Allies or Agitators? How Partisan Identity Shapes Public Opinion about Violent or Nonviolent Protests

ABSTRACT

In recent years, scholars have argued that protests that employ nonviolent tactics attract greater support and are therefore more likely to succeed than those that use violence. We argue that how protest tactics are perceived is not a purely objective determination, but can be influenced in part by observer characteristics – in particular, by partisan identity. We conducted a survey experiment on two independent samples through the MTurk platform, randomly assigning protester group identity and tactics. Results show that when controlling for assigned tactics, self-identified Republicans but not Democrats perceive higher levels of violence when a disliked group is protesting. The effect is strongest in regard to tactics that are nominally the least disruptive. The findings have implications for theories of nonviolent protest, the legitimacy of repression, and the prospects for marginal groups to influence policy in polarized societies.

KEYWORDS:

• Protest tactics
• partisanship
• public opinion
• nonviolent strategies
• survey experiment

Acknowledgments

We thank Blaine Robbins for comments on the survey design. We also Steven Pfaff, Kerice Doten-Snitker, and the anonymous reviewers for feedback on previous manuscripts. Finally, we thank all the participants in the pretests and the surveys.

Disclosure Statement

No potential conflict of interest was reported by the authors.

Data Availability Statement

The data described in this article are openly available in the Open Science Framework at https://doi.org/DOI:10.17605/OSF.IO/TPA6U.

Open Scholarship

This article has earned the Center for Open Science badge for Preregistered. The materials are openly accessible at https://doi.org/DOI:10.17605/OSF.IO/TPA6U.

Supplementary Material

Supplemental data for this article can be accessed on the publisher’s website at https://doi.org/10.1080/10584609.2020.1793848.

Notes

1. To ensure validity of the responses, we restricted eligibility to answer the survey to MTurk workers who lived in the United States and had at least 100 previous approvals and an approval rate of at least 95%. Each assignment was paid $1.10. The median survey completion time was 8.5 minutes.

2. According to the 2010 census, Carter was the 49^th most common surname in the US, 58% of whose bearers identified as white and 36% as black. See http://howmanyofme.com/search/and https://www.census.gov/topics/population/genealogy/data/2010_surnames.html

3. Typically, it is adolescents and younger men who are portrayed in the media as prone to violence (Muschert, 2007).

4. This is intended as a measure of agreement with the protest based solely on the “group” treatment. However, in practice, because it follows the vignette, readers may also take into account information from the tactical treatment. Nevertheless, since the assignment of tactical treatments is random, by averaging across tactical treatments we can approximate a direct measure of how much respondents relate to the group’s goals.

5. This question was used by McLeod and Detenber (1999, p. 12), who did not find a significant effect.

6. The bill did not pass.

7. 23 participants were dropped from the analysis because they failed both checks or completed the survey in under three minutes.

8. See https://www.census.gov/quickfacts/fact/table/US/PST045218

9. In addition to the visualizations, ANOVA tests also yield significant results.

10. For visual purposes, we collapsed the scale of 7 into a binary one – pro-Republican or pro-Democrat. The middle category is omitted in this analysis.

11. Although in the visuals there appears to be some effect on the tactic “rocks” as well, as shown in Table 1 the effect was not statistically significant.

12. The initial positive slope between Intensity 1–3 in category 2 (Somewhat disagree) occurs because respondents are more likely to answer category 1 (Disagree completely).

13. We removed 40 respondents who failed the manipulation checks or if the response time was within 2 minutes (median response time was 3.9 minutes). We also restricted respondents to be those who have not participated in the previous study to produce independent samples. Compared to Study 1, the sample had an almost identical composition in gender, race, age, and education. The sample in Study 2 had a slightly higher level of income (The mean level of income was 2.79 in Study 1, while it was 3.11 in Study 2).

14. Regressions with Protest group: Abortion * Prefer Republican as the focal covariate yielded small effect sizes. However, as shown in Table 2 this is due to the asymmetric effect and heterogeneity between violent and nonviolent strategies.

15. The table shows bivariate coefficients. Including all the controls in Study 1 yields almost identical results (see A5-A7 in online appendix).

16. It may also be the case that the effect is produced by Democrats, who respond favorably toward anti-racist policing rhetoric. We thank a reviewer for pointing this out. Further research could test this by including a different “pro-Democratic” group condition.

17. We thank the reviewers for providing this suggestion.

Additional information

Notes on contributors

Yuan Hsiao

Yuan Hsiao is a doctoral candidate in Sociology and a Master’s student in Statistics at the University of Washington. His major work explores the intersection of digital media, social networks, demography, and political behavior, and has been published or forthcoming in American Sociological Review, Social Forces, Social Networks, New Media and Society, among other outlets. He also serves as a statistician that applies quantitative methods in different fields, and the collaborations have appeared in outlets in multiple disciplines.

Scott Radnitz

Scott Radnitz is Associate Professor in the Henry M. Jackson School of International Studies and at the University of Washington. His research focuses on identity, protests, and informal politics. He is the author ofWeapons of the Wealthy: Predatory Regimes and Elite-Led Protests in Central Asia(Cornell University Press, 2010).