What You Can't Say: Social Image, Communication and Political Correctness
2019-2020 SCAD Graduate Fellow Luca Braghieri shares insights from his SCAD-sponsored research.
A few months ago, a host of prominent writers and academics - including David Brooks, Noam Chomsky, Nicholas Christakis, Francis Fukuyama, Jonathan Haidt, Deirdre McCloskey, Steven Pinker, Bari Weiss, and Fareed Zakaria - signed an open letter published in Harper's Magazine titled “A Letter on Justice and Open Debate”. According to the authors, “the free exchange of information and ideas […] is daily becoming more constricted” due to a culture of “intolerance of opposing views, a vogue for public shaming and ostracism, and the tendency to dissolve complex policy issues in a blinding moral certainty." Broadly speaking, the letter argues that, in many quarters, people feel pressure to publicly espouse views on a set of sensitive socio-political topics that they may not privately hold, and worry that such misrepresentations may impoverish the quality of public discourse.
My research aims to evaluate the argument above in the context of college campuses, where the quality of public discourse and the diffusion of information, both among students and between students and instructors, is a matter of first-order importance.
In order gain empirical traction, I ran a survey experiment at the University of California Santa Barbara in which I asked students to report the extent to which they agreed with a set of politically sensitive statements. The topics involved immigration, cultural appropriation, reparations for slavery, the removal of confederate statues, preferred gender pronouns, trigger warnings, etc.
Participants in the survey experiment were assigned to one of two treatments: a Private Treatment or a Public Treatment. Participants in the Private Treatment were ensured that their answers to the survey would remain completely anonymous; participants in the Public Treatment were not given such assurance of anonymity and were in fact given hints suggesting that their answers might be shared with other UCSB students, together with their names. The treatment manipulation aimed to vary the extent to which subjects were concerned about their reputation when answering the subsequent set of questions.
Five main results emerge from the experiment: first, compared to participants in the Private Treatment, participants in the Public Treatment tend to skew their answers in the direction that is generally perceived to be more socially acceptable at UCSB. Since the socially acceptable positions at UCSB generally correspond to the ideological position of self-identified liberals, the answers of participants in the Public Treatment of the experiment tend to be skewed in the liberal direction compared to the answers of participants in the Private Treatment. Second, the effects are highly heterogeneous across self-reported political ideology. Specifically, there is little difference between the answers that participants who self-identify as liberal report in private and in public. Conversely, there is a large difference between the answers that participants who self-identify as moderates or conservative report in private and in public. Third, when asked in an open-ended question whether it is acceptable to disrupt the talks of controversial speakers invited to campus to deliver lectures, participants in the Public Treatment are more likely to argue that the disruptions are acceptable and less likely to discuss issues related to freedom of speech than participants in the Private Treatment. Fourth, the answers of participants in the Public Treatment are less informative than the answers of participants in the Private Treatment according to a host of theoretical measures of informativeness. An intuitive way of thinking about this finding is as follows: if an observer tried to use the participants’ responses to the survey to predict their demographic characteristics and behaviors, the observer would make more accurate predictions using the answers of participants in the Private Treatment than using the answers of participants in the Public Treatment. Fifth, information loss is exacerbated by the fact that the natural audience in the environment, namely other college students, are in part naive about the ways in which social image concerns distort their peers' public statements. Specifically, students seem to have a hard time understanding the demographic groups that, when assigned to the Public Treatment, tend to skew their answers in the socially acceptable direction. Students fail to identify political ideology as the key dimension that induces their peers to publicly state opinions that they do not privately hold; conversely, and contrary to the findings in the experiment, they think that the opinions that whites and males hold in private are very different from the ones they portray in public.
Overall, the results lend support to the argument I set out to evaluate: students seem to feel pressure to publicly espouse views that they do not privately hold, and such misrepresentations impoverish the quality of public discourse. The impoverishment is partly because the statements that students make in public are not as informative as the ones they make in private, and partly because misreporting makes it hard for other students to figure out whose public statements to trust.