Introduction to Freedom of Expression in an Age of Social Media, Misinformation, and Political Polarization

In January 2022, the Knight Foundation released a study of American attitudes toward freedom of expression, which builds on regular student surveys that the foundation has conducted since 2004. With three other scholars (i.e., Katherine Glenn Bass of Columbia University; Daron Shaw of the University of Texas at Austin; and David Wilson of the University of California, Berkeley), we served as advisors in the development of the national survey of 4,000 Americans that was fielded jointly by Knight and Ipsos. The Knight Foundation study paints a complicated portrait of American public opinion regarding speech and expression, with wide gaps between support for freedom of expression in the abstract and support for particular examples of expression.

The Knight Foundation study paints a complicated portrait of American public opinion regarding speech and expression.

As part of this study, we put out a call in early 2021 to scholars to submit proposals for experiments that consider how people perceive different dimensions of freedom of expression. The preregistered experiments that we selected, which were fielded by Ipsos along with the Knight survey, reach the heart of the way people translate abstract support into political attitudes and behaviors. This symposium presents the (often-unexpected) findings of these studies.

Ranging in topic from police suppression of protests to “flagging” inappropriate content on social media to self-censorship, the experiments measure how identities of speakers, content of speech, and identities of the audience all intersect to affect the way that people apply and understand freedom of expression.

THE SURVEY

Before turning to the articles in this symposium, we briefly discuss the survey itself. Fielded in the summer of 2021 by Ipsos (N = 4,366) ¹, the survey included an oversample of respondents who do not identify as white and asked questions on various topics related to freedom of expression. On some questions, participants were randomly assigned to different question-wording options in order to track how they react to concrete (and, at times, politically contentious) cases and handle tradeoffs of competing values. The data are publicly available. ²

Although the survey included many insights that are detailed in the full report, we briefly highlight five patterns. First, when asked whether freedom of speech is important to them, the majority of respondents—more than 90%—reported that it is “extremely” or “very” important. ³ Similarly, 87% reported that “free-speech rights” are either “extremely” or “very” important for democracy. ⁴

Second, the survey also suggested some uncertainty in the respondents’ understanding of the First Amendment. Although 58% correctly answered that the First Amendment only prevents restrictions on free speech by the government, 40% incorrectly believed that it also prevents restrictions on the private sector. ⁵ In a different question, 35% of respondents answered incorrectly that “Barring someone from social media is a violation of their First Amendment rights.” ⁶

Third, Black respondents felt significantly less protected by the First Amendment than any other racial group. When asked, “Does the First Amendment protect people like you?,” only 61% agreed with the statement—compared to more than 80% agreement among other measured racial groups. Moreover, when asked on a scale of 1 (very difficult) to 7 (very easy) how easy or difficult it is for people who look like them to “use their free-speech rights without consequence,” Black respondents gave a response of 3.3, on average—again, lower than other groups. White respondents rated their ability to use free-speech rights at 4.7—and gave similar ratings when asked how easy or difficult it is for other racial groups.

Fourth, the survey highlighted partisan differences, which were most significant on issues that have been most politicized by partisan elites (e.g., athletes kneeling during the national anthem). Notably, when asked about a case that went to the US Supreme Court but that was not salient in the media (i.e., “A high school student making insulting comments about their high school on social media while not on campus”), Democrats and Republicans evaluated the scenario similarly.

Fifth, the survey highlighted the conditionality of support for freedom of expression. Although almost all respondents believed that freedom of expression is important when they were asked about it in the abstract, levels of support changed after the survey introduced specifics (e.g., misinformation on social media, protests, and jokes in the workplace). These differences seem to hinge on the details: Who is speaking? What are they saying? The studies described in this symposium delve more deeply into how these questions of “who” and “what” intersect in the way that people consider freedom of expression.

THE SYMPOSIUM

On their own, the articles each tell a compelling story about attitudes toward freedom of expression. Combined, they form a nuanced portrait of the conditionality inherent in the way that people consider and apply freedom of expression as they form political opinions and make decisions.

Focusing on the “who” in speech, Jessica Feezell, Meredith Conroy, Barbara Gomez-Aguinaga, and John Wagner consider how the identity of the speaker affects support for freedom of expression. In “Who Gets Flagged?,” the authors gave respondents the opportunity to “flag” a Twitter post—that is, to let Twitter know that a post violates “community standards.” Feezell et al. showed participants a series of social media posts, randomizing the race and gender of the poster who made an inflammatory post. They considered two questions: (1) whether women and people of color are especially likely to have their posts flagged; and (2) whether sharing identities affects flagging (e.g., “Are people less likely to flag posters who are like them?”). The authors do not find any evidence that certain people are more likely to have their posts flagged, but they do find that certain respondents are generally more likely to flag inflammatory posts regardless of the poster’s identity.

Shifting from who speaks to the content of speech, Andrew Lewis and Eric McDaniel consider how people view freedom of expression in a religious context. In “Religious Freedom Backlash,” they relied on a vignette study in which a religious leader gives a speech to policy makers. The authors varied whether the speech focused on themes of religious egalitarianism, Christian nationalism, or social dominance. They hypothesized that these different themes would affect the extent to which people support religious expression and that these differences also would be partisan. Democrats, the authors argued, are more likely to be supportive of religious freedom when primed with the theme of egalitarianism; Republicans would be more supportive when primed with the themes of Christian nationalism or social dominance. The authors find considerable partisan differences but not those that they expected. Although the treatments had no effect on partisans, they produced a backlash among Independent respondents. When presented with any of the religious messages, Independents became less supportive of religious expression.

Tony Carey and Ángel Saavedra Cisneros bring together the “who” and the “what” of speech in “Policing Protest: An Examination of Support for Police Suppression of Protest.” They used photographs and text to vary several aspects of a protest, including the racial identity of those participating in the protest and the issue at the heart of the protest. Carey and Cisneros also considered another dimension: whether media covers the protest through a “freedom-of-speech” frame or a “social-order” frame. They then asked participants whether they agreed with the following statement: “Sometimes police need to use force to suppress protesters.” The authors find that regardless of treatment and across all conditions, African American respondents were the least likely to support police using force to suppress protesters. White respondents, in contrast, were unconditionally most likely to support use of force.

In their article, “Tolerance for the Free Speech of Outgroup Partisans,” Roberto Carlos, Geoffrey Sheagley, and Karlee Taylor also jointly consider “who” speaks and “what” they say using two question-wording experiments. In the first experiment, they asked respondents how much they agreed or disagreed with the statement, “I would never support restricting my or someone else’s freedom of speech.” They randomly assigned some respondents to a modification of the statement: “…even if it means [Democrats/Republicans] will be able to say hurtful things.” The authors find that support for speech is highest when there is no chance of anyone making hurtful statements.

In a second experiment, Carlos, Sheagley, and Taylor asked about speakers on a college campus, varying whether campus speakers were described as presenting “only [liberal/conservative] perspectives.” They find that the inclusion of specifics increases support for freedom of expression. In both experiments, support for restricting speech did not vary based on whether the out-party or the in-party was making hurtful statements. However, the authors find that Republicans were more supportive of the speaker across all treatments.

It is one thing to consider freedom of expression in the context of others’ speech, but how do people think about constraints on their own expression? In the final article of the symposium, “Freedom of Expression in Interpersonal Interactions,” Taylor Carlson and Jaime Settle used a conjoint experiment to consider self-censorship. The authors tracked the circumstances in which people do and do not feel comfortable offering their true opinions. Using a conjoint design, they varied factors such as the context of the conversation; the relationship to the conversation partner; the race, gender, and partisanship of the conversation partner; and the conversation partner’s political engagement. They then asked participants: “Imagine that you were having a conversation about politics in the scenario described above. How likely would you be to express your true opinions in that conversation?” Carlson and Settle find that people prefer to share opinions in-person rather than online, with someone familiar, and with someone who shares their identities. Especially prominent was the preference for co-partisans, which was evident in both parties. In other words, even if Democrats and Republicans differ in their general attitude toward free expression, they all felt that they could express themselves more freely among their own co-partisans.

These introductory reflections only scratch the surface of the many important findings in these five scholarly articles. It is with great pleasure that we share the research in this symposium with PS readers. We thank the Knight Foundation for supporting this effort. We also thank the PS editorial team for publishing this scholarship, which covers an important but underdeveloped topic in political science: freedom of expression.