Skip to main content
Log in

Need for Chaos and Dehumanization are Robustly Associated with Support for Partisan Violence, While Political Measures are Not

  • Original Paper
  • Published:
Political Behavior Aims and scope Submit manuscript


Recent, high-profile acts of partisan violence have stimulated interest among academics and the public in the etiology of support for such violence. Here, we report the results of a study that measures support for partisan violence with both abstract items (e.g., support for general partisan violence) and those focused on more specific acts (e.g., support for a partisan motivated shooting). We additionally examine intra-individual and intergroup correlates of support for partisan violence. Across three data collections (total N = 2003) and using seven unique operationalizations of support for partisan violence, we find the most consistent and typically largest relationships with individuals’ reported “need for chaos” (e.g., agreement with statements like: “Sometimes I just feel like destroying beautiful things”) and the extent to which they dehumanize members of the opposing party. System justification and social dominance orientation also consistently relate to support for partisan violence. These results suggest a common set of correlates for abstract and specific violence that reflect internal psychological (intra-individual) tendencies and intergroup evaluations. It also seems that partisan violence resembles other types of intergroup violence, as the relevant variables are not unique to the political domain.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2
Fig. 3
Fig. 4

Similar content being viewed by others

Data Availability

All data, analysis code, and study materials are available at


  1. Their initial item is a slight variant of this one, asking “What if [OUT-PARTY] win the 2020 presidential election? How much do you feel violence would be justified then?”.

  2. Westwood et al. (2022) also include a non-political condition to compare partisan with non-partisan violence.

  3. Westwood et al. (2022) also asked whether Wright should face criminal charges.

  4. They additionally ask about protesting without a permit, and vandalism. We included these items in our study, but do not report the results given they are not clear cases of violence (Westwood et al., 2022, p. 5).

  5. Petersen et al. (2023) link individuals’ need for chaos to their support for engaging in distinct violent acts to promote their “group’s” political rights and interests, where group is said to refer to political, religious, or social group.

  6. There are various other ideologies that one could study. For example, Armaly et al. (2022) report a moderated correlation with the ideology of Christian nationalism (moderated by perceived victimhood, white identity, and support for the Qanon movement). Piazza (2023a) finds a relationship with populism that he defines as an ideology (p. 3). As noted, the variables considered in our studies are not exhaustive, but rather exemplars of intra-individual and intergroup correlates of support for partisan violence.

  7. Notably, these authors do not invoke partisanship in their violence measures. Instead, they ask whether violence is an acceptable way to express disagreement and achieve important objectives.

  8. Piazza (2023b) finds a relationship between polarization and support for political violence (and replicates it with macro-level data). However, he employs a measure that is different from the most commonly used approach in the affective polarization literature. His measure asks whether the other party is a threat to the nation’s well-being. He notes that his construct/measure may involve the mechanism of dehumanization (as well as moralization and mobilization), a construct that we treat distinctly.

  9. Kalmoe and Mason (2022b) include an item about dehumanization as part of their moral disengagement battery.

  10. From a pure descriptive perspective, in our data reported below, the bi-variate correlations between support for partisan violence and these variables are moderate at best (e.g., averaging around .30 and only in one case exceeding .50).

  11. Study 1 ended up with 17 participants who identified as pure Independents (i.e., they did not lean toward one of the major parties); we exclude them from the analyses. Patterns of results remain unchanged with these participants included. Studies 2–3 samples were composed entirely of partisans or leaners.

  12. As mentioned, we embedded attention checks in all our studies. In study 1, we included two catch items (e.g., “To ensure data quality, please select option number 3”), and excluded any respondent who failed either of these catch items from our analyses. In study 2, we included the two catch items plus an additional catch item and a short mock vignette with an attention check question that asked about facts in the vignette. We excluded respondents who failed any of these attention checks from our analyses. In study 3, we included the two catch items and any respondent who failed either of them was immediately terminated from the survey.

  13. In their research, Kalmoe and Mason (2022b) used 4- or 5-point scales while Westwood et al. (2022) asked about support on a five-point scale and justification as a dichotomous item. Our own use of distinct scales for the Kalmoe and Mason and Westwood et al. items was largely inadvertent (the use of 100-point scales for SPV follows Mernyk et al., 2022) and should not influence the relevance of distinct correlates, which is our focus.

  14. See SI A for question wordings and SI C (Tables SI3–SI25) for the alphas and means and standard deviations for all variables.

  15. To re-scale items that were not measured on a 0 to 100 scale (those measured on a 0 to 100 scale did not need to be re-scaled), we subtracted one from the response (since all were on scales that started at 1) and then divided the score by the maximum number of scale points that could be moved (e.g., for a 1–7 scale, that is 6 scale points and thus, we divided the score by 6). We then multiplied this value by 100, such that the resulting score ranged from 0 to 100 and represented the percentage likelihood of supporting violence.

  16. We operationalize need for chaos by taking the average across items (see SI A). Arceneaux et al. (2021) suggest instead creating four groups: low chaos, rebuilders, medium chaos, and high chaos. When we take that approach, our results remain consistent with what we report (the categories demonstrated substantively very large effects in the anticipated directions).

  17. In SI C (Figures SI1–SI89), we plot (with scatterplots) the relationships between each correlate and each outcome variable, across studies. We also include the distributions of the variables in each plot.

  18. Republicans exhibit significantly higher system justification and social dominance scores, but interactions between partisanship and each fall short of significance.

  19. Kalmoe and Mason (2022b, p. 84–85) find evidence of sexism relating to SPV, which is consistent with our findings regarding system justification because those high in sexism also tend to be high system justifiers (e.g., Jost and Kay, 2005).

  20. In SI C (Tables SI27–SI30), we merge data sets across all three studies, and observe notable effects of need for chaos and dehumanization among available variables.

  21. Even more recent, Uscinski et al., (2021, p. 884) state, “Studies about support for political violence in the United States are few and far between…”.

  22. As noted, Piazza (2023b) reports a relationship with polarization but uses a distinct operationalization that may envelope dehumanization (which contrasts with typical measures of polarization that focus on dislike).

  23. In each study, need for chaos and dehumanization are significantly correlated, respectively, at .25, .39, and .23. These modest, but non-trivial, relationships suggest there may be a subgroup of particular relevance. Additionally, in additional analyses, we created quartile bins for need for chaos, dehumanization, system justification, and social dominance. We then replaced the continuous measures of these variables with the quartile bins and re-ran the regression analyses described in the main text. We find that, in nearly every case, the correlations are isolated to the top or top two quartiles. This suggests an acute focus on those above the median on these variables. The full results of these regression models are reported in SI C (Tables SI31–SI42).


Download references


We thank Isaias Ghezae for his invaluable help with the paper’s figures.


The study was funded by Northwestern University and Stanford University.

Author information

Authors and Affiliations


Corresponding author

Correspondence to James N. Druckman.

Ethics declarations

Competing interests

The authors have no competing interests to declare.

Ethical Approval

The study was deemed exempt by the Stanford University Institutional Review Board and performed in accordance with the American Political Science Association’s ethical standards.

Informed Consent

All study participants gave informed consent.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file1 (PDF 15600 kb)

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Landry, A.P., Druckman, J.N. & Willer, R. Need for Chaos and Dehumanization are Robustly Associated with Support for Partisan Violence, While Political Measures are Not. Polit Behav (2024).

Download citation

  • Accepted:

  • Published:

  • DOI: