Article Summary
Introduction
Much of the concern about affective polarization in the US is rooted in research finding that partisans dislike and distance themselves from members of the other party. These authors argue that this concern may be partially unwarranted, and that most people may be less hostile toward members of the other party than previously thought. They show that reports of widespread hostility are exaggerated by vague survey questions that encourage respondents to rely on inaccurate stereotypes about the opposing party. When questions are reworded to ask people how they feel about ordinary, moderate members of the opposing party, there is much less evidence of partisan hostility. The authors conclude by arguing that this “limited hostility” finding is probably a more accurate picture of most people’s day-to-day interactions with members of the other party.
Analytical Approach
The typical survey measure of affective polarization asks participants to report their feelings about “Democrats” or “Republicans” in general. The authors are concerned that these questions make participants rely on their stereotypes about what kinds of people “Democrats” and “Republicans” usually are. To understand whether these stereotypes influence measures of affective polarization, the authors create an experiment in which they ask questions about affective polarization that describe “Democrats” and “Republicans” more specifically. They vary two parts of the question. First, to examine participants’ stereotypes about partisans being politically extreme, they randomly vary details about partisans’ ideology: (a) no details, (b) “moderate,” or (c) “liberal/conservative.” Second, to examine stereotypes about partisans being highly engaged in politics, they randomly vary details about how often partisans talk about politics: (a) no details, (b) “rarely talk about politics,” (c) ‘occasionally talk about politics,” or (d) “frequently talk about politics.” So, questions might ask participants how they feel about “Democrats”, “moderate Republicans who rarely talk about politics,” or “liberal Democrats who frequently talk about politics.”
These two experimental manipulations mean the experiment has 12 different conditions, shown below:
| Table of Experimental Conditions | Discussion Frequency | ||||
|---|---|---|---|---|---|
| No details | Rare | Occasional | Frequent | ||
| Ideology | No details | 1 | 2 | 3 | 4 |
| Moderate | 5 | 6 | 7 | 8 | |
| Liberal/conservative | 9 | 10 | 11 | 12 | |
In a 13th condition, instead of asking participants how they feel about members of the parties, the authors ask participants to report their beliefs about what a typical Democrat and a typical Republican are like. In particular, they ask questions about partisans’ ideology (whether they are moderate or extreme) and frequency of discussion (whether they talk about politics rarely, occasionally, or frequently). They compare these beliefs to participants’ actual responses about their own ideology and discussion frequency to identify what kind of stereotypes participants hold about partisans.
Main Findings
The paper has three basic findings. (1) People overestimate how extreme members of the other party are, as well as how frequently they talk about politics. (2) People express more hostility toward the other party when they’re asked about partisans who are described as politically extreme or talking a lot about politics. (3) When survey questions just ask about “Democrats” or “Republicans” with no additional details (condition 1 in the table above), people express a lot of hostility toward the other party because their stereotypes fill in those details for them. In fact, they express nearly as much hostility on these questions as they do when explicitly asked about extreme partisans who frequently discuss politics!
Implications
This study shows that affective polarization may not be as severe a problem as many think. Many people hold inaccurate stereotypes about members of the other party being politically extreme and engaged. However, most members of both parties are relatively moderate and relatively uninterested in politics. When survey questions reflect this reality and ask how participants feel about moderate, politically uninterested partisans, people seem much less hostile toward the out-party than suggested by previous research. Thus, from a methodological perspective, scholars need to carefully consider what kinds of survey measures they use to estimate affective polarization and who participants may be imagining when they express partisan hostility.
Substantively, this paper also suggests that survey research on affective polarization likely overstates the extent of partisan hostility in everyday life. When we interact with other ordinary people, we don’t typically rely just on stereotypes because we have much more information about them than we do when answering vague survey questions about “Democrats” or “Republicans.” The information we have about their personal ideology and engagement in politics may thus outweigh the stereotypes we have about partisans in general.
Finally, the importance of stereotypes about the other party suggest that correcting these stereotypes may reduce partisan hostility. If scholars and journalists can inform the public that most people are relatively moderate, these stereotypes may become less harmful. Alternatively, if people can be encouraged to rely on their stereotypes less—for instance, by instead considering their past interactions with ordinary members of the other party—the influence of partisan hostility in everyday life may be lessened.
Questions left unanswered
Much of the substantive conclusion of this article focuses on how people are likely much less hostile toward ordinary, moderate members of the opposing party than past survey questions have led us to believe. However, for people to get along with members of the other party, they need to be willing to talk with them in the first place. The authors highlight that most people regularly interact with at least a few members of the other party, but it may nonetheless be the case that affective polarization and stereotypes about the out-party make people less likely to connect with members of the other party. The nature and severity of the social consequences of affective polarization remain an open question for the literature.
On a more granular note, the authors focus on stereotypes about political ideology and engagement, but past work has also highlighted a variety of other negative stereotypes that people tend to hold about the out-party (for instance, that they are selfish or unpatriotic). Do these kinds of stereotypes also influence survey responses about affective polarization? If so, how should scholars account for them methodologically, and what relevance do they have for people’s willingness to interact with ordinary members of the other party?
Methods and Analysis
Was the study and its analyses pre-registered?: Yes
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 mostly do not separate feelings about the in-party from feelings about the out-party in their discussions. Some past work suggests that people stereotype and feel negatively about extreme and engaged co-partisans as well, not just out-partisans. In Supplementary Information 10, the authors show that participants who were asked questions about co-partisans who occasionally or frequently talk about politics felt more negatively about them. However, participants also reported more negative attitudes toward co-partisans who rarely talk about politics (compared to the no-details condition), and the ideology manipulation did not influence in-party affect either. Both of these latter findings conflict with past work and the authors might have briefly discussed them.
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
Druckman, J. N., Klar, S., Krupnikov, Y., Levendusky, M., & Ryan, J. B. (2022). (Mis)estimating affective polarization. The Journal of Politics, 84(2), 1106–1117. https://doi.org/10.1086/715603
Bibtex
1
2
3
4
5
6
7
8
9
10
11
12
13
@article{doi:10.1086/715603,
author = {Druckman, James N. and Klar, Samara and Krupnikov, Yanna and Levendusky, Matthew and Ryan, John Barry},
title = {(Mis)estimating Affective Polarization},
journal = {The Journal of Politics},
volume = {84},
number = {2},
pages = {1106-1117},
year = {2022},
doi = {10.1086/715603},
URL = {https://doi.org/10.1086/715603},
eprint = {https://doi.org/10.1086/715603
}