Article Summary
Introduction
When scholars, journalists, and members of the public say “The American public is polarized” or “The American public isn’t polarized,” what do they actually mean by ‘polarized’—and who’s right? This paper explores several different ways that mass (i.e., public) polarization has been conceptualized and measured, and tests whether each of these arguments is accurate.
The author first considers mass ideological polarization. There are two scholarly camps on whether the U.S. public is ideologically polarized. One camp argues that polarization should be determined based on whether, across issues, Democrats are consistently liberal and Republicans are consistently conservative. This is what the author calls the ‘ideological consistency’ account. Most scholars in this camp argue that the mass public is quite polarized because most Democrats are liberal and most Republicans are conservative. A second group of scholars, the ‘ideological divergence’ camp, argues that the public is not very polarized because most people have a fair amount of common ground on their issue positions—they don’t tend to disagree intensely with members of the other party. Accordingly, as a first step in testing the extent of mass polarization, the author examines the evidence for these two accounts.
A largely separate body of scholarship explores how much the U.S. public thinks politics is polarized, regardless of how polarized it actually is. Scholars interested in this phenomenon have argued this perception of ideological polarization has increased over time and that it may be contributing to ‘real’ polarization because it shapes how people form attitudes and feel about the parties. The author’s second step, then, is testing whether people perceive more polarization now than they did in the past.
Finally, the author examines affective polarization, which refers not to differences in issue positions (or perceptions of differences in issue positions), but instead to whether people feel positive or negative toward each party. People who are affectively polarized feel warm and favorable toward their own party but cold and unfavorable toward the other party.
Analytical Approach
To examine the ideological consistency and ideological divergence accounts of polarization, the author first reviews the past evidence presented by the ideological consistency camp. Then, the paper considers the ideological divergence argument in three steps. First, the author examines the U.S. public’s responses to a 7-point “extremely liberal” to “extremely conservative” scale used frequently in political science. The American National Election Study (ANES), a long-running survey conducted on representative samples of the U.S. publics, has asked respondents to complete this ideology scale every 2-4 years between 1972 and 2012. The ideological divergence account suggests that the public is polarized if a lot of people are liberal, a lot of other people are conservative, and there aren’t many people in between—that is, if the distribution of ideology is bimodal. To test this formally, the author calculates the distribution’s bimodality coefficient over time. The author also examines the overlap coefficient, a measure of how much common ground exists between self-identified Democrats and Republicans (rather than all Americans). After examining responses to the 7-point ideology scale, the author calculates both the bimodality and overlap coefficients a second time, this time based on people’s overall ‘latent ideology,’ or their positions across a range of policy attitudes.
Second, the author examines how perceived polarization has changed over time. Here, the author uses the overlap coefficient with the ANES data again, but this time relying on the public’s placements of the parties on a range of policy issues. For instance, given the topic ‘how much should the government spend on services such as health and education’, a participant might think the Democratic party falls at one end and the Republican party at the other end (or both somewhere in the middle). The author also considers whether these perceptions differ between Democrats, Republicans, and independents. Finally, the author examines how much people’s perceptions of polarization are wrong—that is, how much more polarization they think exists than actually does—by comparing the overlap coefficients for Democrats and Republicans’ real attitudes with the overlap coefficients for the attitudes people perceive them to have.
Third, the author reviews a range of different kinds of evidence presented on the subject of affective polarization. The author particularly focuses on considers three kinds of evidence: (1) Work using ‘feeling thermometers’ that ask people to rate their feelings toward the parties on 0-100 scales (i.e., from very cold to very warm, or very unfavorable to very favorable); (2) work focusing on social distance, or how comfortable partisans feel about the prospect of interacting with members of the other party; (3) work focusing on partisan stereotypes (e.g., how much members of each party feel the other party is typically intelligent, patriotic, or selfish).
Main Findings
The question of “how polarized is the American public?” fundamentally depends on the answer to “what kind of polarization?” The author finds that the mass public in the United States has not become more ideologically divergent. That is, there are still many people who do not identify as extreme liberals or extreme conservatives and hold a mix of moderate policy positions. However, the subset of the public who identify as Democrats or Republicans is more polarized and has become increasingly party-consistent in their policy views. In other words, even though the public in general has not polarized, Democrats and Republicans in particular have less in common than they did 50 years, and each party has a more-and-more clearly defined set of policy positions.
People also perceive more ideological polarization in the U.S. than they used to. This is particularly true for Democrats and Republicans. However, given the increase in actual polarization between Democrats and Republicans, there is no reason to believe they overestimate it any more than they used to. If anything, the public’s beliefs about division between Democrats and Republicans are increasingly accurate.
Finally, Democrats and Republicans in the U.S. public are clearly more affectively polarized than ever before. Partisans’ feeling thermometer ratings of their own party have not changed, but their ratings of the other party have become substantially more negative. Partisans also feel more socially distant from the other party and stereotype them more.
Implications
This paper’s primary contribution is to organize the existing scholarly literature on polarization and put it all into conversation. This paper helps track the different arguments made about polarization, compare them explicitly, and then test them directly with clear, apples-to-apples data.
This article also presents and develops several ways to measure polarization, some of which have not been taken up in the literature. The bimodality and overlap coefficients are rarely used to study polarization even in 2023, and the author’s application of the bimodality coefficient to politics was novel.
Questions left unanswered
This article was focused on describing the past and present (as of 2016) state of polarization. Accordingly, it leaves open the questions of “what causes these kinds of polarization?” and “what do these kinds of polarization do to politics?” The author also notes that other kinds of mass polarization exist that are harder to examine over time, most notably geographic sorting of the parties and polarization of basic values. Finally, this article was focused on mass polarization. The state of elite polarization, and the links between elite polarization and mass polarization, are issues left to other scholars and other articles.
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 author relied mostly on interpreting past results rather than providing original analysis for the discussions of ideological consistency and affective polarization. Given the range of approaches to measuring these phenomena, more empirical attention may have been warranted in this review to help readers understand the strength of evidence for these phenomena and how they have changed over time.
Open Data & Analyses
Does the article make the replication data publicly available?: Yes
Does the article make the replication analysis scripts publicly available?: No
Article Citation
Lelkes, Y. (2016). Mass polarization: Manifestations and measurements. Public Opinion Quarterly, 80(S1), 392–410. https://doi.org/10.1093/poq/nfw005
Bibtex
1
2
3
4
5
6
7
8
9
10
11
12
13
14
@article{10.1093/poq/nfw005,
author = {Lelkes, Yphtach},
title = "{Mass Polarization: Manifestations and Measurements}",
journal = {Public Opinion Quarterly},
volume = {80},
number = {S1},
pages = {392-410},
year = {2016},
month = {03},
doi = {10.1093/poq/nfw005},
url = {https://doi.org/10.1093/poq/nfw005},
eprint = {https://academic.oup.com/poq/article-pdf/80/S1/392/6864582/nfw005.pdf},
}