Abstract: Recent essays by Jeffrey Sachs and Matt Yglesias have asserted that widespread concerns about free speech on American college campuses are essentially a moral panic with no basis in fact. In this blog post, we show that such skepticism is not justified by the survey data the skeptics point to. When the analysis is properly focused–on iGen college students since 2015, rather than on Millennials at large over the last ten years–the GSS does in fact show a recent downturn in support for controversial speakers. We show the same trend in a much larger survey of college students from the Knight Foundation, which was released the week after the skeptics published their essays. [To see part 2 of this post, click here]
Click here for a downloadable and printable PDF of this post.
This is the first post of a multi-part series exploring the issues raised by the skeptics. See also:
Over the past two weeks, Jeffrey Sachs (a political scientist at Acadia U; not the economist at Columbia) has made the argument that There Is No Campus Free Speech Crisis, as he put it in a long twitter thread on March 9. Matt Yglesias then expanded on Sachs’ argument in a post titled Everything we think about the political correctness debate is wrong, and Sachs expanded his case in a Washington Post Monkey Cage essay with a similar title: The ‘campus free speech crisis’ is a myth. Here are the facts. Sachs and Yglesias both draw heavily on analyses of the speech questions in the General Social Survey, which were plotted and analyzed well by Justin Murphy on Feb. 16. In this blog post we will show that a reliance on older datasets and the failure to formulate the question properly have led Sachs and Yglesias to a premature conclusion. Something is changing on campus, but only in the last few years.
Sachs and Yglesias claim that the current wave of concern about speech on campus that began around 2014 (with media reports about safe spaces and trigger warnings), and that intensified in 2015 (after the Yale Halloween controversy, and the earlier publication of The Coddling of the American Mind, by Lukianoff & Haidt) is a classic moral panic. They believe it is merely a media frenzy in response to a few high profile incidents. In a typical moral panic, people on one side of the political spectrum get riled up because stories about outrageous incidents appeal to their desire to believe the worst about a group on the other side. Sachs and Yglesias claim that conservatives and conservative media have gleefully exploited a handful of campus stories to fuel hatred of left-leaning students, or “social justice warriors,” when in in fact nothing has changed on campus.
Given how frequent moral panics are, especially as political polarization and cross-party hatred increases, and as social media makes it easy to whip up a panic, it is vital to have skeptics. It is important for people with different biases and prior beliefs to dig into survey data that bears on the question. It is also crucial to formulate the question properly. What exactly is it that has changed, or not changed, on campus in recent years?
Here are the three major positions in the current debate, along with our proposal for how each should be operationalized.
A) Lost Generation Theorists:
The claim: A generation of young people has turned, en masse, against free speech. There has been a big shift in the population, and we should all be very alarmed that millions of students are turning into “SJWs” (social justice warriors).
Operationalization: The shift is so big that we should see a big decline in the average level of support for free speech, or tolerance of speech that can be perceived as controversial or even “hateful”, among young people in general (or perhaps among just college students in particular) compared to previous generations. The change will not be just a few percentage points; it should be so big that if we plot out how attitudes change across generations, we should see a sharp downturn.
B) New Dynamic Theorists:
The claim: A new “illiberal” dynamic has emerged on many college campuses. Some students feel empowered to shout down speakers or otherwise prevent them from being heard, call out microaggressors, and even defund newspapers that publish articles they find offensive. College administrators have generally failed to punish students for this sort of behavior. On campuses where this illiberal behavior is tolerated, many students perceive a climate in which they must be very careful about what they say. Students in such climates say that they are “walking on eggshells” and are less likely to speak up honestly about what they believe. This is not a claim about the average college student; it is a claim about a change in the dynamicson many college campuses. It is this change in dynamics that Haidt and Lukianoff wrote about in The Coddling of the American Mind. They began to see this change at some of America’s most progressive colleges, in 2014, and published the article in August of 2015. (Their main concern was that this dynamic represented not just a worrisome attitude toward freedom of expression, but that students were learning ways of thinking that were detrimental to their mental health, including a new idea that students would be harmed by coming in contact with ideas that made them uncomfortable.)
Operationalization: Data collected after 2015 on how comfortable college students feel speaking up in class, or on campus, should show a decrease compared to data collected before 2014. The focal point is not the views of the average young person about controversial speech in general; it is the perception of the average college student about how easy it is to speak up, dissent, or challenge dominant views on campus. This new illiberal dynamic has been most visible at four-year residential schools known for their progressive politics, and located in the most progressive parts of the country — New England and the West Coast. That is where most of the “anecdotes” come from (e.g., Yale, Brown, Middlebury, UC Berkeley, Reed, Evergreen State).
The claim: Nothing has changed, there is nothing to see here. As Sachs put it: “There is no campus free speech crisis, the kids are all right, those that say otherwise have lost all perspective, and the real crisis may be elsewhere.” As Yglesias put it: “everything we think about the PC debate is wrong,” and “the alarm… is grounded in a completely mistaken view of the big picture state of American society and public opinion, both on and off campus.”
Operationalization: There should be no changes, no trend away from support for free speech, in data collected from college students after 2015, compared to college students before 2014. Yglesias does us the favor of offering these three specific empirical statements as his summary of what he believes the data say: 1) “Overall public support for free speech is rising over time, not falling.” 2) “People on the political right are less supportive of free speech than people on the left.” 3) “College graduates are more supportive than non-graduates.”
In this post we’ll examine the debate between the Lost Generation Theorists, who predict a big change in the average level of support for free speech, and the Skeptics, who assert that in fact there has been no change (or else continued change in the direction opposite of that predicted by the lost generation theorists). In our next post we’ll examine the debate between the New Dynamic Theorists and the Skeptics. (Well focus on data about who is afraid to speak up on campus, and why, and whether that fear is increasing.)
Before we delve into the relevant datasets, we must note a crucial point: This debate is not about millennials in general over the past 10 years, it is about iGen college students after 2015. Many people seem to think that the millennial generation ends with birth years around 1998-2000, but several researchers studying generational trends have noticed elbows in the data—turning points—beginning with adolescents who were born around 1995. Jean Twenge calls the generation after the millennials “iGen,” for “internet generation.” She shows that they have higher rates of anxiety, depression, self-harm, and suicide. They have a lower appetite for risk taking, and higher desires for safety. This generation first arrived on college campuses around 2013. By 2017, the millennials were largely gone (at least from the elite residential schools where most students come straight from high school and graduate in four years). So the key period of change, if there is one, is from 2013-2017.
The key transition year, if there is one, should be 2015. Before then, few Americans, and (as far as we can tell) few college students had heard of safe spaces, trigger warnings, microaggressions, cultural appropriation, and other terms and practices that are sometimes used to limit what some people can say, do, or wear (hence these practices are seen by many as illiberal). Those terms may have been common before 2015 on some of the most progressive campuses, such as Oberlin and Brown, but they were not well known on the vast majority of campuses. We (Jon and Sean) did not hear about them at all until 2014. They became nationally known during the student protests of 2015.
In particular, the microaggression concept has spread very widely and is encouraged on some campuses, and even in some high schools. This is the idea that people can be held responsible and “called out” for seemingly small things that someone else took offense to, regardless of the intention of the speaker. It is impact that matters, not intent, so everyone must be extra careful about what they say, being mindful of how the most sensitive person might react. In response to the protests of 2015, hundreds of schools created “bias response teams” to which anyone can report acts of bias, including microaggressions, anonymously (a Cato survey in 2017 and briefly discussed later, found that 65% of currently enrolled college students support the creation of BRTs … please note this result is in the internal cross-tabs that are available upon request from the Cato Foundation). If there are new ideas or norms about speech spreading across college campuses, then the 2015-2016 academic year is the best candidate for the year when they spread nationally.
In our analyses below we will focus on datasets that include data from college students, collected in 2016 or later. Any dataset that ends in 2014 or earlier is simply not useful for examining whether or not things are changing on campus. We examine three datasets with relevant current data.
1) The GSS is not ideal for resolving this dispute, but even still, it shows evidence of a change for iGen.
With a large nationally representative sample going back to 1972, the General Social Survey (GSS) is generally the best dataset for tracking longitudinal changes in American attitudes. It is useful for showing how Millennials differ from previous generations. The Skeptics are correct to note that the Millennials are, if anything, more tolerant of offensive speakers. See Justin Murphy’s graphs. (But note that the GSS uses the Stouffer method, which overestimates tolerance, as we’ll show in our third post).
But there is a problem with using the GSS to evaluate the Skeptics’ skepticism: In any given year the GSS samples only a small number of people aged 18 to 22. Specifically, since 1972 the sample sizes in the GSS for 18 to 22 year olds range from a low of 83 in 1993, to a high of 167 in 2000. Even among the 18-22 year olds sampled, many were not enrolled in college. In other words, the GSS does not contain much data on students who are enrolled in college at the time of the survey. Thus, it is not a good dataset for investigating hypotheses about current college students. For iGen, the GSS has responses from just 99 members that fall within the age range of interest in 2016, and of those 99 only 32 were enrolled in college at the time of administration. Thirty-two is too few to plot (since not all subjects even answered all items), but if we plot the data for all 99 members of iGen, we do indeed see a change, on willingness to permit controversial speakers to speak.
The GSS asks about six such speakers*, using a question stem of: “There are always some people whose ideas are considered bad or dangerous by other people. For instance…”:
— Somebody who is against all churches and religion.
— A person who believes that Blacks are genetically inferior.
— A man who admits he is a Communist.
— A person who advocates doing away with elections and letting the military run the country.
— A man who admits that he is a homosexual.
— A Muslim clergyman who preaches hatred of the United States.
For each speaker, respondents are then asked:
— If such a person wanted to make a speech in your (city/town/community) against churches and religion, should he be allowed to speak, or not?
— If such a person wanted to make a speech in your community claiming that Blacks are inferior, should he be allowed to speak, or not?
— Suppose this admitted Communist wanted to make a speech in your community. Should he be allowed to speak, or not?
— If such a person wanted to make a speech in your community, should he be allowed to speak, or not? (Militarist)
— Suppose this admitted homosexual wanted to make a speech in your community. Should he be allowed to speak, or not?
— If such a person wanted to make a speech in your community preaching hatred of the United States, should he be allowed to speak, or not?
You can see the average ratings of subjects in the 2016 GSS dataset, divided by generation, for all six speakers.
Figure 1. GSS data shows a possible new trend: iGen is a little less “tolerant” of four of the potentially “offensive” speakers.
In his Monkey Cage essay, Sachs plots the trend lines for just two of the six speakers–militarist and homosexual–and these two do show rising tolerance across time. But do those two lines really show us anything about changing attitudes toward speech, or do they merely reveal the declining offensiveness of the speaker? Asking Americans if a “homosexual” should be allowed to give a speech may have been a meaningful question in 1972, but it is a strange thing to ask today, after a revolution in attitudes about sexuality. The gradually rising line for “militarist” is harder to interpret. The fact that tolerance for this person rises with each subsequent generation may be related to the finding that each subsequent generation is less convinced about the necessity of democracy. It may not be a sign of increasing tolerance of offensive speech in general.
If we look at the other four items, they are inconsistent with Sachs’ claim that “each generation of young people has been more tolerant than the last.” The dips are not large enough to justify the claim of a Lost Generation, but at the same time these dips contradict the Skeptics’ claim that there is nothing to see, nothing going on. (We’ll need to wait until we see the 2018 GSS data to know if these dips are an aberration due to the small sample size, or if they are the first glimpse of a trend for iGen).
2) The Knight Foundation Surveys of Current College Students Show the Same Trend
Given that the GSS does not contain large sample sizes of students currently enrolled in college, it is important to draw on other sources of data that do. Just a week after Sachs and Yglesias posted their skepticism, the Knight Foundation, in collaboration with Gallup, released a survey that assessed current college students’ (N = 3,104) attitudes towards first amendment issues (see here for a summary of the survey’s findings). Among current college students, support for protecting free speech is indeed very high in the abstract (i.e., when asked about free speech generally, without context), yet when a more specific context is given, support for restrictions on expression is high:
Figure 2: Knight Foundation 2017 survey shows that support for free speech in the abstract is high (left-hand panel), yet support for specific restrictions on expression is high too, among many groups (in right-hand panel).
Most students in all categories support policies restricting speech that is “intentionally offensive to certain groups.” Such speech is legal in the public square, but within a college community it seems reasonable to have policies discouraging or punishing slurs and verbal aggression. But then what should a college do about students who are not trying to be offensive, they are just expressing their political views, and someone takes offense? Favoring restrictions on such speech is a clear declaration of illiberalism. Thirty percent of the college students in Knight survey made that declaration.
That’s a very high number, but do we have reason to think it might be increasing over time? Yes. The Knight Foundation conducted a similar survey in 2016, and asked some of the same questions in both years. Here is one such question: “If you had to choose, do you think it is more important for colleges to create a positive learning environment for all students by prohibiting certain speech or expression that are offensive or biased against certain groups of people OR to create an open learning environment, where students are exposed to all types of speech and viewpoints, even if it means allowing speech that is offensive or biased against certain groups of people?”
Figure 3: Knight Foundation 2017 survey shows some movement away from prioritizing “open learning environment” on campus over the past year.
As you can see in figure 3, the preference for “open learning environment” over “positive environment” shrank from 56 points to 41 points in just 12 months. That change may not be large enough to justify the Lost Generation Theorists, but it contradicts the Skeptics assurances that there is nothing to see, or that “everything we’ve been told is wrong.” And, of course, we don’t have data on this item from before 2014. Perhaps the preference was much larger back then, and the change over three years has been large. We just don’t know.
3) The Cato/YouGov Survey Shows that iGen and Millennials are more supportive of restricting speech
Finally, the Cato Institute, in collaboration with YouGov, conducted a survey near the end of summer 2017 on free speech and tolerance. The sample was composed of 2,300 respondents, and included an oversample of 769 current college and graduate students. They included some questions that essentially gauge support for blasphemy laws, which have been used for thousands of years to protect holy people, groups, and objects. If a country passes laws mandating punishment for insulting or disrespecting a person, group, or object, it is a blasphemy law, and it is incompatible with the first amendment. What percentage of Americans would support blasphemy laws? The Cato/YouGov survey had a series of items that began with this stem: “Would you favor or oppose a law that would make it illegal to say offensive or insulting things in public about…” It then offered ten groups. Is there a general trend toward greater speech tolerance among today’s young people, as the Skeptics maintain?
As you can see in figure 4, the two youngest age groups are the most supportive of blasphemy laws. Averaging all ten speakers together, the 18-24 group (blue line, mostly iGen) supported blasphemy laws 44% of the time, followed by 43% for the 25-34 group (red line, all Millennials). The least supportive of such laws were the 55-64 group (Boomers, 33%) and the 65 and over group (older boomers and Silent Generation, 33%).
Figure 4: Cato 2017 Free Speech and Tolerance survey, conducted in collaboration with YouGov shows that the two youngest age groups are the most likely to say they would support a law that would make it illegal to publicly say offensive or insulting things about various groups.
The skeptics are right to demand data. If the last two years have taught us anything, it should be that America’s affective polarization is now so strong, and so easy to exploit by “moral entrepreneurs,” to say nothing of foreign intelligence services, that we all must be more careful about accepting claims without evidence. The skeptics are also right that the data don’t show a sudden giant change. We do not have a Lost Generation. Young people have not suddenly turned against free speech in such numbers that we see cliffs in the data.
But the skeptics are wrong in their most basic and provocative claims that “The ‘campus free speech crisis’ is a myth” or that “Everything we think about the political correctness debate is wrong.” Let us return to Yglesias’ three specific claims:
1) “Overall public support for free speech is rising over time, not falling.” That may have been true until recently, but on some measures iGen college students are more willing to restrict or ban speech, or disinvite speakers, in order to maximize other values, particularly inclusion.
2) “People on the political right are less supportive of free speech than people on the left.” That has long been true historically, but as we’ll show in our next post, it seems to be less true today, and may even have reversed.
3) “College graduates are more supportive than non-graduates.” That was surely true until recently. It is probably still true about recent college graduates, but it may no longer be true about students who are in college today, and who were shaped by the college environment since 2015. As we will show in our next post, the cultural dynamics around speech are shifting on many campuses. We cannot draw on data from Millennials to assess what is happening on campuses today.
Survey research is complicated. Responses are heavily influenced by small details of wording. No one survey is definitive. Some survey data that we have presented shows that iGen college students are more willing to restrict speech than were millennial college students. Some survey data will likely not show that trend. The change in speech norms is so recent (we believe) that the question is not yet settled. Data collected in 2018 will give us a much clearer picture as to whether norms on campus really changed between 2013-2017. We will update this post as new data come in.
In our next post, The Skeptics are Wrong Part 2, we examine the claims of the “New Dynamic Theorists.” This is the claim that the speech climate on many campuses has changed in the last few years such that more students feel that they are “walking on eggshells” and are afraid to speak up honestly. Campus dynamics can change rapidly even if there is no change in the views of the average college student.
But the data we have presented here are sufficient to challenge the charges made by the Skeptics. We think this conclusion is likely to withstand scrutiny and further waves of data: American college students today are, on average, more willing than students of recent generations to restrict speech and speakers on campus, in pursuit of other moral and political goals, particularly inclusion.
* = Sullivan, Piereson, and Marcus (1979) demonstrated that the items of the GSS Murphy analyzed generally overestimate political tolerance. When respondents are asked to come up with the group they most dislike and rate them, reported tolerance drops, sometimes precipitously. We explore this problem and its profound implications in our third post, coming soon.