing the expressive freedom of students for politically biased reasons or discriminating in grading.
More generally, some left-wing academics felt they could not express their views in class due to student sensibilities or pressure to remain neutral, though others felt it was important to let students know where they stood in order to have a good discussion:
“Yes, tend to be wary of expressing personal politics as management very against this. Would love to discuss it more as seems deeply patronizing not to – I teach adults.” – Labour, Remainer
“I try to avoid making any political statements when teaching.” – Labour, Remainer
“It is generally taken that lecturers do not air their personal political views in influencing students. I have had some lively discussions with students over this issue particularly in relation to health care. I have had to temper my own views in the University setting. When undertaking research using IPA [Interpretative phenomenological analysis] I had to ask the university not to make my work readily available to students as I had to set out my preconceived ideas and biases which included my political beliefs. In an increasing litigious era, lecturers/researchers need more protection as students seem to use any angle in order to bolster erroneous claims.” – Labour Remainer
“I avoid explicit discussions of my political views in some of my courses out of fear of student retaliation. I lost my first (and only) tenure track job, ostensibly for poor student evaluations. Oddly, two other queer faculty were denied either tenure or promotion that same year, ostensibly because of poor student evaluations. I know several people who expressed feminist views or were out to their students who had the same experience. Now I am in a renewable contingent position, and I am VERY careful about what I say in the classroom since I know that the ostensibly ‘liberal’ administration can, at any time, just decline to renew my contract.” – Fairly left, Anthropology, US
Finally, some far-left professors mentioned that they refrained from open activism. Here is evidence that in some departments, progressive advocacy is frowned upon:
“Speaking out about issues related to social justice has gotten me reprimanded at my workplace, specifically around race, gender, and gender identity. I have been approached by individuals who do not know me well, and told that I should be less outspoken. Although the department is relatively liberal, they are less so when it comes to individual interaction and dealing with their own bias.” – Very left, Psychology PhD, US
“I frequently have to be mindful of how I express my views on racial diversity in science. I do not want to be labelled an ‘activist,’ though I work hard towards increasing diversity in science. However, the way one must talk about the enduring effects of white supremacy and racism on black people specifically is sometimes too uncomfortable for many of my white colleagues to hear without some softening.” – Very left, Psychology PhD, US
“I have refrained from calling out racist classmates because I wanted to avoid drama. It happens once every few months.” – Fairly left, Healthcare, US
These final sets of comments show that an ethic of impartiality holds, at least to some degree, in the classroom and in departmental settings. While this could be seen as a “despotism of custom” that infringes on the expressive freedom of academics, one could argue that in these instances, an ethos of neutrality counteracts the chilling effect on those with dissenting viewpoints, increasing support for a free speech culture. A balance must be struck between allowing maximum freedom for lecturers and colleagues whilst being mindful of academics’ power (i.e., over grading, class atmosphere, departmental environment) to increase the considerable degree of self-censorship that already exists among conservative staff and students. Though chilling effects among conservative students are mainly due to peer pressure and social media rather than worries about lecturers marking conservatives down, the latter is not, as we shall see, entirely absent from the classroom.89
Due to the position of the lecturer as an arbiter of a student’s grade, norms like impartiality arguably expand the degree of liberty in a university even if they limit the expressive freedom of lecturers at the margin – just as norms of debate which limit heckling or frown on excessively unruly or opinionated answers permit a richer exchange of views. What is vital, however, is that collegial norms rather than formal sanctions or mob cancellation be deployed: lecturers who express their political views in class must not be disciplined, even as it should be permissible for peers to criticize or urge them to work towards a politically non-discriminatory class atmosphere.
Section IIc: Political Discrimination
Political Discrimination: Quantitative Evidence
The above data show that many conservative and Leave-supporting academics, especially in the social sciences and humanities, feel estranged from their departments. Half in Britain and 70% in the US say there is a hostile climate for their beliefs in their departments and a similar number report self-censoring in teaching and research. Twice as many identify against, as with, their departmental cultures. Among current staff in the social sciences and humanities who voted for Trump or Leave, over 8 in 10 say they are not sure they would be comfortable expressing their views to colleagues (64-97% uncomfortable, 3-18% unsure).
Is this mere perception, or is it grounded in reality? One clue that such fears are reasonable lies with the fact that only a minority of Remain and Biden-voting academics thought a Brexiteer or Trump voter would feel comfortable expressing their views to colleagues. If, as I will show, there is evidence for discrimination against conservatives, or even a willingness to expel a staff member for their views, then it is perfectly understandable for conservatives to conceal their beliefs in teaching, research, and collegiate conversation.
In what follows, I turn my attention mainly to the second “iceberg” pyramid in Figure 4 (perpetrator’s perspective). This concerns the leftist and centrist academics who comprise about 90% of the professoriate and largely set the atmosphere at work. In the social sciences and humanities in Britain, about 60% of a conservative’s colleagues will identify as left (45% “fairly left” and 15% “far left”) and 30% as centrist. In America and Canada, my survey would suggest that 75% are on the left (including 25% “far left”). To what extent is the leftist majority likely to discriminate against, or even seek to force out, conservatives?
I broadly follow the methodology of three existing studies in the United States (Inbar and Lammers 2012; Honeycutt and Freberg 2017; Peters et al. 2020).90 These ask people if they would politically discriminate in hiring, on grant applications, and in refereeing a journal article. These studies found substantial political discrimination. Inbar and Lammers, in a 96% left- or moderate-leaning sample of psychology academics, discovered that 38% were at least “somewhat” likely to discriminate against a conservative job applicant. The share willing to discriminate against a paper or grant taking a conservative perspective was 19% and 24%, respectively. Honeycutt and Freberg, surveying those in a wider variety of disciplines, found that among left-wing academics, 16% would discriminate against a conservative paper, 22% against a conservative grant application, and 33% against a conservative job applicant.
But these authors also found that academics on the left and right discriminated against each other in equal measure: among conservatives, 21% would discriminate against a left-oriented paper or grant application, and 32% against a left-wing job applicant. Peters et al. (2020), in a 75% left-leaning sample of European philosophers, found that 48% of academics and graduate students would discriminate against a right-wing hire and 35% would discriminate against a right-wing grant application. Leftist philosophers were more discriminatory than right-leaning philosophers by a full 10-15 points.
My approach follows much of the thrust of these studies but differs in three important ways. First, previous studies contacted academics directly by email, collecting a convenience sample of around 500-600 applicants. Their samples achieved a response rate of around 25%, introducing the possibility of self-selection, though this does not present a major problem for comparing between groups in their samples. This is the method I use for US and Canadian academics, though I cast my net wider and have a lower response rate than previous studies that targeted much narrower subgroups drawn from academic bodies or lists.
For Britain, I drew on YouGov’s panel of those who completed a range of different types of surveys for remuneration and just happened to be academics or retired academics. They are thus less likely to be selectively attracted toward filling out a particular survey. Between 61% and 76% of YouGov’s panel of professors and lecturers responded, resulting in a sample that is more likely to represent the actual population of academics than any other study (UK or US) to date.91 I also surveyed British and North American PhD students on the Prolific Academic platform, where I was able to achieve 86% of the target pool in Britain and 63-72% in North America.
Second, I used an experimental survey design called a list experiment across all samples in order to circumvent the social pressures against admitting to open discrimination.92 List experiments work as follows. Consider Table 3. Half the sample receives list 1 and half gets list 2. List 1 contains 3 statements and list 2 contains the same 3 plus a statement on whether the respondent would discriminate. Subjects are asked how many of the statements they agree with. These statements are designed to tap a number of equally contentious questions, and permit those of different ideological views to find something to agree with. Because people are not questioned directly about any one of the items, it is not possible to know which individual would discriminate or not, but it is possible to calculate an average level of discrimination across the sample. This anonymity is what allows people to answer questions free of pressure to adhere to social norms.
Respondents are allocated randomly into one of the two treatment groups in Table 3. If there is no discrimination, the average number of statements people agree with should be the same in both lists. In this case, assume that people agree with two statements of three in the identical lists. List 1 therefore has an average of 2, so if list 2 has an average score of 2.5, this means that half of survey respondents agreed with statement four and are thus willing to discriminate. The difference between the two lists, .5, would represent the average discrimination level in the sample.
I can subsequently restrict my analysis to subgroups such as Trump or Clinton voters, Leavers or Remainers, men or women, and so on, so long as there are sufficient numbers of individuals in each to learn something. In this case, I would be looking at, say, all men who took the survey, comparing scores for men in list 1 with those in list 2. Just to be completely sure that results are not driven by chance differences in who gets which list, I can use statistical models to control for major differences such as age, income, gender, discipline, work status, political views, and so forth.
A third difference from existing studies is that, due to the list experiment, my question is more direct. Rather than ask whether people would discriminate on a 7-point scale from 1-Not at all through 4-Somewhat and 7-Very Much, mine is phrased as a simple Yes/No/Don’t Know. Thus, I asked British respondents whether they would agree with this statement: “If a known Leave supporter applied for a job at my workplace, I would try to avoid hiring them.” Just 10% of the YouGov study said they would avoid hiring the Leaver, 76% that they would not, and 14% replied that they didn’t know. Among Remain and Left voters, the share willing to discriminate inched up slightly to 12%. While these figures are concerning, they are considerably lower than the numbers obtained in two previous US-based studies. Much of this probably derives from differences in question wording between the 7-point scale and my direct yes/no item. Some may also be due to the nature of my sample, which is arguably more representative of the sector as a whole, encompassing lower-ranked universities and not just top research schools.
In the US, 22% of academics admitted they would discriminate against a known Trump supporter in a direct question (“If a known Trump supporter applied for a job at my workplace, I would try to avoid hiring them”), while in Canada, 26% openly said they would do so. Note that American politics receives considerable coverage in Canada, and Canadian academics readily understand, and arguably identify with, American political categories. Finally, note that these questions were all asked well before the November election and the Capitol Hill riot of January 6, 2021.
Turning to the results, those in list 1 in Table 3, who were asked how many among a list of 3 statements they agreed with, were subsequently asked the discrimination question as a stand-alone item. The difference between this and the revealed discrimination in list 2 helps us understand the extent to which people are concealing their actual willingness to discriminate.
Smaller datasets are less reliable for list experiments because there is natural variation in answers to the three emotive but unrelated list questions that only dissipates with sufficient cases. Accordingly, Figure 99 presents results that are significant at the 1% level across the five larger surveys, and in only one survey – the smallest – is the difference between the lists not significant. This suggests that political discrimination against right-leaning academics is pervasive.
Comparing the difference in average score between the 3-item list and the 4-item concealed list with the average score from the one open question on discrimination, I find that the actual level of discrimination among North American academics is nearly twice as large as stated openly, and over three times as large in Britain. Though this is a dispiriting level of bias, the level of concealment is a positive sign insofar as it suggests that most of those who discriminate sense that this is problematic and must be kept hidden. By contrast, PhD students seem not to be concealing their bias to the same degree. This may be due to weaker anti-political discrimination norms at the PhD level, though the level of discrimination is low in the small UK PhD sample.
In North America, the list experiment reveals that 40% of American academics and 45% of Canadian academics would discriminate against a known Trump supporter. These numbers are approximately twice as high as academics’ stated willingness to discriminate in the direct question.93 Among 396 North American PhD students surveyed, 33% said they would discriminate against a Trump voter in a direct question and 35% in the concealed list condition – the narrower difference between conditions may well be due to the smaller sample, around 200 in each control or treatment group.
In the UK, 32% of academics would avoid hiring a Leaver who applied for a job, whereas only 10% would admit it directly. I therefore assume a concealment multiplier of up to three across similar subsequent UK questions on direct discrimination.
Figure 100 shows the average willingness to discriminate against a Leave supporter across various subgroups in the UK YouGov sample. The leftmost bar reveals that 32% of the sample would discriminate against a Leaver. Thirty-seven percent of Remain supporters, 35% of those identifying as left, 39% of 2019 Labour voters, and 44% of female academics revealed that they would discriminate against a Leave applicant.
On a five-point scale, 23% of the sample agreed with the statement “I would consider myself an activist.” Fifty-eight percent of self-described activist scholars would discriminate against a Leaver compared to 20% of faculty who disagreed with the statement (i.e., are non-activist). Surprisingly, 34% of centrists also revealed that they would discriminate, which runs counter to findings in the previous literature which found centrists to lie intermediate between those on the left and right in their propensity to discriminate. While the differences in willingness to discriminate between major political groups are all statistically significant, the male-female difference only reaches borderline statistical significance, so we should not place much emphasis on this finding.94 Age, and whether a respondent is a professor or lecturer, active or retired, or in the SSH or STEM sector are not significant predictors of discrimination.
The fact that nearly a third of academics are willing to discriminate against a Leave applicant is highly concerning. This means that there will be one biased member on a three-person appointment committee, two on a six-person panel and ten in a 30-strong department, a number of whom will evaluate CVs for a longlist and many of whom will attend job presentations where the departmental “steer” strongly shapes the final choice.
On the positive side, just 10% are willing to openly admit to discrimination – 17% among activists. More importantly, two-thirds of left-wing academics and over 60% of Remainers would not discriminate against a Leaver, even when this view is completely concealed. Younger staff are no more likely to discriminate than the old, suggesting that the problem is not getting progressively worse. This is an important finding in itself, because it indicates that people are unwilling to admit openly that they discriminate on political grounds. This suggests that there is something of a norm against political discrimination, even though it is flouted in practice by an important minority of staff.
In the United States, Figure 101 shows that most of the same academic subgroups that scored highly on political discrimination in the UK also emerged: activists and leftists are more discriminatory than non-activists and non-leftists. Women are somewhat more discriminatory than men, but this did not reach conventional statistical significance. While a 20% minority of right-leaning academics also said they would discriminate against a Trump supporter, it is difficult to be confident of this due to low sample size (which also affects the small group of pro-Trump academics).95
I also asked about people’s willingness to discriminate against a right-leaning term paper from a student. Overall, there is much less political discrimination in marking, with just 4% of North American academics openly admitting they would discriminate, rising to 7% in the concealed condition. This number masks large differences between American (15-16% willing to discriminate in concealed condition) and Canadian (-2%, i.e., effectively zero) respondents. The American figure, though not high, is approaching a concerning level, and guidance should be issued against political discrimination.
The most concerning of all, however, are the results for US and Canadian doctoral students, with 12% openly admitting they would discriminate against a right-leaning term paper, rising to 35% in the concealed condition. Moreover, doctoral students are as likely to give left-leaning term papers a higher grade as they are to allocate them a lower mark (7% for both). The sample is not especially large (338) and there is a margin of error on list experiments with smaller samples.
Do Left and Right Discriminate Equally?
The list experiment tested for bias against Leavers only. This begs the question of whether conservatives and liberals discriminate equally, as Freberg and Honeycutt (2017) suggest, or whether this problem is accentuated in particular parts of the political spectrum, as Peters et al. (2020) find where leftist academics were sometimes nearly twice as likely to discriminate as those on the right.
The contact hypothesis suggests that those with greater intergroup contact become more tolerant of difference.96 Right-wing academics encounter left-leaning scholars and their writing frequently. It is difficult to reject a majority of one’s colleagues and their output, and assessing it is a routine part of one’s academic job. By contrast, left academics meet conservative academics or encounter right-leaning content much less frequently. Any negative reactions they experience are not meliorated by many positive interpersonal contacts with known conservatives on the faculty. In short, it is more likely that left-wing academics will create and maintain a homogeneous ideological network, impeding interpersonal contact with those who have opposing views. This latter interpretation would accord with evidence from Britain that Remain voters and those on the left are now considerably less comfortable with Leavers and conservatives than the reverse. Those on the left are more likely to unfriend online, socially distance from, and wish for their children to avoid marrying, those on the right.97 Thus we would expect the left to discriminate against the right more than the reverse.
Against this interpretation, the fact that those on the right are a small minority may incline them to be more self-conscious and mobilize to favor their own. This would lead us to predict that the right-wing minority discriminates against the left-wing majority more.
To interrogate this issue, I asked a series of questions to tap both “hard” and “soft” discrimination. Hard discrimination is discerned through the question, “In choosing a job candidate, I would be inclined to support a known centrist over a known leftist with a slightly stronger track record.” The term “leftist” was then replaced with “conservative” to test for anti-conservative bias. This is “hard” discrimination because it involves giving a job to a weaker candidate for political reasons.
Soft discrimination – in which political affiliation is used as a tie-breaker – is measured by a question which asks: “In choosing between two equally qualified job candidates, one a Corbyn/Sanders supporter and another a Leave/Trump supporter, if you had to pick between them, who would you be inclined to choose?” Respondents could pick one of the two, or state “no preference”/”don’t know.” The last was the most common choice.
Political Discrimination in Britain
In the UK, around 30% of Leavers would choose a Leave supporter over a Corbyn supporter, and 29% of Remainers would choose a Corbyn supporter over a Leaver, if both were equally qualified. In both cases, the majority of academics said they would have no preference and would still not favor one of the candidates.
Focusing only on currently active SSH academics shows more animus among Remainers than Leavers: the YouGov survey found 37% of Remainers (N=191), but just 27% of Leavers (N=22), willing to discriminate against a hire from the other side of the Brexit divide. The mailout survey, limited to current SSH academics, shows a similarly lopsided picture, with just 11% of Leavers (N=19) but 22% of Remainers (N=142) willing to discriminate against a hire from the other side when the two are otherwise evenly matched. Here the Peters et al. 2020 findings of lopsided bias appear to be borne out for the Leave-Remain divide, though we shall see that this does not hold for the left-right divide.
UK results by ideology are presented in Figure 102. These show that 50% of right-wing and 40% of left-wing academics preferred their own candidates (as did 38% of Labour voters and 33% of Tory voters, not shown) where credentials are evenly matched. Looking at measures of “hard” discrimination, 20% of right-wing and 15% of left-wing academics would discriminate against those of the opposite camp who had better records. In addition, numbers (not on chart) show that 12% of Tory voters and 13% of Labour voters would discriminate against the other side. Twelve percent of Remainers were willing to discriminate against conservatives with stronger track records than a centrist, while 10% of Leavers would discriminate against leftists with stronger records than a centrist.
The mailout survey of current SSH academics shows that 20% of the very few right-leaning academics would discriminate against a leftist hire, and 12% of leftists would discriminate against a right-leaning hire. Thus, if anything, there seems to be somewhat greater bias from the right this time as compared to the left. One pattern that we also see in the North American data is that discrimination against concrete populist categories like Leaver or Trump supporter is greater than against more abstract and non-specific “conservatives” or right-wingers – a category that includes the less controversial libertarians.
When I control, in Figure 103, for professor or lecturer status, SSH or STEM, age, and gender, and look only at the 5-point left-to-right question, I find that both sides engage in approximately similar levels of discrimination in hiring. This is especially so when I take the measurement error that stems from small samples of “very right” academics into account. This suggests that, following Freberg and Honeycutt (2017), each side is similarly willing to discriminate against the other in hiring.
Does this mean there is nothing to worry about? Hardly. From my list experiment, I know that 12% of Remain-voting academics say they would discriminate against a Leaver in a job, but the actual figure is 37%. So, the true levels of willingness to discriminate could be up to three times higher than stated.
Second, because Leaver (16%), Tory (15%) and right-wing (9%) academics are a small minority of my British survey respondents, there is structural discrimination against them. Though individuals discriminate the same in hiring, the collective (“structural”) impact of a left majority and small conservative minority in academia means that there is three times as much discrimination against Leavers (24.9%) as against Corbynites (7.4%), and nearly twice as much discrimination against conservatives as against the left. Each group is equally biased, but the compositional effect is to slant the playing field against Leavers and conservatives. This is a similar finding to previous studies.
American results in Figure 104 tell a mixed tale. On the one hand, political discrimination against populist right supporters is worse: 56% of left-leaning academics said they would favor a Sanders supporter over a Trump supporter when the two had equal merit, and only 25% said the candidates’ politics would not affect their decision (in Britain, by contrast, 52% of leftist academics said politics would not affect their decision).
Moreover, 21% of American right-leaning academics said they would back a Trumper over a Sanders supporter, considerably lower than the 56% anti-Trump soft discrimination coming from the left.
This might suggest that the American academic left engages in 2-3 times as much soft discrimination against Trump supporters as conservatives do against Sanders supporters, backing the Peters et al. 2020 story of disproportionate left-wing bias. On the other hand, 17% of conservatives and 16% of centrists would discriminate against a leftist hire whereas only 14% of American academic leftists would discriminate against a conservative hire. These findings are more in line with Honeycutt and Freberg’s 2017 finding that discrimination flows both ways, in equal measure.
In terms of the net effect of discrimination among all academics, Trump supporters are heavily disadvantaged in soft discrimination compared to Sanders supporters by a whopping 46-2, showing that for two equivalent candidates, Trump supporters will be strongly discriminated against. Note as well that just 34% of American respondents said that the politics of the candidates would make no difference to their decision, compared to 62% for the equivalent question in Britain.
However, when it comes to discriminating against a hire who has better credentials, the net discrimination against conservatives compared to the left is just 12 versus 9%, a more positive picture than in the UK. This is partly because of the role of centrist American academics, who resemble conservative academics in discriminating nearly twice as much against the left as against the right. By contrast, British centrists align closer to the left, discriminating disproportionately against conservatives and Leavers.
In addition, this question asks about hiring a “conservative,” rather than focusing on the more charged populist category of Trump supporter. It may be that left-wing academics in America make more of a distinction than British leftists between cultural-populist forms of conservatism and others such as fiscal or foreign policy conservatism. The American left has arguably been more focused on cultural categories of disadvantage while the British left has traditionally been more materialist and class-based.
Canadian results in Figure 105 show a pronounced anti-Trump bias, as in America, and a considerable anti-conservative slant, as in Britain. This includes a sharp 62-17 tilt in favor of hiring a Sanders supporter over a Trump supporter to break a tie, with just 32% saying the candidates’ politics would not influence their decision. While conservative academics in Canada have a higher propensity (25%) to discriminate against a leftist than Canadian leftists do toward conservatives (18%), the right make up just 4% of academics in the Canadian sample. Meanwhile, Canadian centrists discriminate evenly between right and left, conferring no advantage on conservatives. Hence when it comes to net discriminatory effects in hiring, conservatives face a penalty of 16% discrimination in Canada compared to just 7% for leftists: worse than the 5:3 ratio in Britain and the 4:3 ratio in the US.
The UK academic mailout survey shows a 21-1 Corbyn over Leave advantage in the case of a tied set of candidates, thus a strong leftist advantage. However, 69% of UK academics said they had no preference among the candidates, and would not take their politics into account. Even among leftists, 61% of UK academics in the mailout survey said politics would not affect their choice. These reinforce YouGov findings showing that most UK academics would not engage in soft discrimination, unlike their US and Canadian counterparts.
These transatlantic differences largely replicate among PhD students. For the 338 North American PhD students in our sample, 65% would choose the Sanders supporter over the Trump supporter, 5% would opt for the Trump supporter over the Sanders supporter, and only 25% said they would have no preference. The numbers for SSH PhDs on the left are extreme: 82% would select the Sanders supporter, 14% said no preference, and none would choose the Trump supporter.
UK PhDs are more even-handed, with 46% saying no political preference, as compared to 40% for the Corbyn supporter over a Leaver, and 6% for the Leaver over the Corbyn supporter. Focusing on the subset of British SSH leftist PhDs, 59% would choose a Corbynite over a Leaver, but 31% said they would not have a preference.
Norms against political discrimination appear to be somewhat stronger in Britain than North America. While Trump’s personal antics do add a distinct dimension compared to Leave, which may account for some of the difference, one could equally argue that charismatic Leave standard-bearers like Nigel Farage and Boris Johnson also arouse strong criticism. Moreover, the powerful partisan polarization in America cannot explain why Canadian findings resemble those of the United States. A better explanation is that partisans in both countries are exposed to similar US media outlets, which interacts with their ideological biases.
Transatlantic differences in the propensity to discriminate are likewise evident with respect to right-versus-left hiring bias. Among North American PhDs, 35% would discriminate against a conservative hire with a stronger track record than a centrist while 14% would discriminate against a leftist with a stronger record than a centrist.
Among British PhDs, 16% – a much smaller number – would discriminate against the conservative hire compared to 7% against the leftist hire. Whereas 42% of North American leftists would discriminate against a better-qualified conservative, just 20% of British leftist PhDs would.
Overall, when it comes to hiring, academics are more even-handed than PhD students, and British PhD students are less biased than North American PhDs. Though each side discriminates against the other in fairly equal measure, the net effect of academia’s leftward skew is to produce more discrimination against conservatives. This is especially the case among PhD students where the net discrimination ratio slants two and a half times against the right.
These patterns are summarized in Figure 106. Another differentiating factor between the two sides of the Atlantic may be a greater skittishness among Americans and Canadians about selecting candidates with an open political skew – whether on the right or left: North American academics and PhD students are more likely than British equivalents to be biased against a left-wing hire as well as a right-wing hire.
Is there Bias in Refereeing Grant and Promotion Applications and Journal Articles?
As in other studies, I find clear evidence of political discrimination in grants and papers, with the playing field strongly tilted against conservatives. Here I ask, following previous studies, “If I was refereeing a grant application, I would be inclined to rate it lower if it took a politically right-wing perspective,” altering “right” to “left,” and “grant” to “paper” in subsequent questions. Grants often have just a 5-10% success rate and are hotly contested because they involve substantial rewards in the form of prestige and research resources, and a zero-sum contest between individuals and departments could be leading to an elevated level of bias among assessors.
American results in Figure 107 show that 24% of leftist academics would rate a right-leaning grant lower while just 16% of right-wing academics would rate a left-leaning grant lower. However, in terms of papers, right and left discriminate against each other at a similar rate (13-14%), and for promotion, right-wing academics are somewhat more likely to discriminate against the left than vice versa (16% vs. 13%).
Weighting for the twofold concealment multiplier, this results in 26-48% of American left-wing academic staff discriminating against a right-leaning promotion, grant, or paper and 26-32% of those on the right discriminating against their left-leaning equivalents.
Despite each ideological segment displaying similar levels of bias against the other, the structural effect of discrimination in the American academy is, as in Britain, slanted against the right. Again, this is because of the leftward skew of the faculty. Thus the chance of a right-leaning paper facing bias overall among American academics is 12% compared to 5% for a left-leaning paper submission. On grants, 20% of right-leaning grants face discrimination compared to 9% of left-leaning grants. For promotions, the ratio is 12 to 6 (i.e., 2:1). In all cases, the right faces at least twice as much structural discrimination. In addition, we saw from our concealed list experiment that actual bias against a Trump supporter is nearly twice as high as what is stated in an open question.
If the actual rate of discrimination is double the amount openly admitted, then right-leaning papers, grants, and promotion applications face a 24-40% chance of discrimination from any given assessor. Multiplied across 4 panelists or assessors, this suggests that academics on the right will face discrimination in the overwhelming majority of paper submissions, grant applications and promotion bids. By contrast, the 5-9% overall rate of discrimination against the left amounts to 10-18% chance of discrimination from any one assessor given the concealment multiplier.
Furthermore, anti-left bias is offset by discrimination in favor of the left. For instance, 6% of American academics would rate a left-leaning paper lower, but 3% would rate it higher, because of its leftist stance. On promotion applications, 5% would rate a left-leaning promotion lower than if it did not adopt a leftist perspective, but 2% would rate it higher because of its leftist outlook. Right-leaning papers and promotion applications are, by contrast, only rated higher by 1% of academics, conferring little counterweight to discrimination. Combining the US and Canada, 5 of the 55 right-wing academics in the sample (9% of the total) rate right-leaning promotion applications lower while 1 (2%) rates them higher. For papers, 4 rate right-wing papers lower (7%) and just 3 higher (5%). Thus, even if one wanted to get the attention of right-wing reviewers, there is no benefit to signalling a right-wing orientation.
In Canada, the pro-left premium is greater: 3% on papers, offsetting the 3% who would mark it lower. For promotions, a left-leaning application gains points from 4% of respondents while just 2% would mark it down. This results in net positive discrimination in favor of those on the left who signal their political beliefs.
In Britain, left-inclined groups (Remain, Left, Labour) are between three and five times more likely to rate right-leaning grant applications lower than left-leaning grant applications. Right-leaners (Leavers and Tories) are between one and two times more likely to rate left-leaning grants lower than right-leaning grants. This lower level of discrimination is partly due to the fact that only 40% of Leave and half of Tory academics in my sample identify as right-wing. Right-wing academics are six times more likely to rate left-leaning grants lower than right-leaning grants. Even so, the highest level of discrimination reported by right-wing reviewers is 18%, compared to 30% for left reviewers. Here I do find the left to be more biased than the right.
A similar pattern can be discerned for refereeing papers. Although the average level of bias is much lower here, this may be due to the different question wording for the grant and papers questions. All told, these figures indicate that while both sides discriminate, the left does so at a higher level. This runs counter to findings for grants and papers in the Freburg and Honeycutt study, but is in line with the work of Peters et al. (2020).
When weighted for the left’s substantial demographic advantage within the professoriate, the net result is a substantial anti-conservative slant. Across the entire sample, grants adopting a right-wing perspective are discriminated against by 22% of academics whereas those taking a left-wing stance only face bias from 9% of the professoriate. For journal articles, conservatives face discrimination twice (9%) as often as leftist authors (4.5%). These figures don’t seem high until I consider my multiplier of 3.2, derived from the list experiment where just 10% of UK academics admitted they would discriminate against a Leaver, but the revealed percentage in the list experiment is 32. Applying this multiplier would suggest that for any given academic who submits a manuscript for publication, there is a 30% chance each reviewer will downrate a right-leaning paper.
It would also suggest a close to two-thirds likelihood that each reviewer of a right-leaning grant application will engage in political discrimination. However, it is likely that the share who would discriminate against a conservative-oriented grant application is less than two-thirds because, first, the wording of the grant question offers two discrimination points (“strongly agree” + “tend to agree”) that are amalgamated, whereas the paper and promotion questions, like that of hiring, are phrased in a binary “rate it lower/rate it higher” manner.98 Second, as the share of biased reviewers rises over 50%, it runs into an increasingly resistant ceiling of meritocratic scholars. A realistic estimate would suggest a multiplier of 2 rather than 3.2, thus 44% of reviewers would discriminate against a politically right-leaning grant. Nonetheless, this is an astounding figure.
Finally, on promotion applications (see Figure 108), I see an intermediate pattern, with 21% of left-wing reviewers willing to rate right-leaning promotion applications lower as compared to 7% of right-wing reviewers who would rate a left-leaning application lower. Here again, with the multiplier, over 40% of left academics assessing right-leaning promotion applications would discriminate against them.
A promotion application with a right-wing perspective will, on average, be ranked lower. Since there are typically 5-10 referees (including the adjudication panel) for a promotion application, there will be, on average, 2-5 voices in the room discriminating against a right-wing candidate. This may be mitigated to the extent that the applicant is able to name referees, but named referees make up a minority of references – especially when the adjudication panel’s votes are included. A paper is also unlikely to be judged strictly on its merits since most journals require at least two referees plus an editor to take a look. This means there is a 60-90% chance of a right-wing paper being rated lower.
While 4.4% of UK academic respondents would rank a left-wing promotion application or paper lower, 3.3% would rank it higher. Thus it makes almost as much sense to advertise one’s left-wing beliefs as to conceal them. American academics stand out, as compared to Britons and Canadians, for their willingness to rate papers and promotion applications lower if they bear a political cast of any kind.
Discrimination by PhD Students
Doctoral students are more discriminatory than academics on these measures. In North America, 24% of all PhD students say they would rate a right-leaning paper lower, 30% would mark a right-leaning promotion application lower, and 33% would rank a right-leaning grant application down. The level of bias in favor of left-leaning papers is also greater than among academics. Ten percent of all North American PhD candidates would rate a left-leaning paper higher while 8% would rate it lower, producing a net incentive for left partisans to signal their beliefs in articles. For promotion applications, 10% say they would rank left-leaning ones higher while 10% say they would rank them lower, producing no net discrimination against an openly left-leaning promotion.
In Britain, 9% of PhD students would rate left-leaning papers higher while just 4% would rate them lower. For promotions, left-leaning applications gain points from 9% of British PhD reviewers and lose them from just 5% of them. Twenty-eight percent of British PhDs would rate right-leaning grant applications lower, 22% would rank a right-leaning promotion application lower, and 20% would mark a right-leaning paper down, with almost none in the opposite direction, leading to high net negative discrimination.
Figure 109 summarizes the level of open anti-conservative bias across five surveys. These figures should be multiplied by a factor of up to two, though the level of social desirability bias in the responses is, given the results of our list experiment, likely to be substantially higher among academics than PhD students. This does not however take away from the fact that norms of non-discrimination appear to weigh more heavily among academics than PhD candidates. Notice as well that grant bids, which involve substantial monetary research resources, attract the highest level of political discrimination across all surveys.
Combining six questions – each collapsed into a yes/no (coded 0/1) discrimination binary on hiring, promotion, and refereeing – permits me to examine which factors predict discrimination against left and right. Controlling for age, gender, SSH or STEM, and identifying as an activist, there is significantly greater discrimination from left academics in Britain and “very right” academics in the US, with no ideological group in Canada standing out as more discriminatory. Bearing in mind sampling noise, these results seem to indicate that academics of different ideological stripes discriminate against each other fairly evenly, as per Honeycutt and Freberg (2017).
From the evidence, it is unclear whether theories about the left’s lack of contact with conservatives leading to misperceptions of the out-group, or the right’s heightened consciousness about being an embattled minority, play the bigger part in motivating discrimination. Since both seem to discriminate against each other at similar rates, a third possibility arises: it may be that partisan bias runs deeply on both sides, regardless of contact or professional context. Some evidence for this interpretation will follow when comparing academics with non-academics where I find that ideology and activism predict political discrimination just as much among non-academics as academics; even though the work context in other industries is more politically balanced and less politicized.
Who Discriminates?
Beyond ideology, are any other factors associated with higher political discrimination? Running relatively similar models of discrimination (based on a 5-6 item composite index covering hiring, promotion, grants, and publication) across five datasets in Figure 110, what comes across is how dominant ideology is when picking out those who discriminate. Compared to centrists, very left academics and PhD students discriminate the most, followed by those who are fairly left.
Once we control for ideology, being an activist only predicts higher discrimination for the North American academic survey, though the coefficient is in the expected direction in all but one survey. Women and minorities are significantly more likely to discriminate against conservatives when we control for ideology, but only among American and Canadian academics.
SSH academics and PhDs don’t discriminate more than their STEM colleagues once you control for their more left-wing ideology and higher activism. It’s also worth noting that in models of discrimination against the left, the only variable that matters is being right-wing when it comes to identifying who will discriminate. Other factors, such as age or SSH vs. STEM, do not reach conventional measures of significance.
Older academics are marginally less likely to discriminate than the young, but this is only statistically significant in one model. In other words, net of ideological differences, we should not expect more discrimination against conservatives in the future except insofar as recruitment to the profession pushes academia further left than it already is today.
Soft Authoritarianism: Chilling Effects of Discrimination
The combination of individual-level discrimination by both sides with a heavy left-leaning majority produces a high degree of system-level bias against conservatives. These findings mean that conservative scholars who self-censor are not paranoid, but acting rationally. A sufficiently large proportion of academics are willing to penalize work that is right-leaning to make it prudent for conservatives to hide their views. This substantiates with data the repeated testimony that there is a climate of political discrimination inside the contemporary university. If conservative academics wish to have papers accepted for publication, to be awarded grants, or to be promoted, it is wise for them to conceal their political views.
Summary of Discrimination Effects
To summarize the discrimination effects, I constructed an index of seven discrimination-related questions for the data, with four questions on hiring, and one each on refereeing a grant application, promotion case, or paper. In the United States, 74% of SSH academics would discriminate against the right, conservatives, or Trump supporters on at least one dimension. For Canada, the equivalent figure is 75%.
Discrimination is about 20 points lower in Britain. Using the UK YouGov data pertaining to Leavers or Conservatives, 44% of the sample would discriminate on at least one of the seven questions. This rises to 54% of currently-employed academics in the social sciences and humanities, 67% of far-left academics, and 74% of currently-employed far-left SSH academics. The UK SSH mailout survey shows a slightly lower share who would discriminate on any dimension: 42% of SSH academics and 66% of far-left academics.
With discrimination this pervasive, it is little wonder that conservative scholars tend to avoid writing papers that signal a conservative perspective, while left-leaning authors tend to openly advocate theirs. A recent study based on a random sample of American legal scholars, inferring political leanings from political donations, found that among the small minority of Republican-supporting academics, work could not generally be assigned a political coloring by coders. By contrast, much of the output of Democrat supporters was easily identifiable as progressive. The authors go on to suggest that:
The most plausible explanation is that if the dominant ethos in the top law schools is liberal or left-wing, then Republicans are likely to conceal their ideological views in their writings. Republican professors might fear that scholarship that appears conservative may be rejected by left-leaning law review editors, and disparaged or ignored by their colleagues, which will damage their chances for promotions, research money, and lateral appointments. This would explain why even [registered Republicans] tilt left. Republicans could suppress their ideological views by avoiding controversial topics, taking refuge in fields that have little ideological valence, focusing on empirical or analytical work, or simply writing things they don’t believe.99
The data from my study indicate that a similar dynamic is at play within the social sciences and humanities more broadly, in both North America and Britain.
Political discrimination greatly impoverishes scholarship and teaching by narrowing the range of research questions that are asked, funded, and pursued, and circumscribing the number of acceptable interpretations of the facts. This is especially true of research areas that touch on sacred progressive values around race, gender, or sexuality. For instance, there has been comparatively little work on left-wing authoritarianism in social psychology, compared to the wealth of scholarship on right-wing authoritarianism. The reasons are fairly clear. Most scholars are progressive and thus have little interest in examining left-wing forms of discrimination and authoritarianism. Second, investigations of left-wing bias or authoritarianism are likely to face more resistance – whether in grant applications or paper submissions – than those focusing on right-wing forms.100
Collegiality is strongly encouraged in academic departments, and provides important career and social benefits to academics, including the minority of right-leaning scholars. Cass Sunstein argues that the more of a social character an institution or field has, the stronger its mechanisms of conformity to group norms will be, and the lower its performance. In such an environment, Sunstein explains why self-censorship occurs: “Group members who care about one another’s approval, or who depend on one another for material or nonmaterial benefits, might well suppress highly relevant information [for organizational functioning].”101 This applies in academia, as it comes to have more of a confessional hue, and one which is defined on the basis of political outlook. Collegiality has its benefits, but one of its downsides is that it fosters the social conditions that can sustain a culture of conformity, rather than one that encourages dissent.
Social Aversion to Conservatives and Gender-Critical Scholars
Political discrimination is not simply an abstract sentiment, but is rooted in the emotions. One of the “non-material” benefits of collegiality, for instance, is a pleasant experience at the workplace, which contributes to well-being. Those who politically discriminate, however, are often uncomfortable socializing with those of other political backgrounds. This is especially true among progressives, likely because, as the More in Common report suggests, at high levels of education, they have far less social interaction with conservatives than less-educated progressives do. This lack of intergroup contact has been shown in other situations to permit misperceptions to flourish unchecked.102
One measure of social distance is whether an academic would feel comfortable sitting next to someone with a different view at lunch, be neutral towards the idea or not feel comfortable. Eating lunch together is a common form of collegial interaction among faculty and thus an important barometer of how non-material factors can alter the attractiveness of a workplace.
Figure 111 shows that just 41% of 2016 Democratic voters would feel comfortable sitting next to a Trump voter, 26% said they would be uncomfortable and 30% “neutral.” Trump and Clinton voters are similarly comfortable (61% vs. 65%) sitting next to a Sanders supporter. Less than half of Trump supporters (48%) and Clinton voters (25%) said they would feel comfortable sitting with a “known proponent of the idea that trans women should not be admitted into women’s refuge centers,” with 21% of Trump supporters and 50% of Clinton supporters saying they would not be.
Canadian data are similar, with 30% on the left uncomfortable with a Trump supporter and 36% comfortable. Fifty-four percent of Canadian academics said they would be uncomfortable having lunch with a gender-critical feminist colleague (i.e., who opposes trans women accessing women’s shelters). Here is evidence that gender-critical feminists may face even greater levels of discrimination than conservatives and Leavers.
The results for the British YouGov data are very similar. Figure 112 shows that under half of Remain voters would feel comfortable sitting next to a Leaver; 16% said they would be uncomfortable and 36% “neutral.” Leavers and Remainers are similarly comfortable (60% vs. 64%) sitting next to supporter of far-left politician Jeremy Corbyn. Less than half of Leavers (49%) and Remainers (34%) said they would feel comfortable sitting with a “known proponent of the idea that trans women should not be admitted into women’s refuge centers,” with 30% of Remainers and 12% of Leavers saying they would be uncomfortable sitting next to such a person. For the UK mailout sample, the corresponding Figures are 12% uncomfortable with a Leaver, 11% with Corbynite, and 30% with the gender-critical scholar. Among this group, 54% said they would comfortably lunch with a Leaver compared to 58% for a Corbynite. This indicates that intra-left animus is also important within academia.
Across all partisan identities, among North American PhD students, 39% would be uncomfortable sitting next to a Trump supporter, 9% uncomfortable with a Sanders supporter, and 59% uncomfortable with the gender-critical colleague. For UK PhDs the corresponding discomfort figures are 11% with the right populist (Leave) supporter, 18% for the far-left supporter, and 54% for the gender-critical feminist, suggesting no bias against right populists. However, in all cases, the share saying they would actively feel comfortable with the idea of sitting next to a populist left supporter was higher than the share comfortable sitting with a populist right supporter.
Overall, gender-critical researchers appear to face the highest levels of social discrimination. There is also a consistent transatlantic pattern of scholars being more comfortable dining with supporters of the populist left than those of the populist right. The anti-populist right skew seems somewhat more marked in North America than in Britain. Though there are differences, the positive news is that just 12-30% of leftist academics said they would feel distinctly uncomfortable having lunch with a populist right supporter, even if only a minority declared themselves “comfortable” doing so and many gave a lukewarm “neutral” response to the idea.
American and Canadian SSH left-wing and right-wing academics in Figure 113 differ by around 50 points in their comfort with the idea of lunching with a Trump supporter when I control for age, gender, and race. However, there is no statistically significant difference between SSH and STEM faculty, or between American and Canadian academics, in their comfort levels.103
Looking at the British YouGov results by left-right ideology and controlling for age, gender, professorial rank, and SSH vs. STEM in Figure 114 shows that those on the left are around 40 points less comfortable sitting next to a Leaver than academics on the right, a somewhat smaller gap than in North America. Nevertheless, there is a lot of variation in comfort level within the left, with just under half of leftists saying they would be comfortable sitting with a Leaver.
There is no obligation to associate with those one has little in common with. However, there is an important difference between freely associating with like-minded people and actively excluding those who are different. In a collegial context, active avoidance of a minority, as distinct from choosing to associate more often with those in one’s in-group, constitutes a form of political discrimination that can produce important chilling and conformity effects.
Figure 114 showed that ideology strongly conditions whether someone is comfortable sitting next to a Leaver. Does this matter for academic freedom? Yes. To understand why, consider what we noticed in Figures 111 and 112 concerning comfort at the idea of having lunch with someone with a different view. The model underlying Figure 115 asks what predicts discrimination against Trump/Leaver supporters or conservatives across seven questions I have encountered, compiled into one index with a score ranging from 0, indicating discrimination on no dimensions, to 1, discrimination on all 7 dimensions. Though ideology is a major cause of political discrimination, even when I hold this constant, the degree of comfort with a Trumper/Leaver remains extremely important.
I noted that those on the left split fairly evenly between half who are comfortable sitting next to a Trump/Leave supporter and half who are neutral or uncomfortable doing so. This variation is important in separating biased from unbiased leftist academics. Thus, even among faculty on the left (the red upper line in Figure 115), those who are most comfortable sitting next to their ideological opposites have less than a 50% chance of discriminating at least once against a Leaver, Friends of Israel supporter, or right-winger in a job, or rating a right-leaning article, grant, or promotion application lower. Leftists who are not comfortable sitting next to a Leaver have a greater than 90% likelihood of discriminating against them on at least one of the seven aforementioned dimensions. In effect, social tolerance is more important than ideology for predicting whether someone will politically discriminate. The academy is much better off with tolerant leftists than intolerant centrists.
The pattern is relatively similar in North America in Figure 116, albeit at a higher level of intolerance, with around 6 in 10 leftists who are comfortable sitting next to a Trump supporter discriminating against the right on at least one dimension, rising to 85% among those who are uncomfortable sitting with a Trump supporter. Ideology in the US and Canada matters almost as much as social comfort for predicting political discrimination against the right whereas in Britain ideology matters much less than social comfort.
Discrimination: Summary
Political discrimination exists across society, as several studies of the off-campus environment show. However, the specific problem in the social science and humanities departments of universities is that the balance of prejudices is highly lopsided due to the ideological makeup of staff, and professors’ beliefs are unusually visible in their work compared to other sectors of society.
The key, therefore, lies in awareness. If the fair-minded majority of academics are able to enforce a norm of non-discrimination on the discriminatory 30-40%, then structural discrimination can be reduced. This would need to encompass alertness to biased panellists’ tendency to tolerate a “dog-whistle” form of discrimination, in which political objections are couched in plausibly deniable language or excessive objections to other aspects of candidate performance or qualifications. The fact that the share who revealed they would discriminate in hiring is 2 or 3 times as large as the number who directly admitted to this in an unconcealed question suggests that committee chairs can highlight the need to avoid political discrimination to panel members and be on the lookout for concealed forms of political bias.
Toward Soft Authoritarianism
These results indicate that a large plurality of left-wing and some centrist academics discriminate against conservatives in aspects of academic life, from hiring to promotion to refereeing. This contributes to a culture of self-censorship, hampering conservatives’ freedom to disseminate the full range of their ideas in research and teaching, limiting the chance to bridge society’s main axis of political antagonism while constricting policy solutions. Political discrimination – acting as much through a chilling effect and the perpetuation of stereotypes of academia as a leftist preserve as through overt rejection – may also help explain why so few conservatives choose a career in academia. This distorts the production of knowledge and the effectiveness of the university. Notwithstanding this injustice, might a thick-skinned conservative manage to survive in academia? Not necessarily.
Political Discrimination and Illiberalism in Academia
Part I looked extensively at support for hard authoritarian measures such as dismissal campaigns or mandatory reading list quotas that restrict academic freedom. Is there any connection to Part II of this report on the soft authoritarianism caused by political discrimination? Absolutely. The kind of person that would discriminate against a conservative entering into, or succeeding in, academia is far more likely to be the kind of person that will endorse measures to push them out.
Conservative and gender-critical academics are keenly aware of political discrimination, and take steps to self-censor their research and teaching to avoid drawing unwelcome attention to themselves in a left-dominated environment. Here Mill’s “despotism of custom,” which chills expression, shades into the harder “cancel culture” politics of institutional sanctions or dismissal actions.
The results below show that softer and harder forms of authoritarianism are connected because “hard” restriction of academic freedom in the name of protecting disadvantaged groups is connected to “soft” political discrimination. This emerges clearly in the analysis in Figure 110 where I modeled the predictors of discrimination – with this outcome measured through an index of political discrimination aggregating the results of seven questions tapping a willingness to discriminate against Leave voters or those on the right.
Recall that in Part I, we saw that younger, far-left, and activist academics are more likely to endorse a campaign to fire a controversial scholar. When political discrimination is added to the model in Figure 117 predicting which academics will support one of four hypothetical campaigns to dismiss controversial scholars from their posts, I find that the willingness to discriminate against those on the right jumps out as by far the strongest correlate of illiberalism. This greatly improves the model’s ability to predict which academics will support a push to fire conservative or Leave-voting dissenters. Here we see clear evidence that soft authoritarianism (i.e., discriminating) serves as a gateway to hard authoritarianism (i.e., backing a firing campaign).
The pattern in Britain (YouGov data) is similar, with political discrimination the strongest predictor of being willing to endorse one of the four dismissal campaigns. This finding holds up across all five surveys, with political discrimination significant among both academics and PhD students.
In both North America and Britain, younger academics are more willing to endorse cancelling controversial staff, even when political discrimination and ideology are held constant. Looking ahead to the future of academic freedom, this may be viewed as a concerning development.
Being a political discriminator in the social sciences and humanities has a somewhat elevated effect on a person’s willingness to endorse the sacking of a fellow academic as compared to STEM fields. Discrimination is connected to hard authoritarianism in both parts of academia, but matters more for hard authoritarianism in the social sciences and humanities in the US and UK (but not Canada). Despite this relationship, it is vital to bear in mind that political discrimination, which contributes to a chilling effect that stifles academic freedom, is considerably more widespread than hard authoritarianism. As a result, some 70% or more of academics who would politically discriminate do not support dismissal campaigns.
Those on the far left who are willing to discriminate are also considerably more illiberal than far leftists who are unwilling to engage in political discrimination. Figure 119 shows that those on the far left who are not willing to discriminate against a right/Trump application, paper, or hire are around 25 points less likely to endorse a campaign to fire another academic. Among far leftists who would discriminate, 45% would endorse at least one authoritarian measure against conservative academic dissenters. The elevated relationship among far leftist discriminators holds independently in both the American and Canadian academic surveys.
Likewise, in Britain, far-left academics who are unwilling to discriminate against a right/Leave application, paper, or hire are around 20 points less likely to endorse a campaign to fire another academic. Figure 120 shows that among far leftists who would discriminate, nearly 3 in 10 would endorse authoritarian measures against conservative academic dissenters.104
Part III: Off-Campus
Is Academia Different?
Earlier we saw that while university professors and lecturers are more liberal than professionals in other sectors, the political gap is smaller if compared to degree-holders outside academia, especially advanced degree-holders. Perhaps, therefore, a chilling effect on free speech, political discrimination, and anti-conservative authoritarianism are parts of professional life and high culture in elite institutions more generally. As Andrew Sullivan writes, “We all live on campus now.”105 As we’ll see, there is a great deal of truth to this hypothesis. Yet it is equally apparent that universities stand out as distinctively hostile environments for conservatives.
Comparison with Non-Academics
As I will show, academics may not be more likely to politically discriminate than other professionals. Indeed, a well-known American study asking 1000 people to judge applications for a scholarship – where CVs that were otherwise identical variously stated the applicant had been president of a Young Democrats or Young Republicans club – found that respondents were biased toward their own party 80% of the time, regardless of merit, but cared little about race. As the authors note, “Political identity is fair game for hatred. Racial identity is not. Gender identity is not. You cannot express negative sentiments about social groups in this day and age. But political identities are not protected by these constraints. A Republican is someone who chooses to be Republican, so I can say whatever I want about them.”106
A major Cato Institute YouGov survey showed that 6 in 10 employees with master’s or doctoral degrees who support the Republicans say that they “are worried about losing [my] job or missing out on job opportunities if [my] political opinions became known.” Almost half of independents with postgraduate degrees agreed, compared with 25% for Democrats. Meanwhile, 88% of Trump-voting degree-holders compared to just 44% of Clinton-supporting degree-holders agreed that “The political climate these days prevents me from saying things I believe because others might find them offensive.” On the perpetrator side, 50% of “strong liberals” said a business executive who was found to be donating to the Trump campaign should be fired. Thirty-six percent of “strong conservatives” said the same for an individual donating to the Biden campaign.107
Moreover, it goes without saying that students also feel the chilling winds of Mill’s “despotism of custom.” In a previous Policy Exchange report utilizing a sample of over 500 British students, Tom Simpson and I found that only 4 in 10 Leave-supporting students said they would be willing to express their views in class. Students also split fairly evenly into a group that supported no-platforming controversial writers Jordan Peterson and Germaine Greer, a group that opposed these actions, and an intermediate, biddable group. Social media rather than lecturers were identified as the source of views on cancel culture and free speech questions.108
American research shows that 55% of students feel that the “campus climate prevents me saying things I believe.” Fully 82% of conservative students said they had self-censored at least once in class, compared to 40% of liberals. On politics, race, gender, and sexuality, about 30-35% of Republican students are reluctant to share their views in class compared to 15-25% for Democrat students. While these numbers show a substantial chilling effect, they indicate that right-leaning students are somewhat less inhibited in expressing their views than right-leaning academic staff.
Of students who feared speaking up, 61% worried about how their peers would judge them and nearly half were concerned about what their professor would think (not entirely an unfounded concern in the US context given that 7-15% of academics we asked revealed they would mark down a right-leaning term paper). Forty-three percent felt that their views might wind up on social media and nearly 40% worried that a complaint may be lodged against them.109 Here we see that both soft forms of authoritarianism based on peer pressure and hard forms involving fears of being disciplined work to produce a chilling effect.
Non-Academic Comparator Study
To compare my academic sample with professionals outside the sector on many of the same questions used in the academic survey, I used Prolific to sample 867 degree-holding employees over age 23 residing in Britain.110 The average age in the sample was 39, with a median household income of around £45,000, with two-thirds homeowners and around 72% married, living as married, or formerly married. Just 15 are university or college lecturers or professors, though some 97 work at universities, permitting us to compare them with other sectors. Survey platform samples tend to lean left and be tech-savvy, and ours is no exception. Eleven percent of respondents identified as right-wing, 46% as left-wing, and the remainder centrist or unstated. Twenty-one percent voted Leave, 73% Remain, with 6% not voting. This puts the sample of degree-holding non-academics about 5 points to the right of my academic sample, but well to the left of the degree-holding public.
Thus, when asked about the views of the median staff member at their workplace, the non-academic sample replied 36% left and 15% right, a 2:1 left:right slant compared to the 9:1 ratio in my UK YouGov academic survey. These non-academic professional workplaces are clearly more centrist than universities.
Limiting ourselves to the small subset of Leave voters in the non-academic survey, I find they report a 1.8:1 left-right balance in their work environment. This figure among non-academic Leavers compares to a 7.4:1 ratio reported by academic Leavers (in my academic survey) when describing their work colleagues’ median ideology. In the Prolific non-academic data, the subsample of 97 university employees (mostly staff rather than academics) report, on average, a 4.8:1 left-right ratio. This is clearly different from the 2:1 left-right ratio reported by those working outside the university sector. The latter seems to be an accurate representation of non-academic professional workplaces: YouGov Profiles’ 56,000-person representative sample shows a 2.1:1 left-right ratio among degree-holders working in non-university professional sectors, which matches that of my Prolific non-academic sample. Universities, then, lie considerably to the left of other professional sectors.
These figures reinforce the picture in Figures 49, 50, and 51, that find, using YouGov’s much larger sample of credentialed professionals, that British academics place significantly to the left of professionals in other sectors – even though university graduates in most major professions also lean left and Remain.
Non-Academic Workplaces Are Less Hostile to Expressive Freedom
Most conservative and Leave-voting professionals experience considerably less apprehension in their work environments than their counterparts in academia. Fifty-four percent of respondents felt a Leaver would feel comfortable expressing their views at the workplace, with only 16% saying such a person would not feel comfortable doing so. Among Leavers, the corresponding figures were 56% and 13% – no real difference. By contrast, we saw in Figure 90 that, among academics, just 37% of current academics of any political stripe think a Leaver would feel comfortable expressing their views, with just 28% of academic Leavers saying so. Among social science and humanities academics, only 18% of Leavers felt they would be comfortable doing so. This indicates that university-educated Leavers feel 2-3 times freer to express their views when working outside of academia.
This is reflected in the views of the 97 people working in the university sector in my non-academic dataset – researchers, administrative staff, and around 13 academics. Only 22% said a Leaver would feel comfortable expressing their views on campus, significantly lower than in any of the other 8 industry categories I recorded. In factories, there was no difference in comfort level between Leavers and Remainers when it came to expressing their opinions. In all other sectors, Figure 121 shows that 51-59% of respondents felt that Leavers would be comfortable sharing their views and that 76-90% of Remainers would as well. While there is a chilling effect for Leavers in all white-collar workspaces, it is two to three times larger in universities.
The share of conservatives and Leavers who feel they work in a hostile environment is also considerably higher inside academia. Seventeen percent of non-academic Leavers and 25% of non-academics who identify as right of center say their workplaces are at least somewhat hostile toward people with their political beliefs. Among current academic staff, the equivalent figures are 43 and 47%, rising to 50% of Leavers and 64% of conservatives in the social sciences and humanities. Here again, I find that academic environments are two to three times more hostile toward conservatives than professional workplaces outside the university.
Figure 122 shows that outside of universities, there is only a modest 8-point difference between Remainers (.08 probability) and Leavers (.16 probability) in the likelihood of a staff member saying their workplace is a hostile environment for their political beliefs. However, among the 97 university employees in the sample, the gap between Leavers and Remainers explodes in size to 54 points, with 4% of Remainers, but 58% of Leavers, reporting a hostile work environment.
