The Mental Health Crisis Does Not Explain Woke

My first report for the Centre for Heterodox Social Science

Welcome to the first report of my new Centre. This report represents the deeper research behind arguments set out in my piece in Unherd, and includes, as part of the report, a theoretical and empirical reply to Greg Lukianoff’s response article in the same publication.

The Mental Health Crisis Does Not Explain Woke [click to download PDF]

Executive Summary

·         The youth mental health crisis is real, with a majority of young people saying they are sad or anxious at least half the time. Among American students, nearly 7 in 10 say they are anxious at least half the time compared to a third of American adults

·         Young people are considerably more woke than older people. Over 7 in 10 American college students support banning a speaker who claims transgenderism is a mental illness

·         Being younger is the strongest predictor of mental illness in the population and, next to liberal-conservative ideology, the strongest correlate of woke attitudes

·         Yet feeling anxious or depressed is far less important than being liberal or conservative in predicting whether someone supports woke ideas like cancel culture or saying their country is racist. Anxious liberals don’t differ much from well-adjusted liberals in their support for such ideas

·         While liberals are more depressed and anxious than conservatives, mental health and liberal-conservative ideology have largely independent effects on attitudes to speech and Critical Social Justice

·         The youth mental health crisis therefore does not explain why young people are more woke than their elders. It cannot account for the Great Awokening - even though both trends arose in the mid-2010s

·         Female students are actually more likely than non-binary students to support no-platforming speakers who oppose transgenderism

·         Gay, lesbian, bisexual and queer students are considerably more likely than other students to support shouting down, blocking or using violence against offensive speakers

·         Within the young population, it is women, LGBTQ people, those on the left of the political spectrum and low achievers that are more likely to report poor mental health. Socioeconomic status and achievement matter, but cultural categories are more closely connected to variation in mental health

·         Lonely students are considerably more likely than those who are not lonely to say they do not have enough time for themselves

·         Non-binary or individuals who question their gender have worse mental health than cis-gendered men and women, but the effect is smaller than for sexual orientation

·         Whites report worse mental health than nonwhites in some surveys. Race is not one of the more important predictors of attitudes toward cancel culture or critical race and gender perspectives.

·         So too for education. And while religion matters for mental health it has no consistent effect on woke attitudes. Rising diversity, education and non-religiosity therefore do little to explain the Great Awokening

·         It appears that the spread of woke ideas due to the rise of smartphones, social media and a more partisan news media contributed to a rise in youth mental illness, rather than the other way round

Introduction

Woke refers to the sacralization of historically marginalized race, gender and sexual minorities. From this set of symbolic attachments we can derive an ethical principle I term cultural socialism: the desire for equality of result and emotional harm protection for totemic identity groups.

Woke, as measured by the number of people targeted in firing campaigns, the scale of campus deplatformings or mentions of Critical Social Justice terms such as ‘racism’ or ‘transphobia’ online took off in the early-to-mid 2010s.[1] At the same time, rates of depression and anxiety doubled or tripled among young people.[2]

This study will make the claim that today’s mental health crisis, especially among youth, does not explain woke. Rather, this belief system evolved from left and liberal sources alongside new patient-affirming humanitarian-therapeutic ideas. It subsequently extended its reach off campus in the 2010s with the rise of social media and partisan news. My claim is that this owes little to society’s state of mental health.

Data Review: The Youth Mental Health Crisis and the Great Awokening

The 2010s is the key period for both the rise of progressive illiberalism and a surge of mental health problems, concentrated among young people, especially young girls. Figure 1, drawn from the Foundation for Individual Rights and Expression (FIRE), College Fix and National Association of Scholars’ databases, shows a rapid spike in deplatformings in the 2010s, largely initiated by left-wing activists. This subsides somewhat after 2020-21 but remains elevated.

Figure 1

Source: College Fix Campus Cancel Culture Database, accessed December 1, 2022; FIRE Scholars Under Fire database, accessed December 4, 2022; Acevedo, D. ‘Tracking cancel culture in higher education’, National Association of Scholars (NAS), accessed December 4, 2022.

Meanwhile, media mentions of social justice terms took off in the mid-2010s, as shown in Figure 2, which focuses on coverage in the New York Times.

Figure 2

Source: https://media-analytics.op-bit.nz/timeline, accessed August 20, 2019.

Figure 3 shows that these subsided somewhat after 2020-21, in line with the trend in deplatformings in Figure 1.

Figure 3

Source: Rozado, David. ‘Is The Great Awokening Really Winding Down? Part II: Evidence from News Media,’ David Rozado Substack, June 19, 2023.

We also know that young people increasingly began to suffer from heightened rates of depression, anxiety and other mental health disorders, as shown for the United States in Figure 4. This trend also took place in other western societies.

Figure 4

Cited in Jonathan Haidt, After Babel Substack, Feb 1, 2023.

Jean Twenge provides a nice summary of several trends in Figure 5, showing that teen depression appears correlated with social media, internet and smartphone use rather than any economic trends.

Figure 5

Technology adoption, teen depression, and national unemployment, 2006-2019. Sources: National Survey of Drug Use and Health, Monitoring the Future, Pew Research Center, Bureau of Labor Statistics. From Twenge, J., ‘Here are 13 other explanations for the adolescent mental health crisis. None of them work,’ Generation Tech, October 18, 2023.

Figure 6 shows the growth of a number of major social media platforms. Tumblr, Instagram and Snapchat all took off during the period in which teen mental illness soared. Tumblr and Instagram in particular have been named as sites in which young influencers began to spread identity-politics related content more widely in the 2010s. ‘Tumblr was the first place many white people…encountered ideas about race and privilege…[the pages of teenagers] included posts about feminism, anti-racism and social justice,’ enthused a progressive Pacific Standard writer in 2018.[3]

Figure 6

The rise of woke, mental illness and social media occurred synchronously, but what caused what? How are these developments connected?

Turning to studies which compare individuals at one point in time rather than aggregate trends over time shows that young people on the political left report elevated mental health problems compared to young people on the right.[4] Other psychologists have likewise noticed that Jonathan Haidt and Zach Goldberg, in studying the post-2010s teen mental health crisis, find that young liberals are two or three times more likely to report mental health problems as young conservatives. The pattern by age and sex for 2020 is shown in Figure 7.[5]

Figure 7

Source: Goldberg,Z. in Haidt, J., ‘Why the Mental Health of Liberal Girls Sank First and Fastest,’ Free Press, March 13, 2023. Sample limited to whites.

In addition, left-wing cultural beliefs appear to be correlated with poor mental health in a way support for redistributive economics is not. George Yancey argues that this is because cultural leftism revolves around empathy toward certain groups and enmity toward their oppressors, leading to an emotive, dichotomous view of the world - which is not conducive to good mental health.[6] Recently, a large-scale academic study from Finland discovered that woke individuals are more likely to be unhappy, anxious and depressed than other people.

The Puzzle of Causation

Do woke ideas make people unhappy, or are unhappy people drawn to the cultural left? By extension, did the rise of woke cause the mental health crisis, or did the rise of mental health problems produce what Matthew Yglesias and others term the ‘Great Awokening’? This study seeks to ascertain the direction of the causal arrow between the mental health crisis and Great Awokening.[7]

A psychological perspective takes the latter stance, holding that the mental health of society has declined, giving rise to a woke culture of fragility and victimhood. A sociology of emotions outlook, by contrast, argues the former view: that ideologies (i.e. mind viruses) are the driver, switching emotions on or off to align with the ideology.[8] In the sociological account, woke ideology comes first, prompting people to view themselves and others as victims with mental health problems. This jacks up mental illness rates.

Two recent books advance the psychological argument that mental health problems are an important force behind woke. Abigail Shrier’s Bad Therapy: Why The Kids Aren’t Growing Up identifies permissive parenting and therapeuticization as central to the youth mental health crisis, which in turn is fueling political extremism. Focusing on one’s thoughts and feelings instead of the world out there leads to more psychosis. Jonathan Haidt’s The Anxious Generation: How the Great Rewiring of Childhood Is Causing an Epidemic of Mental Illness points to smartphone use, hyperparenting and social media as culprits behind the mental health epidemic. The result, as explored by Haidt and Greg Lukianoff in the Coddling of the American Mind, is an explosion of left-wing illiberalism on campus.[9]

The aforementioned authors are open to both psychological and sociological explanations, but they focus mainly on the psychological. Clarifying his views in response to an article I wrote bearing the same title as this report, Greg Lukianoff countered that: ‘Our book argues that teaching others the habits of anxious people results in both a mental health crisis and illiberalism on campus. They work and emerge together: one doesn’t cause the other.’ The causal argument seems to be that bad therapeutic ideas spawned both the mental health crisis and progressive illiberalism. Yet the intellectual history of this therapeutic thinking remains unmapped.[10]

My claim is more forthright: that woke ideas are primary in the causal chain, with a deep intellectual history that reaches back to the humanitarian, egalitarian and expressive individualist ideas which first coalesced as Liberal Progressivism and bohemian left-modernism in the early twentieth century. Left-liberalism lurched to the radical cultural left in the late 1960s, establishing a set of ideas which spread first in universities and hotbeds of left-wing municipal organizing such as Berkeley and San Francisco in the US, the Greater London Council in Britain or Toronto District School Board in Canada.[11]

Using new data from FIRE on the mental health and free speech views of undergraduate students, the data questions the psychological explanation for woke. How so? Many liberals do not have psychological problems and some conservatives do, thus it becomes possible to ask whether mental health problems or ideology underpin woke beliefs in the here and now.

The report will begin by outlining evidence on the relationship between mental health and ideology, especially among young people. I then move to examine the three-way association between mental health, ideology and support for cancel culture to see what this can tell us about their association. In the final section, I examine these relationships in the wider adult population, and also ask how mental health and ideology interact in predicting attitudes along the critical race, feminism and queer theory dimensions.

Part I Youth Mental Health

I begin with the mental health of American young people, asking which factors are most closely associated with better or worse mental health outcomes.

I draw on the Foundation for Individual Rights and Expression (FIRE)’s 2023 survey of 55,000 American students, mainly the 15-20 percent who are enrolled at the top 200 research-intensive universities. I examine how ideology and anxiety interact to predict whether a student will be in favour of the illiberal practice of shouting down a visiting speaker.

FIRE’s massive sample of 18 to 22 year-olds reveals  the magnitude of the youth mental health crisis. Five new questions on FIRE’s 2023 student survey (released in 2024 as the College Free Speech Rankings) address student mental health. Namely:

1.      How often would you say that you feel anxious?

2.      How often would you say that you feel lonely or isolated?

3.      How often would you say that you feel like you have no time for yourself?

4.      How often would you say that you feel depressed?

5.      How often would you say that you feel stressed, frustrated, or overwhelmed?

Response categories for these items were:

1) Never

2) Less than half the time

3) About half the time

4) Most of the time, nearly every day

5) Always

In my analyses, for ease of interpretation, I dichotomize these measures so that categories 1 to 3 (fewer problems) are distinguished from the 4s and 5s (severe mental illness). However, it must be noted that those who say they have mental health issues half the time are arguably also experiencing poor mental health.

As Figure 8 shows, 68 percent of students say they are anxious and 42 percent that they are depressed at least ‘half the time.’ 76 percent report being ‘stressed, frustrated and overwhelmed’ at least half the time. A whopping 37 percent say they are anxious ‘almost every day’ or ‘always’. This is indeed a crisis.

Figure 8

Source: FIRE 2024 College Free Speech Rankings data. N = 55,058.

The five mental health indicators are highly correlated, and load onto one latent variable that captures 60 percent of the variation across the five indicators. There are a number of surprising relationships, most notably, as Figure 9 shows, that lonely people are much more likely to say they have no time for themselves! This suggests that an underlying mental illness factor may be driving negative responses across all five questions.

Figure 9

Source: FIRE 2024 College Free Speech Rankings data. N = 55,102. Bars show raw counts of students. Graph excludes median categories of those who are lonely or have no time for themselves ‘half the time.’

In order to assess what may be driving student mental health, I first amalgamate all five mental health questions into one latent variable based on factor analysis (rotated principal components) which, as noted, captures 60 percent of response variation across the five variables. I could have done this in the form of an index, which would yield a similar result. I next run a statistical analysis in Figure 10 to identify the leading predictors of student mental health. The outcome measure is a combined mental health score across the five questions, with coefficients expressed as standardized effects. The length of each bar shows the size of the effect for each predictor variable, whether in a positive or negative direction.

Results show that being female is the most important correlate of poor mental health in this model (.19). This is followed by sexual orientation, with heterosexuals having significantly better mental health than non-heterosexuals (.13). Those from a higher self-ascribed social class background (on a 5-point scale from highest to lowest) have better mental health than those from lower class backgrounds (.12).

Beyond this, those reporting lower grades have poorer mental health than those on higher grades. Students who identify as non-binary or other gender also have worse health than the cis-gendered. Both effects are a moderate .07 in strength. Those who say they are liberal have worse mental health than moderates or conservatives, in line with past research reported earlier (.06). The ‘very liberal’ left-most point and the two most conservative points on the 7-point ideology scale stand out most from the mid-point in terms of, respectively, poor and good mental health.

Significant but less important effects include religious ‘nones’ having worse mental health than the religiously affiliated, younger students having better health than older students. Other groups with significantly lower mental health include white students and those receiving financial aid.

These results suggest that socioeconomic status and grades matter a lot for mental wellbeing, but are somewhat less important than cultural categories such as gender, sexuality and ideology (not to mention religion and race). Interestingly, LGBQ sexual orientation appears to be more predictive of reduced mental health than non-binary gender. On this note, there has been a pronounced two- to three-fold increase in young people identifying as not heterosexual since 2010. This trend has largely taken place within the most left-liberal segment of youth public opinion.[12]

Figure 10

Source: FIRE 2024 College Free Speech Rankings data. N=52,007. R2=.109.

Another way of conceptualizing how these variables affect mental health is to look at the concrete case of anxiety. Figure 11 shows how the share reporting daily anxiety falls as I change the signature profile demographic. For example, 71 percent of the 120 students who are female, very liberal, not heterosexual, of lower socioeconomic class, with Grade-Point Averages below 3 (the 30^th percentile) and no religious affiliation report chronic anxiety.

As we move right across the chart, I alter one characteristic at a time to see how anxiety levels decline, using capital letters to show which characteristics I changed. The share with high anxiety falls 11 points as I move from lower to higher achieving students. It drops a further 7 points as I substitute higher socioeconomic class students for lower class ones. It drops a further 6 points if Christian rather than nonreligious students come into focus. Anxiety falls another 5 points to 42 percent when moving from non-heterosexual to heterosexual students, and drops 9 points to 33 percent as I change the profile from ‘very liberal’ to non-liberal (i.e. moderate and conservative) female students. The final factor is gender: when I swap male for female students with these characteristics, the subgroup drops 16 points in anxiety, showing the importance of gender for anxiety. Note that this progression is for illustration only and should not be read as equivalent to the systematic evaluation of the predictive power of various characteristics presented in Figure 10.

Figure 11

Source: FIRE 2024 College Free Speech Rankings data. The number of cases per profile is given in brackets.

Moving away from a focus on relatively elite university students reveals similar results. Figure 12 summarizes findings from models predicting how often an 18-20 year-old says they feel sad or anxious, with a 5-point response category that is identical to that used by FIRE. The data is based on two surveys fielded in 2022. In Britain, data was collected by YouGov for 1,552 18-20 year olds, with 500 per each single age. In the US, the sample of 1,502 is from Deltapoll, collected between August 18 and 23.

In both samples, household income or personal income, as well as social class (in UK) has no association with mental health. University students do not differ in their levels from those not attending. However, in Britain, questions on current employment status show that young people not in education or employment are substantially more likely to have elevated sadness or anxiety compared to other young people. Those who have chosen not to attend university but have the marks to do so have better mental health than others, whether university students or non-students. Thus achieved status rather than personal income or class seems to be important for mental health. The FIRE data confirms that grades are important, but also finds social class and financial aid status to be connected to mental health.

US data shows that 3 percent of young respondents identified as something other than male or female. This did not predict poorer mental health. But the 25-30 percent who identified as not heterosexual reported substantially worse depression and/or anxiety, as in the FIRE data.

Results for both British and American young people shown in Figure 12 therefore highlight that being female or non-heterosexual is powerfully associated with being more sad and anxious than males or heterosexuals. In both this and the FIRE data, the effect of being LGBQ does not seem to vary much between LGBQ males or LGBQ females.

Ideology is also important, with a standardized effect size on mental illness of between .08 (UK) and .13 (US). Those on the far liberal or left end of the spectrum did not stand out, with the data showing more of a gradual decline in mental wellbeing as we move from right- to left-wing subjects. All told, these results for young people, only a minority of whom are students, reinforce findings in the FIRE student data - allowing us to generalize beyond the approximately 15-20 percent of college-age Americans attending leading universities.

As in the FIRE data, while material/achievement and cultural factors both matter, the latter are somewhat more consequential for predicting youth mental health. This intimates that cultural shifts may be more important than material ones in accounting for the mental health crisis.

Figure 12

Source: YouGov UK Apr 14-May 4, 2022; Deltapoll US data August 18-23, 2022. N=1,542 (UK) with R2=.106; and 1,195 (US) with R2=.107.

Part II Mental Health and Cancel Culture

I now turn from analyzing the correlates of student mental health to assessing the connection between mental health and cancel culture, first among students, then among adults more generally.

In the FIRE data, 19 percent of the students are conservative and 48 percent liberal, in line with previous FIRE survey waves going back to 2020. A related split, 20 percent Republican and 53 percent Democratic, holds for two-party identification.

In terms of cancel culture attitudes, an astonishing 32 percent of students support shouting down a speaker to prevent them being heard, 18 percent endorse physically blocking a speaker and 11 percent back the use of violence to prevent them addressing an audience on campus. 71 percent feel a speaker who claims BLM is a hate group should ‘definitely’ or ‘probably’ not be allowed to speak on campus while only 12 percent say such a speaker should ‘definitely’ be permitted to speak. 71 percent feel the same way about a speaker who says transgenderism is a mental disorder. 57 percent would bar a speaker who endorses an abortion ban, with only 16 percent ‘definitely’ favour allowing the talk to proceed. The problem of  illiberal progressive student attitudes on US campuses remains acute.

Student attitudes reflect those of young people in general. This has been shown in numerous surveys, and is apparent in Figure 13 which illustrates that young people in the US and Britain are about twice as likely (67-68%) as those over 45 (25-39%) to endorse Google’s firing of programmer James Damore for questioning the firm’s gender equality assumptions in a scientifically-rigorous internal memo.[13]

Figure 13

Source; Kaufmann, E. Taboo: How Making Race Sacred Produced a Cultural Revolution / The Third Awokening (Forum Press/Bombardier Press, forthcoming June 20/May 14, 2024).

Are mentally ill students more likely to support cancel culture? This could explain why rates of mental illness and cancel culture rose in tandem in the 2010s. On the other hand, if mental illness is not strongly connected to cancel culture attitudes, this would suggest that ideology is substantially independent, and that cultural left ideas may have spawned both woke attitudes and mental illness. This is not to claim that a correlational study can prove causation. An instrumental variables approach or, preferably, experimental evidence, would be needed to make more definitive causal claims.

Controlling for gender, year, socioeconomic class, sexual orientation, race, depression, stress and grade-point average allows me to focus squarely on the importance of being anxious. The red line in Figure 14 represents the roughly two-thirds of anxious students and the blue line the one-third who are not anxious. The horizontal axis plots students on a 7-point scale from those who identify as ‘very liberal’ to those who say they are ‘very conservative.’

The net result: almost all of the variation in willingness to shout down a speaker is explained by ideology, not anxiety. Nearly 1 in 2 very liberal students support shoutdowns, and even 1 in 5 strong conservatives do, but there is little difference between the anxious and non-anxious. Mentally ill students, whether liberal or conservative, are somewhat more censorious than their well-adjusted peers, but this doesn’t make much difference to the overall level of cancel culture. Results are the same when I swap depression for anxiety, or use different outcome measures of cancel culture.

Figure 14

Source: FIRE 2024 College Free Speech Rankings data. N=52,019. Pseudo-R2=.062.

I find a similar effect in Figure 15 concerning the question of whether a speaker who says Black Lives Matter is a hate group should be permitted to speak on campus. There is a staggering 45-48 point difference between the most liberal and conservative points of each line but just 4-6 points between the lines, i.e. between students with anxiety and those without.

Figure 15

Source: FIRE 2024 College Free Speech Rankings data. N=52,097. Pseudo-R2=.135.

Figure 16 indicates that the impact of gender, like that of ideology in Figure 14, dwarfs the impact of mental health on cancel culture beliefs. When asked whether a speaker who claims transgenderism is a mental health condition should be permitted to speak on campus, the distance between the top line for females and bottom line for males is a whopping 25 points. By contrast, the difference between the left and right points of each line is almost zero. The overwhelming majority of students back an anti-free speech position on this question. Interestingly, women (red line) are more opposed to a politically-incorrect speaker on the trans issue than are non-binary (including agender, genderqueer, genderfluid or unsure) people, represented by the green line!

The stronger female predilection to shut down contentious speech may spring from evolved caring dispositions, as Cory Clark and Bo Winegard have argued: ‘Women are more likely to experience self-protective emotions such as anxiety and fear, to be more harm- and risk-averse, and to have more empathy and desire to protect the vulnerable. Men, on the other hand, are more likely to take risks and to endorse hierarchy and support for conflict.’[14]

Another perspective is that women are more likely to endorse established communal norms, whether conservative or progressive. Thus female freshmen were more conservative than male freshmen in the early 1970s, during a period of social unrest in which established conservative norms were being challenged. Only after 2004 were women consistently more liberal.[15] Women are also consistently more religious than men.

Finally, it may be that the dominant ideology evolved to embody a feminized outlook, which therefore resonates better with women. For instance, the ‘feminine’ abolition of rude table manners in the early modern period, or dueling in the early 1800s, likely garnered more female than male support.[16]

Figure 16

Source: FIRE 2024 College Free Speech Rankings data. N=52,097. Pseudo-R2=.147.

Women are far more likely than men, and more likely than the non-binary, to prevent controversial speakers from coming to campus to speak on questions such as transgenderism, abortion or Black Lives Matter. They are more likely also to want to keep speakers away who insult conservatives on guns, religion or race, though the strength of this effect is not quite as large.

However, not all cancel culture items are equally correlated. When it comes to keeping conservative speakers off campus, there is a strong correlation between survey items asking about controversial speakers on trans, BLM and abortion bans. However, when it comes to shouting down, blocking or using violence to prevent speech, support for these tactics taps a somewhat different attitudinal dimension of cancel culture. On this set of questions, women are no more (and sometimes less) supportive of cancellation than men, whereas lesbian, gay, bisexual and queer students are, on these items, considerably more illiberal than heterosexuals.

Figure 17 illustrates. The blue line for LGBQ students is considerably more supportive of shouting down offensive speakers than the red line representing heterosexual students. This relationship holds across the political spectrum. In fact conservative students who are LGBQ are some 15 points more likely to favour shoutdowns than their straight conservative counterparts – a larger gap than the 10-point difference between heterosexual and non-heterosexual liberals. Note that this model controls for confounding factors including anxiety and gender.

Figure 17

Source: FIRE 2024 College Free Speech Rankings data. N=52,097. Pseudo-R2=.060.

One way of taking stock of these relationships is to examine the matrix of correlations between four sets of questions:

1.      Mental health indicators

2.      Cancel culture related to preventing controversial conservative speakers from coming on campus

3.      Cancel culture related to shouting down, blocking or using violence against offensive speakers

4.      Political ideology

What the matrix in Figure 18 highlights is that mental health indicators (in blue) cluster tightly, with high correlations of .32 to .65. While the strongest associations involve intuitive pairings such as anxiety-stress and loneliness-depression, we also find, as noted, a substantial counterintuitive connection between a person feeling lonely and feeling that ‘you have no time for yourself’.

Cancel culture questions pertaining to blocking, shouting down and using violence to prevent offensive speech (in pink) are highly associated, at between .41 and .53.

A largely separate dimension of cancel culture attitudes consists of questions to do with keeping controversial conservative speakers off campus (shaded green) – in this case on the transgender and BLM issues. These correlate so heavily with liberal-conservative ideology that all three might even be considered part of an underlying latent attitude.

A set of more modest correlations, in the .10 to .25 range, shown in italicized bold, indicate overlap in the Venn diagram between these three clusters of attitudes. For instance, liberal-conservative ideology predicts support for shoutdowns at a fairly strong .25, and support for blocking speakers at a more modest .12, but is unrelated to support for violence to prevent speech. Ideology is correlated with anxiety, depression and stress at a modest .11 to .15, with conservatives experiencing better mental health than liberals.

The shoutdown and block cancellation cluster has a modest association with the no-platforming of anti-BLM and anti-trans cluster, with correlations in the moderate range .11 to .19. Cancel culture is not unidimensional.

And when it comes to mental health and cancel culture, the only moderately strong association is between anxiety and support for the no-platforming of offensive anti-BLM and anti-trans speakers (.14 in both cases). Links to the more aggressive tactics (shoutdown, block, violence) are weak.

As noted, there are two distinct dimensions to cancel culture in the FIRE data. Liberal-conservative ideology is more strongly associated with views on no-platforming than with attitudes to obnoxious tactics like shoutdowns or blocking. Ideology correlates most strongly with the ‘softest’ aggressive tactic, shoutdowns, somewhat less for blocking and not at all for violence. That is, when controlling for confounding factors, conservatives are as likely as liberals to advocate violence to suppress speech - bearing in mind that almost no students do so.

Women are significantly stronger backers of no-platforming than men. But when it comes to endorsing the use of violence, shoutdowns or blocking to suppress speech, women are less likely than men to approve. Instead, it is LGBTQ individuals who index most highly on the more strident tactics.

Figure 18

Source: FIRE 2024 College Free Speech Rankings data. N = 55,058.

Thus far we have examined questions that are controversial among the left-liberal majority of students. However, FIRE’s surveys also ask about whether speakers who offend conservatives should be permitted to address an audience on campus. Should a speaker be permitted who says guns should be confiscated, white privilege upholds systemic racism, or religious liberty is an excuse to treat gay people badly?

Comparing these questions with those that offend progressives (BLM is a hate group, trans is a mental disorder, all abortion should be illegal) in Figure 19 reveals that anxiety has a slight ideological valence. Specifically, those with high anxiety are 8-10 points more likely than those with low anxiety to want to ban speakers offensive to the left, but 0-4 points less likely than those with low anxiety to advocate deplatforming speakers who offend the right.

Figure 19

Source: FIRE 2024 College Free Speech Rankings data. N = 55,058.

Of course, this could be because – as we saw - there is a correlation between being left-wing and being anxious, hence the anxious will back leftist speakers’ right to offend conservatives. The anxiety question may also be picking up aspects of ideology or measurement error that are not fully captured by the liberal-conservative question.

In order to probe this, I ran statistical models controlling for socio-demographics such as gender, sexual orientation and race. These examine how the interaction between anxiety and ideology changes as we move from strong liberals to strong conservatives in predicting whether a student will favour banning a progressive speaker from campus.

For speakers attacking religious liberty, white privilege and gun confiscation, the results show essentially no impact of anxiety on support for no-platforming the speaker. Only when it comes to speakers attacking trans, BLM or abortion do I see an effect. This suggests that even the modest impact of anxiety on cancel culture may be ideologically-inflected. For instance, it could be that some students are anxious about guns, religion and white people and thus view speakers who attack them as anxiety-reducing. Anxious liberals are more likely than other liberals to favour deplatforming conservatives, but this may be picking up measurement error (noise) around ideology rather than reflecting the impact of psychological states on support for cancel culture.

Patterns in the Adult Population

In this section, I venture beyond the young population to examine the wider adult population, drawing on a survey of 1,317 American adults I commissioned with Qualtrics between July 21 and October 3, 2023. Among adults, 17 percent said they were sad or anxious most of the time or ‘always’, with fully two-thirds reporting low depression and anxiety. Those under 35 had worse mental health, with 27 percent sad or anxious most of the time and 48 percent saying they were never or only sometimes sad or anxious. A similar Canadian survey of 1500 conducted by Maru on September 18-20, 2023 (for another study of mine) using the same mental health question showed a somewhat analogous picture, with 13 percent reporting high sadness and anxiety and 72 percent low levels. For those under 35, the levels were at an elevated 20 percent and 55 percent, respectively. In the Canadian sample, however, there were very few under-25s – with those present in the data reporting somewhat greater levels of high sadness/anxiety, at 26 percent, than the 20 percent for the 18-35s as a whole.

LGBQ sexual orientation was the strongest predictor of mental problems (standardized beta of .20), followed by age (i.e. being younger) at .16, female (.14) and poor (.09), in that order. Neither race, income or education was significantly correlated with mental health while liberal-conservative ideology was also not a significant predictor in this data. The Canadian picture is somewhat similar: being young (.28), female (.13), left-wing (.12) and poor (.10) were the leading predictors of sadness and anxiety. Neither education nor LGBTQ was significant, however. Race was not available in the survey and social media use, while signed in a direction that would indicate a relationship, did not emerge as statistically significant.

Rather than examining cancel culture - an attack on classical liberalism - the next survey probes adherence to critical race, feminist and queer theory, manifested as charges of racism, sexism or homophobia against particular groups which members of these groups may consider demeaning. This is represents an attack on conservatism, i.e. of attachments to nation or ethnic group.

Respondents were asked whether they agreed or disagreed with a set of statements on a 7-point scale from ‘strongly disagree’ to ‘strongly agree’:

·         Italy is racist

·         Chinese-Americans are racist

·         African-Americans are homophobic

·         White Democrats are racist

·         Hispanic-Americans are sexist

·         White Americans are racist

·         United States are racist

·         White Republicans are racist

Importantly, in the case of Chinese, African-Americans and Hispanics, around a third of respondents are unsure whether these groups are guilty as charged. For Italy, the share who say ‘neither’ is over 50 percent. There is clearly greater uncertainty in the responses to questions about these more ‘exotic’ entities whereas only about 20 percent have no clear view on whites, the United States or white Republicans and Democrats.

The results by partisanship are presented in Figure 20, sorted by partisan gap. Democrats, as expected, are much more likely to view whites, the United States and white Republicans as racist. But they are also as or more likely to view the ‘exotic’ groups as racist, sexist or homophobic. Why? Left-wing critical race, feminist or queer theory deems not just whites, but males or heterosexuals, to be ‘privileged’, and therefore to be upholding oppressive structures of racism, patriarchy or heteronormativity even if they do not tick all the intersectional boxes. Having said this, non-white groups tend to fall outside the usual progressive purview, so would not be expected to attract the same level of criticism as whites or the United States.

As a result, I find that Republicans are only slightly more likely (32%) than Democrats (27%) to agree that white Democrats are racist, whereas one might have expected Republicans to respond in a highly partisan manner to this question. In addition, there is virtually no difference between Democrats and Republicans in their tendency to agree that Chinese-Americans are racist or African-Americans homophobic. Democrats are more likely than Republicans to say that Hispanic-Americans are sexist and Italy is racist, but the differences are modest: a gap of 9 and 11 points, respectively.

However, when it comes to the United States, white Americans or white Republicans, a much greater share of respondents say these are racist, and there is now a pronounced partisan skew: for example, 51-66 percent of Democrats call these entities racist, while just 17-33 percent of Republicans do. Party identification is clearly a powerful predictor of perceptions of whether whites, white Republicans and the United States are racist.

Figure 20

Qualtrics survey, July 21-October 3, 2023. N = 1,317 American adults. Other response categories include disagreement or neutrality.

Having noted the partisan pattern in Figure 20, I next move to consider the predictive power of mental health status. Figure 21 presents results by depression-anxiety level, parsing out those with severe depression or anxiety from the 83 percent majority who do not suffer from these problems. The chart is sorted by the gap in outlook between the mentally ill and mentally healthy on critical race, feminist and queer questions.

The first point of note is that those with poor mental health are more likely to agree with critical theory claims across all the groups listed. Moreover, the size of the gap is relatively consistent. While the raw gap ranges from 7 points for African-Americans to 13-18 points for whites, white Republicans and the United States, when we consider these numbers in relation to the gross level of such sentiment (i.e. 22 vs 29 percent), the percentage change is similar for all categories, a very different picture from Figure 20 where Democrats indexed far higher on whites and Americans. In view of the greater uncertainty in responses when people assessed the moral failings of the ‘exotic’ nonwhite or foreign entities, the fact that people who are mentally ill rate all entities as more racist/sexist/homophobic, even those that are minorities or foreign nations, jumps out as noteworthy.

Notice that this story, in Figure 21, presents a different picture from the pattern we saw among students in Figures 14 through 16, where mental health made only a modest difference for views on cancel culture and only predicted progressive illiberalism when the proposed speaker was conservative.

Figure 21

Qualtrics survey, July 21-October 3, 2023. N = 1,317 American adults. Other response categories include disagreement or neutrality.

The foregoing is borne out in statistical analysis. First of all, in OLS regression models using the full 7-point response categories for assessment of a particular entity, a respondent’s mental health has roughly half the power (.13) of ideology (.25) to predict whether an entity is perceived as racist, sexist or homophobic. Thus both mental health and ideology are important in a way that is much less true in the student data.

Second, mental health is basically independent of ideology when it comes to predicting a person’s response to claims about the moral failings of particular groups. The model in Figure 23 shows that mental health status has a broadly constant effect across all ideological categories. The same pattern holds when modeling responses to the other groups analyzed here: that is, it does not change when considering ‘exotic’ or dominant groups, or those deemed friendlier to the Democrats or Republicans.

As with Figures 14 and 15 in the FIRE student data, mental health status appears to have an independent effect from ideology, with no significant interactions across the political spectrum. Bad mental health sours a person on the group in question, irrespective of which one it is, and regardless of whether the person is conservative or liberal. As in the FIRE data, introducing mental health variables into the model does not make much difference to the predictive power of ideology. This indicates that the effect of a person’s ideology on their critical race/feminism/queer theory views (i.e. accusing a group of being racist) is not an artifact of the correlation between being a liberal and having worse mental health.

As in the FIRE student data, partisanship matters much more than mental health for predicting how a person will evaluate claims of racism against whites, Americans, white Republicans or white Democrats. However, for the minority groups or Italy, mental health status is a more important predictor than ideology of whether a person will call these entities racist, sexist or homophobic.

Figure 22

Qualtrics survey, July 21-October 3, 2023. N = 1,314 American adults. Other response categories include disagreement or neutrality. Pseuo-R2=.085. Interactions not significant.

Interestingly, minority groups did not differ much from whites when their group was being criticized (black, Hispanic, Chinese). Hispanics, in fact, were somewhat more likely than other groups to agree that Hispanics are sexist. On the other hand, the questions about whites and Americans all elicited significantly greater opposition from whites as compared to minorities, who were more likely to agree with the statements.

Discussion: Making Sense of the Relationship Between Mental Illness and Illiberalism

My view is that a sociology of emotions approach best fits the data, and that we are in the midst of a feedback loop, detailed in Figure 23. Liberal-individualist modernism promotes boundary-transgression while cultural socialism lionizes victimhood. In other words, left-modernist western culture creates a therapeutic sensibility based on victimhood and transgression. Social media and partisan news, entering in the 2010s, accelerated the transmission of these ideas off campus, producing both woke activism and declining mental health.

Popularized left-modernism (i.e. woke) inclines psychologically susceptible individuals to lean towards insecure modes of thought, removing externally-defined structures and social roles. It shapes the therapeutic victimhood ideology which feeds back into left-modernism, energizing it. Increasingly loose social boundaries are accompanied by more rigid political and ideological tribes. However, the latter do not provide the same existential security and social ordering that well-defined social bonds of gender, ethnicity, family or nationhood do.

People’s inclination toward resilience is actively suppressed in the culture. One result is rising mental distress, as indicated in the bottom rectangle in Figure 23. Social media is a force multiplier for cultural socialism mainly because it accelerates the flow of left-modernist ideas from elites to young people, often via influencers. Higher social media exposure has a modest negative impact on mental health - plotted as the thin-dotted line running down the right of the chart - but does not explain the 2010s woke youthquake.

Instead, the pathway is more cultural. The arrows at the top left of Figure 23 indicate that higher education matters for wokeness but less so for LGBT identity or mental health. Anxiety and depression are as prevalent among those who don’t attend university as among those who do (though highly-educated young women have had the biggest recent takeoff in poor mental health). White and highly-educated young people are bathed in a more intense and politicized victimhood culture than most minorities (apart from those at elite schools).[17] Thus it is highly-educated young white women and LGBT identifiers who are in the vanguard of this new sensibility.

Woke, manifesting as approval of cancel culture and critical race/gender theory, is centred in the young highly-educated leftist population. Young graduates provide the advance guard for cultural socialist beliefs and activism, with women and LGBT identifiers overrepresented. Highly-educated minorities are more culturally socialist than other minorities and can be prominent campus activists, but, in numerical terms, they are not the driving force.

Social media, as shown in the top right of the chart, oxygenates cultural socialism by enabling flash mobs and online approval for attacks on hate figures while facilitating the circulation of outrage from progressive media sites. Along with the new clickbait model of journalism, social media allowed longstanding critical race, gender and queer theory ideas to cross the academia/media boundary in the 2010s and spread into the wider society.

The correlation between mental health problems and woke beliefs arises mainly because left-modernism’s victimhood and transgression culture, fanned by social media, encourages an experimental approach to social roles while nourishing a sense of personal victimhood. This increases depression and anxiety. This is not to say that the Haidt-Lukianoff-Twenge mechanism of social media directly impacting mental health to create a fragile new generation has no purchase. While social media’s role as a social accelerant is more important than its impact on individual mental health for the Awokening, the dotted lines on the right side of Figure 23 show that social media also directly affects mental health.

The victimhood narratives of the cultural left resonate more with those in search of new values and sources of meaning, or who identify as weak. This means young people, women, far leftists and sexual minorities, and, to a lesser extent, those experiencing poor mental health. Nonwhites, who might be expected to identify with these ideas, are insulated on two counts. First, because they are less invested in partisan and ideological identities, and second because they often come from more traditional cultures with stronger identity boundaries. Only among young minorities educated at elite universities is there widespread support for progressive illiberalism.

Figure 23

In short, the woke youthquake is, in the main, not a result of the youth mental health crisis. The depressing takeaway is that cancel culture and negative attitudes toward free speech, objective truth, majority groups in society, or the nation as a whole, are likely to flourish even if we discover a magic pill to cure mental health.

Eric Kaufmann is Professor of Politics at the University of Buckingham, Director of the Centre for Heterodox Social Science and author of the upcoming Taboo: Why Making Race Sacred Led to a Cultural Revolution (Forum Press UK, June 6)/The Third Awokening: A 12-Point Plan for Rolling Back Progressive Extremism (Bombardier Books USA, May 14).

www.heterodoxcentre.com

---

[1] For the trend in cancellations, see FIRE’s Campus Deplatforming Database and Scholars Under Fire Database. For the big data trend in mentions of Social Justice terms online, see Rozado, D. (2022). "Themes in Academic Literature: Prejudice and Social Justice." Academic Questions 35(2).

[2] Twenge, J. M. (2017). iGen: Why today's super-connected kids are growing up less rebellious, more tolerant, less happy--and completely unprepared for adulthood--and what that means for the rest of us, Simon and Schuster.

[3] Nagle, A. (2017). Kill all normies: Online culture wars from 4chan and Tumblr to Trump and the alt-right, John Hunt Publishing; Mounk, Y. (2023). The identity trap: A story of ideas and power in our time, Random House, pp. 84-88.

[4] McGreal, S. (2021). “The Unexpected Relationship Between Ideology and Anxiety.” Psychology Today, September 20.

[5] Goldberg,Z. in Haidt, J., ‘Why the Mental Health of Liberal Girls Sank First and Fastest,’ Free Press, March 13, 2023.

[6] Lahtinen, O. (2022). "Construction and validation of a scale for assessing critical social justice attitudes." Scandinavian Journal of Psychology; Yancey, G. (2023). “Identity Politics, Political Ideology, and Well‐being: Is Identity Politics Good for Our Well‐being?,” Sociological Forum, Wiley Online Library.

[7] Fischer, Molly. 2018. “The Great Awokening: What Happens to Culture in an Era of Identity Politics?”, The Cut, Jan. 8; Yglesias, Matthew. 2019. “The Great Awokening.” Vox, Apr 1.

[8] Turner, J. H. and J. E. Stets (2005). The sociology of emotions, Cambridge University Press.

[9] Shrier, A. (2024). Bad Therapy: Why the Kids Aren't Growing Up, Swift Press; Haidt, J. (2024). The Anxious Generation: How the Great Rewiring of Childhood Is Causing an Epidemic of Mental Illness, Random House.; Lukianoff, G. and J. Haidt (2018). The coddling of the American mind : how good intentions and bad ideas are setting up a generation for failure. New York City, Penguin Press

[10] Lukianoff, G. ‘What’s behind the campus mental health crisis?,’ Unherd, March 21, 2024; Kaufmann, E. ‘The mental health crisis does not explain wokeness,’ Unherd, March 18.

[11] Pew, James, ‘Fulcrum and Pivot: The New Left Remaking of Toronto School Policy,’ Woke Watch Canada Substack, March 10, 2024.

[12] Kaufmann, E. (2022). "Born this way? The rise of LGBT as a social and political identity." Center for the Study of Partisanship and Ideology, Report(6).

[13] Wakabayashi, Daisuke, ‘Google Fires Engineer Who Wrote Memo Questioning Women in Tech,’ New York Times, August 7, 2017.

[14] Clark, C. and B. Winegard, ‘Sex and the Academy,’ Quillette, October 8, 2022.

[15] Stolzenberg, E. B., et al. (2020). "The American freshman: National norms fall 2017." Higher Education Research Institute, UCLA.

[16] Elias, N. (1994). The Civilizing Process. Oxford, Blackwell.

[17] Kaufmann, E., ‘Diverse and Divided: A Political Demography of American Elite Students,’ CSPI, October 3, 2022