Article Summary
Introduction
While previous studies have demonstrated that partisan affective polarization can lead to behavioral consequences in political and nonpolitical contexts, the extent of partisan prejudice remains unclear. Previous research has shown that affective polarization has increased and that partisans dislike the other side more than in previous decades, but it’s unclear how far citizens are willing to take this dislike and how influenced they are by party elites. In their article titled “The Limits of Partisanship,” Lelkes and Westwood address this gap by examining the boundaries of partisan prejudice in America. Specifically, they explore how severe partisan prejudice has become and whether partisan affect drives Americans to disregard democratic norms. It’s particularly important to understand if partisan animosity is driven more by like of one’s own party vs dislike of the other party, as the answer to this question will influence how willing people are to actually punish someone from the other party. The authors investigate this question of how far partisans are willing to take their feelings toward the in and out party because it is important to understand if Americans are willing to actually violate democratic norms or if they just support rhetoric from their party elites.
To answer these questions, the authors systematically investigate the limits of partisan prejudice through a series of five studies that utilize a typology of prejudice. Their findings suggest that while affective polarization is linked to certain types of prejudicial behavior, it primarily involves in-group favoritism rather than out-group hostility. The authors note that intentional harm to the opposition is limited, even among those most affectively polarized.
Analytical Approach
Based on Allport’s framework, this paper defines three levels of prejudice: the promotion of negative speech, the avoidance of group members, and tolerance for behaviors that violate democratic norms. This paper uses five studies to examine the role of partisanship and affective polarization in these three levels of prejudice towards in-group and out-group partisans, respectively. Affective polarization, used as a moderating variable in the analysis, is measured pre-treatment by the difference in the feeling thermometer score between a respondent’s own party and the opposition party.
The following tables summarize the design of the five studies and their experimental conditions.
Study 1: affective polarization and the promotion of hostile rhetoric (N = 556)
| Treatment conditions (randomized) | Read a news opinion article: Fox News blaming congressional gridlock on Democrats | Read a news opinion article: MSNBC blaming congressional gridlock on Republicans |
|---|---|---|
| Outcome variable | Participants’ endorsement of including the article on the news organization’s website | |
Study 2.1 & 2.2: affective polarization and avoidance of partisans
| Study 2.1: partisan prejudice in group selection (N = 661) | ||||
|---|---|---|---|---|
| Conditions | Player 1 | Player 2 | Player 3 | Player 4 |
| Education | Some college | High school | College degree | College degree |
| Gender | Female | Male | Female | Female |
| Age | 30 | 32 | 35 | 31 |
| Marital status | Married | Single | Married | Married |
| Political affiliation | Independent | Independent | Republican | Democrat |
| Outcome variable | Choose three players from the 4 player profiles to form a team to complete a series of puzzles (incentivized) | |||
| Study 2.2: perceived partisan bias among the excluded (N = 153) | ||||
| Method | Participants were told that after the team leader viewed their demographics, they were excluded by the team leader for further participation. They were then asked to select the cause of not being picked. | |||
Study 3.1 & 3.2: affective polarization and discriminatory behavior
| Study 3.1: tolerance for the suppression of political action (N = 395) | ||
|---|---|---|
| Treatment conditions (randomized) | Read a newspaper story: Police Use Tear Gas on Peaceful Young Democrat Protest | Read a newspaper story: Police Use Tear Gas on Peaceful Young Republican Protest |
| Outcome variable | Agreement on the decision to use tear gas on the protesters. The cost of this behavior. | |
| Study 3.2: tolerance for political corruption (N = 475) | ||
| Treatment conditions (randomized) | Read a newspaper story: Donations from Millionaire Businessman to Democratic Super PACs in Question. | Read a newspaper story: Donations from Millionaire Businessman to Republican Super PACs in Question. |
| Outcome variable | Support for the investigation of the businessman. | |
In studies 1 and 3, the randomized component of the treatment condition is the partisanship of an actor (congressional representatives, protestors, and donors). This allows Lelkes and Westwood to causally identify the effect of such partisanship in those studies, which is estimated conditional on whether the respondent has the same or different partisan identification as the treatment. The authors then estimated whether these co-partisan and opposing partisan effects are moderated by the respondents’ level of affective polarization. This allows them to determine whether partisans display prejudice for or against in vs. out group members and whether that prejudice is a function of differences in feelings toward such members.
Main Findings
Based on the findings from the five studies conducted, this article suggests that partisans do demonstrate partisan prejudice and affective polarization plays a significant role in shaping certain manifestations of prejudicial behavior. However, the research indicates that affective polarization is more strongly associated with in-group love rather than out-group hate.
In the first study respondents were less likely to share an article criticizing their own party than one critical of the opposing party. However, there was an asymmetric effect of affective polarization on the willingness to suppress or promote hostile rhetoric directed towards the in-group party and the out-group party. The relationship between affective polarization and sharing behavior was stronger for co-partisan criticism than opposition criticism; more affectively polarized co-partisans were much more likely to suppress criticism of their own party than share news critical of their opponents.
Furthermore, the second set of studies similarly showed partisans had a strong preference for forming teams with co-partisans at the exclusion of opposing partisans, even when doing so eliminated more educated individuals from helping solve puzzles. The studies also revealed that individuals who are more affectively polarized were more likely to exclude opposing partisans because of their partisanship. When asked to give reasons why they themselves could have been excluded from the group, the most common response was political party.
The last set of studies, however, shows the limits of partisan prejudice. Majorities of respondents, regardless of shared partisan status, disagreed with the police action and agreed with investigating illegal activities. While affective polarization in both studies increases willingness to protect co-partisans, it does not affect attitudes toward opposing partisans.
Implications
Overall, the results suggest partisans do display prejudiced partisan attitudes, but typically for the benefit of their own party and at lower levels when prejudice becomes more extreme. Importantly, this research helps provide context to the negative consequences of partisanship found in previous work, which often suggests dire consequences for our democracy, and investigates the limitations of partisan prejudice. Although prior studies have found that partisan affect is linked to various negative outcomes, this article proposes that these negative consequences are largely restricted to preferential treatment of co-partisans, with support for more egregious norm violations dissipating as the level of prejudice increases. Furthermore, this study addresses the puzzles surrounding the understanding of partisan bias by distinguishing between different prejudicial behaviors, going beyond zero-sum thinking of in-group favoritism and opposing animosity, and investigating prejudice as a continuum. This article also has broad implications for studies of American politics, indicating a need to re-examine the assumption of symmetry and the nature of division across party lines.
Questions left unanswered
As the authors themselves state, they are unable to directly answer the question of if affective polarization causes prejudiced behavior, as they are unable to directly manipulate the respondents’ level of affective polarization. The authors also identify that there is a disconnect between what partisans are willing to accept in terms of vitriolic rhetoric from their party elites and the actions they may be willing to take or want party elites to take against members of the out party, so an additional unanswered question is how elite rhetoric does or does not influence partisan behavior.
Finally, the authors at the beginning of the paper ask “just how bad has partisan prejudice become in America?” and it is unclear whether they truly found “the limits of partisan prejudice.” Allport (1954) has two additional levels of prejudice left unanalyzed in the paper: physical attack and extermination. While these would be difficult to attitudinally measure, it seems possible partisans would be willing to endorse some level of violence. While there are attenuating effects at higher levels of prejudice in the study, they are not zero, so we do not yet know what the true limits of partisan prejudice could be.
Methods and Analysis
Was the study and its analyses pre-registered?: No
Did the study rely on proxy variables to measure polarization?: No
Were standard p-value thresholds used (p<.05 or 95% Confidence Intervals that don’t overlap zero)?: Yes
- Largest p-value presented as significant: 0.05
Were correlational results interpreted with causal language?: No
Limitations / Weaknesses
The authors themselves identify a number of limitations in their studies. First, the measures of prejudice used in the studies are ultimately attitudinal, not behavioral, meaning the studies may have limited external validity with regard to behavioral demonstrations of prejudice. Second, they note that the studies with randomization presented in the paper (1 and 3) are randomly varying partisanship in a number of settings, not affective polarization itself. While they can identify the effects of partisanship conditional on levels of affective polarization, they cannot causally identify the effect of affective polarization. Third, the authors recognize that their faux newspaper articles are presented independent of partisan media attention that often accompanies political events, which may exacerbate partisan discrimination. Finally, the paper can only speak to the American political system, so similar effects may not extend to other contexts. Additionally, the authors briefly report not finding consistent differences in the responses of Democrats and Republicans in their experiments. It is plausible such a lack of a result is due to sample size constraints in the individual experiments, which range from n = 395 to n = 611 (and, in the supplement to study 2, n = 153). Since the main effects are estimated by splitting the sample into co-partisan and non-copartisan effects, estimating a within-party effect functionally cuts the sample into quarters.
Open Data & Analyses
Does the article make the replication data publicly available?: Yes
Does the article make the replication analysis scripts publicly available?: Yes
Article Citation
Lelkes, Y., & Westwood, S. J. (2017). The limits of partisan prejudice. The Journal of Politics, 79(2), 485-501.
Bibtex
1
2
3
4
5
6
7
8
9
10
@article{lelkes2017limits,
title={The limits of partisan prejudice},
author={Lelkes, Yphtach and Westwood, Sean J},
journal={The Journal of Politics},
volume={79},
number={2},
pages={485--501},
year={2017},
publisher={University of Chicago Press Chicago, IL}
}