Do colleges have a free speech problem?

What a study of social norms in Saudi Arabia teaches us about free expression on campus

“If the doors of perception were cleansed every thing would appear to man as it is, infinite. For man has closed himself up, till he sees all things thro’ narrow chinks of his cavern.” – William Blake

A few weeks ago, the Foundation for Individual Rights and Expression (FIRE) released their annual College Free Speech rankings. The rankings are based on a survey of more than 55 thousand students across 248 U.S. college campuses. The survey asks students about their comfort expressing ideas, tolerance for liberal and conservative speakers, and other factors. These baseline rankings are then adjusted upward or downward based on data kept by FIRE of campus incidents like speaker disinvitations and sanctioning or supporting faculty and student groups for free speech-related issues.

Harvard ranked 248^th out of the 248 schools that were ranked. Not only did Harvard rank last, but FIRE singled us out for special scorn, stating that Harvard’s score was six standard deviations below the average, and more than two standard deviations below the next highest school.

My first instinct – and maybe yours – was to think they cooked the books to generate a headline-worthy result. So, naturally, I burrowed into the survey methodology like the good social scientist that I am.

My conclusion was that we have a serious problem with free expression on Harvard’s campus. It’s not quite six sigma bad, because the point values for speech sanctions are somewhat arbitrary and there are sample selection issues with the FIRE database (everything gets more attention when it happens at Harvard, and some of the cases they document weren’t speech issues at all).

Still, the student survey results are deeply discouraging. Harvard gets lower scores than most other schools on a variety of measures of free expression. The three questions where we score particularly badly are:

·         Comfort expressing ideas in the classroom, in dining halls, etc. (193 out of 248)

·         Acceptability of shouting down speakers or other disruptive blocking of campus speech (198^th)

·         Perceived likelihood that the administration clearly defends speech on campus (183^rd)

Even if we ignore specific incidents, the polling puts us near the bottom of the schools that were surveyed, and below most of our competitors. The FIRE survey results can’t be dismissed with methodological quibbles.

Yes, we do have a free speech problem at Harvard

The natural question is – why? One possible answer is that Harvard policies discourage speech more than policies at other universities. FIRE analyzed universities’ written policies about free speech and gave Harvard’s policies a “yellow” rating on a green/yellow/red light scale. People sometimes suggest that we should adopt the “Chicago Statement”, a set of principles about free expression on campus developed by the University of Chicago in 2014 that has since been adopted by more than 100 other universities around the country. I would like to see Harvard affirm these principles as well.

However, I don’t think university policy has much to do with our free speech problem at Harvard, nor do I think the Chicago principles would solve it. My reason is simple - most behavior that punishes or affirms free expression exists within the boundaries of university policy. Students and faculty are less worried about formal sanctions, and more worried that they will be shouted down or socially ostracized for expressing their views. Policy cannot easily fix what seems to me like a problem of social norms.

Another possibility is that we (faculty, staff and students) are ideological extremists that mostly agree with each other and are seeking to vigorously punish dissent. There is a bit of truth to this. After all, only 3 percent of Harvard faculty call themselves politically conservative. 91% of Harvard college seniors have an unfavorable view of Donald Trump.

I think we need more political diversity on campus. But I do not think that political extremism is the explanation for our free speech problem. If we really did have ideological uniformity on campus, people would feel pretty good about their ability to express themselves, because they’d know that everyone agrees with them! In fact, our students and faculty hold a wide range of views on controversial issues, even in cases where one side is clearly more popular.

For example, 32% of Harvard seniors view the Palestinian Boycott, Divestment, Sanctions (BDS) movement favorably, 18% view in unfavorably, and 50% have no opinion or need more information. 63% say they view race-conscious affirmative action favorably, compared to 15% who are unfavorable and 22% who have no opinion or need more information. In both cases one view is clearly more common than the other, but not overwhelmingly so, and many people sit on the fence.

We should be able to have conversations about these issues. Most faculty think so too – in a recent survey, 57 percent agreed that “Harvard should give controversial speakers a platform, even when many faculty or students object to their views”. To me, this suggests that there is a hunger on campus for more dialogue about difficult issues.

Cleansing the doors of perception

A third reason why free speech may feel threatened on campus is that we mistakenly believe our colleagues’ views are more unanimous than they actually are. I bet if you asked Harvard college students to guess how many of their classmates view race-conscious affirmative action favorably, they would guess a higher number than 63%. And if you asked them how many of their politically progressive classmates view it favorably, I bet they’d say “almost all of them”, when in fact many do not.

A recent review article by Leo Bursztyn and David Yang argues that misperceptions of others’ beliefs are pervasive, and that in-groups are especially miscalibrated. In other words, people think everyone in their group shares the same views, and that everyone in the outgroup believes the opposite.

Think for a moment about how this dynamic plays out in a classroom or in a faculty meeting. Suppose a politically progressive person offers a commonly held progressive view on an issue like Israel-Palestine, affirmative action, or some other topic. Fearing social sanction, people in the out-group remain silent. But so do in-group members who disagree with their group’s stance on that particular issue. They stay silent because they assume that they are the only ones in the group who disagree, and they do not want to be isolated from their group. The only people who speak up are those who agree with the original speaker, and so the perception of in-group unanimity gets reinforced.

This dynamic is highly discouraging for free expression. But there might be a way out. Informing people that others agree with them may make them more willing to share their views. There is safety in numbers.

Women working outside the home in Saudi Arabia

Evidence of the power of perceptions comes from an extremely clever paper by Bursztyn, Gonzalez, and Yanagizawa-Drott in the American Economic Review a few years ago. In 2017, only 4 percent of Saudi women worked outside the home and less than 15 percent worked at all. The Saudi custom, called male guardianship, requires a husband’s approval before any married woman can work. However, private views among Saudi men were running well ahead of social norms. In a survey administered by the authors, 87 percent of Saudi husbands agreed with the statement “In my opinion, women should be allowed to work outside the home.”

Although this view was widely held privately by the men who were surveyed, they greatly misperceived the views of their peers. In that same survey, which was administered to small groups of 30 Saudi husbands living near each other, participants guessed that only 63 percent would agree with the statement. This 24-percentage point “wedge” (87 minus 63) is a measure of people’s misperceptions.

The authors were interested in whether correcting misperceptions would change behavior. So they randomly assigned half of the men to a treatment group where they were informed that 87 percent of their session-mates agreed with the statement.¹ The control group was told nothing about others’ beliefs, so we can assume that their belief was 63% rather than 87%.

All participants were then given some information about a Saudi company that connects job-seeking Saudi women with employers, and they were given the option to sign their wives up for the service.

Saudi husbands in the treatment group were much more likely to sign their wives up. What’s more, correcting their misperceptions led to other behavior changes that persisted for months after the experiment. The percentage of wives applying for a job outside the home increased from 5.8 percent to 16.2 percent, a 180% increase. Job interviews increased from 1.1% to 5.8%, and actual employment 3-5 months later increased by 27%.

Even more encouragingly, the change in beliefs expanded beyond employment. Husbands were more likely to say that they would sign their wives up for driving lessons, suggesting a broader impact on women’s rights. Finally, treated respondents were asked 3-5 months later to guess at the beliefs of others in their neighborhood, and they were much more likely to believe that others thought women should be allowed to work outside the home. The change in beliefs persisted beyond the study, rather than reverting to baseline.

If misperceptions matter, how can we correct them?

This paper tests the classic theory in social psychology of “pluralistic ignorance” – where most people hold an opinion privately but believe incorrectly that other people believe the opposite. Evidence of pluralistic ignorance turns up on all kinds of controversial issues – racial segregation, climate change, alcohol use on campus…you name it. We’ve been here before. It’s social science meets “The Emperor’s New Clothes”, and I think it explains a lot about the speech climate on college campuses. Fear of social isolation silences dissenting views within the in-group, which reinforces the belief that such views are not widely shared.

One practical implication is that we should establish a baseline of beliefs within the group before engaging in difficult conversations. The old-fashioned way to do that is for people to get to know each other privately first, in one-on-one settings. Private conversations tend to build trust, which gives people space to say things like “yes I’m progressive, but my views differ on topic X.”

A higher-tech solution is to make frequent use of classroom polling technology. If you want to talk about a controversial issue in class (or in a faculty meeting), it might help to first elicit views anonymously and share them with the group, so that people can get a sense of where they stand.

Correcting misperceptions won’t fix all our speech problems on campus. But if we want people with unpopular views to share them anyway, we can give them a boost by showing them the data. Personally, I would fear the consequences of offering an opinion much less if I knew that 20% of the room agreed with me. And if nobody agreed with me, I’d like to know that too!

1. The average “wedge” was 24 percentage points, but of course it varied across the 30-person groups that were assembled. The treatment informed groups of the actual percentage who agreed with the statement, which averages to 87% but wasn’t always exactly 87%. The authors were very careful to ask other questions about the labor market to disguise the intent of the study. They also went to great lengths to ensure that private beliefs weren’t shared with peers and that respondents knew their beliefs would remain private.