Article Summary
Introduction
Scholars have argued that polarization among elected officials in the United States has a variety of harmful effects on citizens and democracy in general. However, polarization is not just one phenomenon. Political elites in the United States are polarized both on the issues and their incivility toward each other. This author asks whether the public responds differently to elites who are polarized on the issues than to elites who are uncivil and suggests that these two aspects of elite polarization may have three separate kinds of consequences for citizens. First, elite issue polarization may make citizens trust politicians less because they feel the elites are too politically extreme to represent ordinary Americans. Similarly, elite incivility may make politicians seem unlikely to observe ‘the rules of the game,’ further reducing trust in politicians. Second, elite issue polarization may make it easier for people to identify their own party’s preferred position and follow that cue, increasing attitudinal polarization between Democrats and Republicans in the public. This might be further strengthened if elite incivility promotes anger, which makes people more likely to rely on party cues. Third, elite issue polarization may increase citizens’ affective polarization because it increases the stakes of voting and amplifies in-group favoritism. Similarly, elite incivility may increase citizens’ affective polarization because it reinforces anger and stereotypes about the out-party.
In short, elites are becoming more polarized on the issues and more uncivil. These trends may undermine citizens’ political trust, increase their attitudinal polarization, and increase their affective polarization. But because existing research treats elite issue polarization and incivility as interchangeable, it is unclear whether they have different consequences in practice.
Analytical Approach
To explore the potentially distinct effects of elite issue polarization and incivility, the author conducts a main experiment and three follow-up studies. In the main experiment, the author recruited a nationally representative sample of 1,279 people in the U.S. Participants were assigned to read a short vignette about Congress discussing either off-shore drilling or regulations for air traffic controllers. These vignettes randomly varied how divided the politicians were on the issue (either highly polarized or not especially polarized) and how civil the politicians’ discussion was (either civil or uncivil). A final group of participants were given no information about politicians’ polarization or civility, representing a control condition. Then, participants were assigned to read another vignette about the other issue featuring the same level of issue polarization and civility (or control).
| Experimental Conditions | Target of Threat | |
|---|---|---|
| In-party | ||
| Type of Threat | Party-based | 1 |
| Issue-based | 3 | |
After reading the two vignettes, participants in the main study were asked questions about their trust in politicians, their own policy attitudes (to capture attitudinal polarization), and their feelings about the parties (to capture affective polarization).
In the follow-up studies, the author considered three extensions of the main study’s design. In the first follow-up, the author used the same basic design but adjusted the vignettes so that they featured stronger incivility in case the main study’s findings were driven by relatively mild impoliteness.
In the second follow-up, the author considered whether meta-commentary included in the main study’s vignettes might have affected the findings. To do so, the author presented participants with only one vignette and manipulated only incivility. Most importantly, the author did so only with quotes from the politicians rather than by also describing the tone of debate as “harsh” or “respectful.”
In the third and last follow-up, the author considered whether the pattern of findings was unique to the experimental setting. To do so, the author conducted a non-experimental survey asking about participants’ existing beliefs about elite issue polarization and incivility, and tested whether those were related to participants’ own trust, attitudinal polarization, and affective polarization.
Main Findings
Across all studies, both elite issue polarization and elite incivility make citizens more affectively polarized. However, only elite issue polarization makes citizens more attitudinally polarized themselves, and only elite incivility makes citizens trust politicians less. Thus, elite issue polarization and elite incivility have related—but distinct—effects on the public.
Implications
This article’s finding that elite issue polarization and elite incivility have different effects is important. Many studies examine public responses to either issue polarization or incivility, but the author emphasizes that scholars interested in these topics need to take care to separate them. In particular, the author recommends three things. First, scholars should take care to manipulate either issue polarization or incivility with words that do not also implicitly cue the other construct. Second, because people may infer the presence of one construct from cues of the other, scholars should include information about both issue polarization and civility to ensure that manipulations precisely target the construct of interest. Third, scholars should include post-treatment manipulation checks to ensure that manipulations successfully influenced the construct of interest and rule out concerns about confounding.
After examining the control condition in both the main study and the first follow-up study, the author notes that the effects of the civility manipulation on both trust and affective polarization were driven primarily by beneficial effects of the civil condition, not by harmful effects of the uncivil condition. That is, people who read an uncivil vignette reported trust and affective polarization no differently than people who read the story without any information about issue polarization or civility. When they read a civil vignette, however, participants reported more trust and less affective polarization. Thus, people seem to already think that politicians are highly uncivil, and further exposure to elite incivility may be unlikely to continue undermining trust in politicians or increasing affective polarization.
Questions left unanswered
The finding that incivility did not influence issue polarization suggests a range of possibilities for future work. The author particularly highlights that the designs of the studies in this paper explicitly conveyed to participants that both Democrats and Republicans were uncivil—that is, the manipulations conveyed two-sided incivility. In this symmetric context, incivility may not promote attitude change. However, in the real world, if people feel that the other party is the only party behaving badly, incivility may have different effects. The author notes that the effect of incivility on affective polarization partially undermines the “incivility only matters when it is one-sided” possibility but nonetheless suggests that future work should take up this question more seriously.
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?: Yes
In the author’s discussion of the final, non-experimental follow-up study (introduced on p. 1469), they describe non-causal relationships from the survey as “effects” rather than “associations” or “relationships.”
Limitations / Weaknesses
As noted above, the final study in the article (introduced on p. 1469) is non-experimental. It relies on survey data to examine baseline correlations between perceptions of incivility, perceptions of issue polarization, and the three theoretical outcomes in the study (i.e., affective polarization, trust, and attitudinal polarization). The author’s wording suggests that these relationships are “effects,” but given that the data are correlational, they would be more accurately described as relationships or associations. The evidence in this last study is not causal. However, as the author conducted several true experimental studies, the broader argument of the article is still well supported.
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
Skytte, R. (2021). Dimensions of elite partisan polarization: Disentangling the effects of incivility and issue polarization. British Journal of Political Science, 51(4), 1457–1475. https://doi.org/10.1017/S0007123419000760
Bibtex
1
2
3
4
5
6
7
8
9
10
@article{skytte_2021,
title={Dimensions of Elite Partisan Polarization: Disentangling the Effects of Incivility and Issue Polarization},
volume={51},
DOI={10.1017/S0007123419000760},
number={4},
journal={British Journal of Political Science},
publisher={Cambridge University Press},
author={Skytte, Rasmus},
year={2021},
pages={1457–1475}}