Article Summary
Introduction
As concern grows among social scientists that increasing disdain for opposing partisans may drive support for anti-democratic politics and political violence, interest in depolarization interventions has increased dramatically. Many of these interventions have proven quite effective in reducing affective polarization among respondents through moderating negative sentiments toward opposing partisans. But do these depolarization interventions reduce anti-democratic attitudes? Many researchers leveraging such depolarization interventions conjecture on downstream effects on anti-democratic attitudes, but few directly test that relationship.
To that end, Voelkel et al aim to directly test the effectiveness of depolarization interventions on reducing support for anti-democratic candidates, policies, and violence. Using 3 interventions over 3 studies, they replicate previous findings supporting the claimed effects of interventions on affective polarization, extend the interventions’ effects to behavioral measures of affective polarization, and determine if a downstream relationship exists with anti-democratic attitudes.
This work is important for the tailoring of specific interventions to anti-democratic attitudes. Anti-democratic attitudes and violent behavior are certainly threats to a well-functioning civil society, so if depolarization efforts can effectively alleviate such ills while also reducing disdain for opposing partisans, efforts should be made to maximize their reach and effectiveness. However, if they are not effective in doing so, as the authors find, researchers and practitioners must consider alternative interventions tailored to the distinct psychological processes behind anti-democratic support.
Analytical Approach
Voelkel et al leverage both correlational and experimental analysis to shed light on the alleged relationship between depolarization interventions, affective polarization, and support for anti-democratic policies and political violence.
The authors use the standard feeling thermometer approach to measure affective polarization post-treatment, along with a battery of questions on support for an anti-democratic candidate, suppressing out-party votes through redistricting, and support for violence against opposing partisans. They also employ behavioral measures of out-party affect through two games. In the “dictator game,” respondents must allocate money to fellow partisans or opposing partisans. In the “joy of destruction” game, respondents must decide how much money they are willing to sacrifice to reduce out-party earnings.
The authors leverage 3 experimental interventions over 3 studies, described in detail in the table below. Respondents were randomly assigned to 1 (study 1) or 2 (studies 2 and 3) interventions. In the out-party friendship intervention, respondents in the treatment condition were primed to consider friends who belong to the out-party group. In the corrected policy misperceptions treatment, respondents’ registered perceptions of out-party partisan beliefs and had those beliefs subsequently corrected with real information. Finally, in the warm elites relation intervention, respondents were informed of the close friendship of two famous politicians from opposite parties. Studies 2 and 3 were preregistered. Study 1 was not.
Table 1: Randomized conditions in Voelkel et al. 2023
| Intervention | Study | Treatment | Control |
|---|---|---|---|
| Outparty friendship salience | Study 1 (n = 2,341) | “Although you are [a Democrat/an Independent who is closer to the Democratic Party/an Independent who is closer to the Republican Party/a Republican], you probably know people who are [Republicans/Democrats]. Think about one such [Republican/Democrat] that you like and respect a great deal. This person could be a friend, relative, neighbour, co-worker, or just someone that you know. Please explain why you feel this way about this person” | NULL (skip to dependent variable measures) |
| Corrected policy support misperceptions | Study 1, Study 2 (n = 2,021) | Respondent presented with 1 of 5 policy scenarios and asked baseline questions on outparty policy support. Respondents were then given real responses from in- and out-party members from a similar scenario. | NULL |
| Warm elites relation | Study 3 (n = 4,023) | Participants watched video about the friendship of Joe Biden and John McCain. | Watch video about history of neckties |
Studies 2 and 3 randomized the post-treatment order of affective polarization, anti-democratic attitude, and political violence items. Bayesian analysis was used as a follow-up to null results to confirm the relative performance of the null and alternative hypotheses, and results are confirmed using a meta-analysis of existing research on depolarization interventions on anti-democratic attitudes.
Main Findings
Voelkel and coauthors successfully replicate the negative effect of depolarization interventions on affective polarization, and show the effects persist in the behavioral measures as well. However, none of the interventions significantly reduced anti-democratic support or support for political violence. These results persist after follow-up Bayesian analysis and a meta-analysis of previous work.
Starting with the correlational results, the authors find significant correlations between feeling thermometer operationalization of affective polarization and behavioral measures, confirming at least some overlap in attitudinal underpinnings. However, they find only weak, inconsistent, or nonexistent correlations between affective polarization and anti-democratic attitudes and support for political violence.
In the experimental treatments, all interventions are associated with reductions in affective polarization through the reduction in negative out-party affect, with reductions in out-party affect ranging from around 2 to 4 points on a 101 point scale. These reductions persist in the behavioral games, with similar effect sizes. However, the misperception correction intervention had no effect on respondent behavior in the joy-of-destruction game.
The experimental interventions had no effect on downstream attitudes related to democracy and violence. Across all studies and all conditions, the authors detected no effect. These effects are confirmed in the Bayesian analysis, where even though some effects are not null, they are in the opposite of the expected direction.
Implications
This study yielded both positive and negative implications for researchers and practitioners in the social sciences. Positively, it showed depolarization interventions do work when analyzed with regard to their intended target, both attitudinally and behaviorally. This is hopeful for efforts to reduce affective polarization in the American public. While scaling considerations are undoubtedly present, there is hope in this study that negative sentiment toward opposing partisans can be ameliorated for a better functioning civil society.
A negative implication, however, is the benefits of depolarization efforts stop at affective polarization. The existence of anti-democratic attitudes and support for political violence is tempered by its relatively small baseline, but that baseline proves durable to depolarization efforts. As practitioners consider avenues through which such attitudes can be attenuated, they must consider interventions more specifically tailored to the psychological uniqueness of anti-democratic support. There is seemingly no “one-size-fits-all” approach to more undesirable political attitudes.
Questions left unanswered
While this study finds no effects of depolarization efforts on anti-democratic attitudes, that does not mean there are no downstream effects at all. Perhaps such efforts could increase related positive behaviors, such as willingness to compromise. Additionally, only 3 depolarization efforts were considered, and even though these have been proven to successfully reduce affective polarization, perhaps less effective interventions in that domain could prove effective in other domains.
Finally, the full scope of anti-democratic attitudes potentially impacted by depolarization efforts is unknown. The researchers hold a fairly high bar for the depolarization efforts, as the operationalization of measures of anti-democratic support make very clear they are about the erosion of democracy. Certain anti-democratic practices, however, may be more perceptually benign, and such perceptions could make attitudes related to those practices more malleable.
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 authors insightfully correct a common but erroneous assumption among political scientists about the downstream effects of certain depolarization interventions on anti-democratic attitudes. By their own admission, they do not test all such interventions; just ones “perceived to be particularly promising ways to reduce affective polarization.” Herein may lie the basis of the subsequent null results; the interventions are the most effective for affective polarization, meaning they’re most aligned both psychologically and instrumentally with the underpinning of affective polarization, not anti-democratic attitudes. It is possible that less effective but more generalized interventions may have more success. Take, for instance, recent work documenting what partisans have in mind when answering affective polarization questions. Druckman and Levendusky (POQ 2019) find respondents are more often answering the question with regard to party elites than to voters. Given the interventions in question often have more interpersonal framings, it seems possible the estimated reduction in affective polarization is due to a change in framing from elite evaluation to interpersonal sentiment. The question wordings for the anti-democracy and political violence items, however, seem less ambiguous and therefore more resistant to such changes in framing. Future work should consider how interventions change the ways in which respondents evaluate survey items.
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
Voelkel, J.G., Chu, J., Stagnaro, M.N. et al. Interventions reducing affective polarization do not necessarily improve anti-democratic attitudes. Nat Hum Behav 7, 55–64 (2023). https://doi.org/10.1038/s41562-022-01466-9
Bibtex
1
2
3
4
5
6
7
8
9
10
11
12
13
14
@article{voelkel_interventions_2023,
title = {Interventions reducing affective polarization do not necessarily improve anti-democratic attitudes},
volume = {7},
issn = {2397-3374},
url = {https://doi.org/10.1038/s41562-022-01466-9},
doi = {10.1038/s41562-022-01466-9},
number = {1},
journal = {Nature Human Behaviour},
author = {Voelkel, Jan G. and Chu, James and Stagnaro, Michael N. and Mernyk, Joseph S. and Redekopp, Chrystal and Pink, Sophia L. and Druckman, James N. and Rand, David G. and Willer, Robb},
month = jan,
year = {2023},
pages = {55--64},
}