Article Summary
Introduction
Outside the US, many political systems feature more than two parties, which raises the question of how to measure affective polarization in these multi-party systems.
The author poses three key questions to take account of these party system differences: How can we effectively measure affective polarization in multi-party settings? How do levels of affective polarization compare between countries? And how does it relate to factors such as ideological polarization, democratic values, and political participation?
By addressing these questions, this article provides a conceptual framework for measuring affective polarization in multi-party contexts. Responses from individuals in multi-party systems can vary significantly. For instance, some individuals might have positive feelings toward two political parties and negative feelings toward another party, while others may feel positive about only one party and negative about two others. Furthermore, parties can vary in their importance within multi-party systems, and the size and relevance of these parties should be taken into account when calculating affective polarization scores in systems with more than two parties.
This article proposes ways to compute affective polarization scores in multi-party settings, enabling meaningful comparisons of affective polarization levels between the United States and other countries worldwide.
Analytical Approach
The author develops several measures of affective polarization in multi-party settings and mainly focuses on a measure that averages the absolute difference between the like-score for each party and the average like score across all parties. To take account of the varying party sizes in multi-party settings, the author proposes to weigh the average by the vote share of each of the parties. As a result, (dis)like for a larger party will have more weight in individuals’ affective polarization scores than smaller ones. The author also considers variants of this score (e.g., mean distance from the most-liked party) in the conceptual framework and examines the associations between these variants in the empirical section of the article.
The article uses data from the Comparative Study of Electoral Systems (CSES) project, a cross-national survey project that has been fielded multiple times since 1996. Overall, the data entails 166 surveys and 51 countries, including the United States. The author first descriptively analyzes the proposed weighted polarization score along with the alternative measures and compares levels of affective polarization across countries. The analysis then turns to studying a number of correlates of affective polarization: partisanship, ideological (i.e., polarization on a left-right/liberal-conservative continuum), democratic values, and political participation. Each of these correlates is considered separately in the analysis and examined with linear regression models that add (fixed effects) control variables for each country survey.
Main Findings
The analysis shows that the US has a lower level of affective polarization compared to most other countries worldwide, even lower than European countries such as Sweden or Ireland. However, there are stark differences between partisans and non-partisans in the US: people who identify as either Republican or Democrat are much more affectively polarized than individuals who do not self-declare as partisans. This difference between the two partisan vs. nonpartisan groups is the most pronounced in the US compared to other countries. Furthermore, the author finds that the variants of the affective polarization score do not deviate significantly from each other, suggesting that the exact computation of affective polarization scores in multi-party systems makes only a marginal difference. Lastly, the author reports that affective polarization is positively correlated with identifying with a political party and negative partisanship (i.e., respondents reporting that they would never vote for a specific party).
At the same time, while the association is strong, the author highlights that it is far from perfect, suggesting that partisanship and affective polarization are not identical concepts. The author also shows that affective polarization is positively associated with ideological polarization, democratic values (i.e., considering voter turnout important and believing that it matters who is in power), and participation in elections. However, affective polarization is negatively correlated with satisfaction with democracy.
Implications
A key implication of the paper is that there are different measures by which affective polarization in multi-party systems could be gauged, but these differences do not lead to substantial variations in affective polarization scores across countries. Furthermore, despite the significant attention American society has received in scholarly work on affective polarization, it exhibits a relatively low level of affective polarization compared to most countries worldwide. The conceptual framework for comparable measures of affective polarization enables researchers to compare levels of affective polarization across countries, including the US. This is a valuable contribution for future research projects that seek to study affective polarization cross-nationally and assess the extent of partisan animosities in a comparative perspective.
Questions left unanswered
The author points to several unanswered questions that could be addressed in future research, drawing on the proposed measures of affective polarization in multi-party settings. First, as the author has identified substantial variation between countries, the question arises: What explains different levels of affective polarization? Among other factors, do certain economic conditions or political institutions help explain the varying levels of affective polarization, or do individual-level characteristics account for most of the varying degrees of partisan animosities?
Second, how are short-term events, such as elections or other salient political occurrences, affecting affective polarization? Third, as the author highlights throughout the article, the comparative evidence presented in the analysis is based on correlations. Future research may leverage experimental designs to examine the causal direction of the variables studied in the article.
Methods and Analysis
Was the study and its analyses pre-registered?: No
Did the study rely on proxy variables to measure polarization?: Yes
Like-dislike scale for political parties (and not their supporters) was used; response scale ranges from 0-10 (and not from 0-100)
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
While most respondents in the US can be expected to have heard of and formed an opinion about the two dominant parties (Republicans and Democrats), this might not be the case in countries with more than two parties. Some countries, such as The Netherlands and Israel, feature a high number of political parties, and citizens might not be aware of all of them or may not have contemplated their own stance toward these parties extensively. Consequently, this could result in more unstable responses to questions about party likability in multi-party systems compared to stable two-party systems. This potential issue is partially mitigated by the proposed multi-party affective polarization score, in which smaller parties are assigned lower weights than larger ones. Nevertheless, the presence of a greater number of parties may lead to more volatile attitudes and feelings towards these parties, potentially making comparisons between two-party and multi-party systems more challenging.
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
Wagner, M. (2021). Affective polarization in multiparty systems. Electoral Studies, early view.
Bibtex
1
2
3
4
5
6
@article{Wagner.2021,
author = {Wagner, Markus},
year = {2021},
title = {Affective polarization in multiparty systems},
journal = {Electoral Studies},
}