Yves here. As positions on political issues become even more polarized, this post examines one route for creating social tolerance for expressing dissident views.
By Leonardo Bursztyn, Professor, Department of Economics, University of Chicago; Georgy Egorov, Professor of Managerial Economics & Decision Sciences, Kellogg School of Management, Northwestern University; Ingar Haaland, Associate Professor, University of Bergen; Aakaash Rao, PhD student in Economics, Harvard University; and Chris Roth, Professor of Economics and Management, University of Cologne. Originally published at VoxEU
Dissent plays a vital role in driving social change, but can be limited when individuals fear social sanction for expressing opinions about controversial issues. This column explores the function of rationales in facilitating dissent on both sides of the US political spectrum. Using a simple theoretical framework, it shows that liberals are more willing to post a tweet opposing movements to defund the police – and indeed face fewer social sanctions – when their tweet implies they have read scientific evidence supporting their position. Analogous experiments with conservatives demonstrate that the same mechanisms facilitate anti-immigrant expression.
Dissent is a vital driver of social change (Adena et al. 2020, Cantoni et al. 2014, Enikolopov et al. 2019, Gagliarducci et al. 2018, Manacorda and Tesei 2016). Dissenters usually draw upon rationales — narratives that provide arguments supporting dissenting causes — when expressing their views. Rationales might spur dissent because they are persuasive – that is, they change people’s private opinions and thus their behaviour. But in many situations, dissent is limited not because people support the status quo, but rather because they fear the social sanctions associated with publicly expressing their opposition. In fact, 62% of Americans agree that “[t]he political climate these days prevents me from saying things I believe because others might find them offensive” (Ekins 2020).
For example, consider a Democrat who opposes the movement to defund the police. She may be reluctant to express this opinion around fellow Democrats, as opposing defunding the police could be interpreted as a signal of racial intolerance. Now, suppose that a new study is widely circulated suggesting that defunding the police would encourage violent crime. This new study might increase her willingness to publicly oppose police defunding even if the study does not change her private opinions, as long as she is able to attribute her views to the study. The availability of this rationale provides an explanation other than racial intolerance for her position. Rationales can provide social cover and facilitate the expression of dissent.
In a new paper (Bursztyn et al. 2022a), we present experiments exploring the power and potential limitations of rationales. Across the political spectrum, dissent is often expressed — and suppressed — on social media, where rationales from both mainstream and fringe sources proliferate (Fujiwara et al. 2020) and where people often face large social costs for expressing controversial opinions. We experimentally investigate the expression and interpretation of dissent on social media, focusing on two controversial domains: liberals’ opposition to defunding the police and support for deporting illegal immigrants.
Experimental Design and Results
In a first experiment, liberal respondents read a Washington Post article written by a Princeton University criminologist arguing that “[o]ne of the most robust, most uncomfortable findings in criminology is that putting more officers on the street leads to less violent crime” (Sharkey 2020). They then choose whether to join a campaign opposing the movement to defund the police and, if so, whether to authorise a tweet promoting the campaign. (Importantly, our experiment is designed such that no tweets are actually posted.) The experimental manipulation varies the availability of a social cover in the tweet while holding fixed other potential motives to post. In the Cover condition, respondents’ tweets suggest that they were shown the article before joining the campaign, while in the No Cover condition, participants’ tweets indicate that they were shown the rationale after joining the campaign. The implied timing in the Cover condition provides these respondents with a social cover – the (implicit) justification that they joined the campaign because they were persuaded by the article’s claims. The timing implied by the No Cover condition removes this social cover. Differences in ‘willingness to tweet’ therefore cannot be explained by the persuasiveness of the rationale, as all respondents in both groups read the article; nor can such differences be explained by respondents’ beliefs that the rationale will persuade their followers, as both versions of the tweet contain an identical description of the article.
Figure 1 shows that the availability of a social cover strongly affects public behaviour: respondents are 12 percentage points more likely to authorise the tweet in the Cover condition than in the No Cover condition. A placebo and a number of further experiments help rule out other potential explanations for the treatment effect.
Figure 1 Willingness to post anti-defunding tweet
We conduct a second experiment, again with liberal respondents, to examine how the availability of the rationale shifts an audience’s inferences about the motives underlying dissent and the resulting sanctions levied upon dissenters. Figure 2 shows that when interpreting a previous participant’s decision to publicly oppose defunding the police, respondents see participants in the Covercondition as less racially prejudiced than those in the No Cover condition. They are also less likely to deny the Cover participant a $1 bonus, indicating that rationales lower the social sanctions associated with dissent.
Figure 2 Fraction who believe partner donated to NAACP (top) and fraction who deny partner bonus (bottom)
We also study the effects of rationales among a different sample (conservatives), and in a different policy context (anti-immigrant policies). Here, supporting the immediate deportation of all illegal immigrants from Mexico is a stigmatised opinion that people may be reluctant to publicly express, but a similar rationale as studied previously – concerns about crime – may shift inference about motives and thus decrease social sanctions. As in our first experiment, in the Cover condition, respondents’ tweets indicate that they were exposed to the rationale – a clip of the Fox News Channel’s anchor Tucker Carlson arguing that illegal immigrants commit violent crimes at vastly higher rates than citizens – before joining the campaign, while in the No Cover condition, respondents’ tweets indicate that they were exposed to the rationale after joining the campaign. As shown in Figure 3, respondents are 17 percentage points more likely to post the tweet in the Covercondition than the No Cover condition. A further experiment shows that this rationale once again has strong effects on inference: as shown in Figure 4, respondents matched with a participant who chose to post the Cover tweet are 5 percentage points more likely to believe that this participant authorised the pro-immigrant donation and 7 percentage points less likely to deny their matched participant the bonus.
Figure 3 Willingness to post pro-deportation tweet
Figure 4 Fraction who believe partner donated to US Border Crisis Children’s Relief Fund (top) and fraction who deny partner bonus (bottom)
Our evidence showcases the importance of rationales in facilitating dissent on both sides of the political spectrum. Our theory and evidence highlight the mechanisms by which individuals and institutions can influence public behaviour by shaping the availability of rationales and perceptions of their credibility. Our findings have important implications for how the expression of dissent responds to the availability of new narratives that become widely known.
First, rationales are only effective when observers believe that they genuinely change the dissenter’s beliefs. An obscure or non-credible rationale likely fails to shift inference, and may even backfire, if it is informative of the dissenter’s underlying type. In the extreme, if only intolerant people tend to read a particular source, citing a novel rationale provided by this source will fail to generate social cover. Thus, the endorsement of rationales by prominent figures such as politicians or celebrities may generate particularly large ‘social amplifiers’. Such figures may not only be more credible and directly persuade more people, but they may also be able to generate common knowledge such that dissenters can claim they were exposed to the rationale without seeking it out directly from stigmatised sources. Anti-minority politicians, for example, may enable supporters to speak their mind more openly and spread propaganda through their social circle, an effect documented for Nazi propaganda in the Weimar Republic by Satyanath et al. (2017). In another paper (Bursztyn et al. 2022b), we apply this framework to understand scapegoating of minorities during crises. The strength of the ‘social amplifier’ channel depends not only on the number of individuals who hold stigmatised views, but also on the number of individuals who could not express these views prior to the rationale becoming widespread.
Conversely, groups seeking to suppress dissent have strong incentives to silence or marginalise potential sources of rationales because these tactics reduce the perceived probability that people will be exposed to rationales ‘by chance.’ Such tactics may include censoring certain figures or otherwise disallowing them a public platform (e.g. disinviting campus speakers), or branding particular media sources or speakers as fringe. This helps explain why censorship techniques such as China’s Great Firewall can be very effective in suppressing dissent, even when citizens can bypass them with relative ease (Chen and Yang 2019). If successful, these tactics can create and sustain a ‘political correctness’ culture in which, for better or worse, certain rationales are ineffective because citing the stigmatised source undermines social cover. By questioning the credibility of rationales or tying them to stigmatised positions, a vocal group can silence a majority.
See original post for references