Article Summary
Introduction
A significant body of research documents the negative consequences of affective polarization in American politics, including increases in social distance, professional discrimination, stereotyping, and intolerance. However, other research suggests a positive externality: an increase in political participation. Theoretically, this seems plausible. Affective polarization may generate mobilizing emotions such as anger, which seem to have a positive relationship with political participation. Additionally, affective polarization may raise the perceived stakes of an election, motivating higher rates of turnout. In their paper, Ahn and Mutz critically assess whether increases in affective polarization are associated with increases in political participation.
The authors list multiple contributions of their paper, the most important of which is that they are the first to focus their analysis on the relationship between affective polarization and political participation. Second, they expand the scope of the study of affective polarization by comparing measures of affective polarization with respect to parties and presidential candidates, with the expectation being the latter generates more intense behavioral responses. Third, they utilize both self-reported and validated voter turnout, allowing for a comparison between reported and actual behavior. Finally, they are able to assess the relationship causally by implementing a panel study, reducing the threat of omitted variable bias.
Analytical Approach
The authors utilize two types of analyses: cross-sectional and panel. In the cross-sectional study, the authors use data from the American National Election Studies (ANES) from 1980 to 2020. The panel study relies on data from a nationally representative probability panel collected by Amerispeak/National Opinion Research Center from 2016 to 2020. For some ANES presidential election years and all of the panel study, voter validation was performed to determine actual voting behavior in addition to pre- and post-election self-reported intention/voting.
The main independent variables are measures of affective polarization provided by 0-100 feeling thermometers. The standard affective polarization measures takes the absolute difference in feeling thermometers between the two parties, whereas the candidate polarization measure takes the absolute difference between the two presidential candidates. The resulting differences are binned 1-20 to account for clustering around round numbers ending in 0 and 5.
The main dependent variables are intent to vote, self-reported turnout, and validated voting, which are all measured dichotomously. Additional analyses are performed with a dependent variable for campaign participation, which is based on an index of 5 campaign activities rescaled 0-1. Cross-sectional analyses was performed with logistic regression and calculated both within- and across-years, and panel analysis was done with a difference-in-difference design between the 2016 and 2020 elections. Standard demographic controls (partisan strength, employment status, education, income, gender, race, and age) were included in the models.
Main Findings
Taken together, the findings from the validated vote analyses suggest affective polarization (for both candidates and parties) has little to no impact on turnout. In both the cross-sectional and panel analyses, neither affective polarization nor candidate differences were associated with a significant or substantial increase in validated voting for any individual year.
These results stand in contrast to the cross-sectional results for self-reported voting behavior and campaign participation. In those results, candidate thermometer differences were consistently positive associated with political participation, and affective polarization had smaller but generally positive effects as well. This suggests self-reported behavior is likely a function of the same affective pressures voters experience with regard to affective polarization, but are not necessarily grounded in reality. Indeed, post-hoc analyses suggest strength of partisanship was a significant predictor of one’s overreporting of their own voting behavior.
Descriptively, the authors do show both types of affective polarization, self-reported voting behavior, and campaign participation have risen from 1980 to 2020. Extrapolating the effects from the cross-sectional validated voting models, the authors suggest about 32% of the 1.5 percentage point increase in turnout from 2016 to 2020 is attributable to increases in candidate thermometer differences.
Implications
The findings of this paper cast doubt on the notion of affective polarization being a “double-edged sword.” To the extent any positive externalities of affective polarization exist, they are likely small and candidate-focused. This suggests a further reconsideration of turnout being the “gold standard” for judging U.S. democracy; turnout has increased in recent years, but at the cost of increased polarization and perceptions of election illegitimacy.
More technically, this paper emphasizes the importance of validated vote turnout measures for analyzing political behavior. The findings between the self-reported variables and the validated measures diverge drastically, suggesting there are unmeasured cognitive or emotional processes at play when survey respondents respond to political participation questions. Researchers should beware when those psychological processes are potentially related to their behavior of interest.
Finally, the results emphasize the importance of presidential candidates in affective judgements. Differences in candidate feeling thermometers were consistently stronger indicators of self-reported voting than affective polarization of parties. This suggests a potential in-road against polarization through better candidate selection, although it also leaves open the possibility of more nefarious actors entrenching animosity even further.
Questions left unanswered
The authors note they do not attempt to separate the two mechanisms (mobilizing emotions and increase of stakes) linking affective polarization and political participation because polarization likely reflects both and their focus is on the robustness of the relationship between polarization and participation. However, this does leave unanswered whether the lack of a relationship is because affective polarization does not increase mobilizing emotions/perceptions of stakes, or if those mechanisms do not actually cause increases in political participation.
Additionally, the title of the paper “does hating the other side motivate voters?” is never truly answered, as the analyses are focused on measures of affective polarization rather than component measures of in-group love versus out-group hate. Recent research suggests these component measures have different effects on behavior depending on setting (Rudolph and Hetherington, 2021), so it would be interesting (and easy for a replicator given the availability of data) to perform the analyses on the component affective measures.
Methods and Analysis
Was the study and its analyses pre-registered?: No
Did the study rely on proxy variables to measure polarization?: Yes
Authors relied on the original 0-100 feeling thermometer, but also added the 0-100 feeling thermometer for presidential candidates. Both measures were binned into categories from 1-20 for analysis.
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
This study is one of the most careful considerations of the effects of affective polarization in the literature, and the authors should be commended for their thoroughness in evaluating a very clear research question. Due to the time period of the panel (2016-2020), the authors cannot rule out the lack of an effect is in part due to the candidates contesting the election. In particular, only the Democratic candidate changed between elections, and we may expect variation in candidate feeling thermometer differences to be fairly stable between elections (this may not be the case, so some descriptives would be useful). The authors suggest we might believe Trump is particularly divisive and therefore drives any aggregate result, but because Trump is the one constant in the panel, we cannot critically assess this claim. Additionally, while not a limitation of the study, it would be interesting to know how the electorate changes if we think affective polarization has some small effect on turnout. The authors grapple quite heavily with the normative component of turnout in democracy. Previous work suggests certain GOTV efforts increase inequality within the electorate (Enos, Fowler, and Vavreck 2014), so is the same potentially true for affective polarization? More plainly, if affective polarization motivates turnout, does it do so in a way that makes the electorate more representative? Given the small estimated effect size, this is likely a question only answerable with a larger sample size.
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
Ahn, C., & Mutz, D. C. (2023). The Effects of Polarized Evaluations on Political Participation: Does Hating the Other Side Motivate Voters? 87(2), 243–266. https://doi.org/10.1093/poq/nfad012
Bibtex
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
@article{10.1093/poq/nfad012,
author = {Ahn, Chloe and Mutz, Diana C},
title = "{The Effects of Polarized Evaluations on Political Participation: Does Hating the Other Side Motivate Voters?}",
journal = {Public Opinion Quarterly},
volume = {87},
number = {2},
pages = {243-266},
year = {2023},
month = {05},
issn = {0033-362X},
doi = {10.1093/poq/nfad012},
url = {https://doi.org/10.1093/poq/nfad012},
eprint = {https://academic.oup.com/poq/article-pdf/87/2/243/50873561/nfad012.pdf},
}