Single-year change in views of democracy and society and support for political violence in the USA: findings from a 2023 nationally representative survey

Abstract

Background

A 2022 survey in the USA found concerningly high prevalences of support for and personal willingness to engage in political violence, of beliefs associated with such violence, and of belief that civil war was likely in the near future. It is important to determine the durability of those findings.

Methods

Wave 2 of a nationally representative cohort survey was conducted May 18-June 8, 2023; the sample comprised all respondents to 2022’s Wave 1. Outcomes are expressed as weighted proportions; changes from 2022 to 2023 are for respondents who participated in both surveys, based on aggregated individual change scores.

Results

The completion rate was 84.2%; there were 9385 respondents. After weighting, 50.7% (95% confidence interval (CI) 49.4%, 52.1%) were female; weighted mean (SD) age was 48.5 (25.9) years. About 1 in 20 respondents (5.7%, 95% CI 5.1%, 6.4%) agreed strongly/very strongly that “in the next few years, there will be civil war in the United States,” a 7.7% decrease.

In 2023, fewer respondents considered violence to be usually/always justified to advance at least 1 of 17 specific political objectives [25.3% (95% CI 24.7%, 26.5%), a 6.8% decrease]. However, more respondents thought it very/extremely likely that within the next few years, in a situation where they consider political violence justified, “I will be armed with a gun” [9.0% (95% CI 8.3%, 9.8%), a 2.2% increase] and “I will shoot someone with a gun” [1.8% (95% CI 1.4%, 2.2%), a 0.6% increase]. Among respondents who considered violence usually/always justified to advance at least 1 political objective, about 1 in 20 also thought it very/extremely likely that they would threaten someone with a gun (5.4%, 95% CI 4.0%, 7.0%) or shoot someone (5.7%, 95% CI 4.3%, 7.1%) to advance such an objective.

Conclusions

In this cohort, support for political violence declined from 2022 to 2023, but predictions of firearm use in political violence increased. These findings can help guide prevention efforts, which are urgently needed.

Background

In the USA, experts both inside and outside of government have documented an increase in incidents of political violence (the use of force or violence to advance political objectives) (Armed Conflict Location Event Data Project 2019) and have issued repeated warnings about the potential for such violence to disrupt democratic processes in America and jeopardize Americans’ health and safety (Kleinfeld 2021; Walter 2022; Kalmoe and Mason 2022; Armed Conflict Location Event Data Project 2022; Federal Bureau of Investigation and Department of Homeland Security 2022; Office of the Director of National Intelligence 2021; House Select Committee to Investigate the January 6th Attack on the United States Capitol 2022). Political violence is a public health problem.

In 2022, we conducted Wave 1 of a nationally representative cohort survey exploring Americans’ support for and willingness to engage in political violence and the prevalence of beliefs associated with such violence. The first report (Wintemute et al. 2024) from that survey presented concerning findings for the population as a whole, among them that nearly one-third of Americans (32.8%) considered violence usually or always justified to advance at least 1 of 17 specified political objectives and that 13.7% strongly or very strongly agreed with the statement that “in the next few years, there will be civil war in the United States.” Additional reports have examined variation in these and other measures with party affiliation and political ideology (Wintemute et al. 2022) and focused on particular populations of interest: MAGA Republicans (Wintemute et al. 2024); firearm owners (Wintemute et al. 2024); and those who hold racist beliefs, endorse the use of violence to effect social change, or approve of specific extremist organizations and movements that have been linked to violence (Wintemute et al. 2023).

This study presents initial findings from Wave 2 of the survey, conducted in mid-2023 among respondents to Wave 1. The focus of the analysis is on change from 2022 to 2023 in the findings presented in our first report (Wintemute et al. 2023), based on linked observations for 9385 participants who responded in both years. We also expand our findings on predicted firearm use in political violence, assessing differences between respondents who do and do not consider political violence justifiable.

Methods

Methods for Wave 2 of this cohort survey closely followed those for Wave 1 (Wintemute et al. 2023). Wave 2 was designed by the authors and administered online in English and Spanish from May 18 to June 8, 2023 by the survey research firm Ipsos (Ipsos 2021a). The study was reviewed by the University of California Davis Institutional Review Board (protocol 187125: exempt from full review, category 2, survey research). Before participants accessed the questionnaire, they were provided informed consent language that concluded, “[by] continuing, you are agreeing to participate in this study.” The study is reported following American Association for Public Opinion Research guidelines (American Association for Public Opinion Research 2021).

Participants

Participants for Wave 1 were drawn from the Ipsos KnowledgePanel, an online research panel that has been widely used in population-based research on violence and firearm ownership (Kravitz-Wirtz et al. 2021; Wintemute et al. 2022; Schleimer et al. 2020; Miller et al. 2022; Miller and Azrael 2022; Salhi et al. 2020). To establish a nationally representative panel, KnowledgePanel members are recruited on an ongoing basis through address-based probability sampling using data from the US Postal Service’s Delivery Sequence File (Ipsos. Knowledge 2021b). Recruitment into KnowledgePanel involves repeated contact attempts, if necessary, by mail and telephone. Recruited adults in households without internet access are provided a web-enabled device and free internet service, and a modest, primarily points-based incentive program seeks to encourage participation and promote participants’ retention in KnowledgePanel over time (Ipsos. Knowledge. 2021b).

A probability-proportional-to-size procedure was used to select a study-specific sample for Wave 1. All panel members who were aged 18 years and older were eligible for selection. Invitations were sent by e-mail; automatic reminders were delivered to non-respondents by e-mail and telephone beginning 3 days later (Ipsos 2021a).

The Wave 1 survey was conducted May 13 to June 2, 2022. It included a main sample, which had a completion rate of 53% and provided the study population for our initial report (Wintemute et al. 2023), and oversamples of firearm owners, transgender people, combat veterans, and California residents that were recruited to ensure adequate power for planned analyses. Compared with main sample nonrespondents, main sample respondents were older and more frequently white, non-Hispanic; were more often married; had higher education and income; and were less likely to be working (Wintemute et al. 2023).

Including the main sample and oversamples, Wave 1 comprised 12,947 respondents. Of those respondents, 11,140 (86.0%) remained active members of KnowledgePanel on Wave 2’s launch date and were invited to participate in Wave 2. (The 1807 Wave 1 respondents who were not active members of KnowledgePanel on Wave 2’s launch date had left the cohort through normal attrition.)

A final Wave 2 survey weight variable provided by Ipsos adjusted for the initial probability of selection into KnowledgePanel and for survey-specific nonresponse and over- or under-coverage using design weights with post-stratification raking ratio adjustments. As with the 2022 sample, the weighted 2023 sample is designed to be statistically representative of the noninstitutionalized adult population of the USA as reflected in the 2021 March supplement of the Current Population Survey (Ipsos. 2021a).

Measures

Sociodemographic data were collected by Ipsos from profiles created and maintained by KnowledgePanel members. Survey questions that supplied data for this analysis covered 3 broad domains: beliefs regarding democracy and the potential for violence and civil war in the USA, beliefs regarding American society and institutions, and support for and willingness to engage in political violence.

Our primary outcome measures again concerned political violence. Violence was represented by the phrase “force or violence,” defined in the questionnaire as “physical force strong enough that it could cause pain or injury to a person.” “Force or violence to advance an important political objective that you support” was used in questions about respondents’ support for and willingness to engage in political violence.

As in 2022, respondents were asked about the extent to which they considered political violence to be justified “in general” and then about justification for its use to advance specified political objectives. Example objectives include “to return Donald Trump to the presidency this year,” “to preserve an American way of life based on Western European traditions,” and “to stop police violence” (Tables 6 and 7). Responses for 17 objectives were collected in both years. In 2022, 9 of 17 were presented to all respondents and 8 were paired, with respondents randomized for each pair to see 1 item; each respondent was presented with 13 of 17 objectives. In 2023, all 17 items were presented to all respondents.

Respondents who considered political violence to be at least sometimes justified for at least 1 of these 17 objectives in 2023 were asked about their personal willingness to engage in political violence: by type of violence (to “damage property,” “threaten or intimidate a person,” “injure a person,” “kill a person”) and by target population (examples: “an elected federal or state government official,” “a police officer,” “a person who does not share your religion”) (Tables 8 and 9).

All respondents were asked about the likelihood of their future use of firearms in a situation where they consider political violence to be justified (examples: “I will be armed with a gun,” “I will shoot someone with a gun”) (Table 10).

The full text of all questions reported on here, including sources for questions from prior surveys by other investigators, is in the Supplement (see Additional file 1).

Implementation

Ipsos translated the questionnaire into Spanish, and interpreting services staff at UC Davis Medical Center reviewed the translation. Thirty-three KnowledgePanel members participated in a pretest of the English language version that was administered May 5–9, 2023.

Respondents were randomized 1:1 to receive response options in order from either negative to positive valence (example: from ‘do not agree’ to ‘strongly agree’) or the reverse throughout the questionnaire. Where a question presented multiple statements for respondents to consider, the order in which those statements were presented was randomized unless ordering was necessary. Logic-driving questions (those to which responses might invoke a skip pattern) included non-response prompts.

We employed unipolar response arrays without a neutral midpoint (e.g., do not agree, somewhat agree, strongly agree, very strongly agree). The literature is not in agreement on whether such midpoints should be included (Chyung et al. 2017; Westwood et al. 2022). We were persuaded by the studies reviewed by Chyung et al. (Chyung et al. 2017), which suggest that such midpoints allow respondents to choose “a minimally acceptable response as soon as it is found, instead of putting effort to find an optimal response,” a behavior known as satisficing. According to those authors, satisficing is particularly common when respondents are uncomfortable with the topics of the survey or under social desirability pressures, and both conditions apply here. Our analyses focus on responses above the “somewhat” or “sometimes” level to minimize the impact of potential satisficing on the results.

Statistical Analysis

To generate prevalence estimates, we calculated weighted percentages and 95% confidence intervals (CI) using PROC SURVEYMEANS in SAS version 9.4 (SAS Institute, Inc., Cary, NC) and Complex Samples Frequencies in IBM SPSS Statistics, version 29 (IBM Corp., Armonk, NY).

Each survey item was ordinal and was subject to non-response. We report weighted frequencies for each item for each possible response. In addition, we summarized each item’s non-missing responses for a given year by assigning integer values to ordinal levels to produce an item score and then averaging them.

To rigorously describe between-year changes in survey responses, we accounted for the longitudinal study design by computing within-individual change scores and then summarizing those. To compute differences in percentage choosing a particular response, we created indicator variables for each year for each item and each possible response and then computed the within-individual change score between the two survey years for each item and response level. To compute differences in mean response scores, we computed within-individual change scores for the item scores, restricted to the sample of respondents with non-missing responses to the item in both years. Between-year comparisons on whether respondents considered violence justified for at least 1 of the 17 specified political objectives were restricted for each respondent to the 13 items presented to that respondent in 2022.

Tables present findings for all respondents (main sample and oversamples) in 2022 and 2023, mean differences from 2022 to 2023 for each response option, mean item scores, and mean differences in item scores.

Results

Of 11,140 panel members invited to participate as part of the main study sample, 9385 completed the survey, yielding an 84.2% completion rate. The median survey completion time was 25 min (interquartile range, 18.6 min). Item non-response for items included in this analysis ranged from 0.5% to 4.0%; only 3 items had non-response percentages above 3.0% (see Supplement, Additional file 1).

After weighting, half of the respondents (50.7%, 95% CI 49.4%, 52.1%) were female; 62.7% (95% CI 61.2%, 64.1%) were white, non-Hispanic (Table 1). The weighted mean (SD) respondent age was 48.5 (25.9) years. Nonrespondents were younger than respondents (mean (SD) ages 52.5 (17.5) and 57.0 (16.5)) and less frequently male and white, non-Hispanic (Table S1).

Table 1 Personal characteristics of respondents

Democracy and the Potential for Violence

More than 60% of respondents in 2023 (62.3%, 95% CI 60.9%, 63.7%) perceived “a serious threat to our democracy,” and 84.6% (95% CI 83.4%, 85.7%) considered it very or extremely important “for the United States to remain a democracy”—decreases from 2022 of approximately 5% in both cases (Table 2). About 1 respondent in 6 (16.1%, 95% CI 15.0%, 17.1%) agreed strongly or very strongly that “having a strong leader for America is more important than having a democracy,” a 2.3% decrease from 2022. Strong or very strong agreement with the statement that “the 2020 election was stolen from Donald Trump, and Joe Biden is an illegitimate president” (16.7%, 95% CI 15.7%, 17.7%) did not change significantly from 2022 to 2023.

Table 2 Beliefs concerning democracy in the USA

There were decreases from 2022 to 2023 in the proportions of respondents agreeing strongly or very strongly with 2 of 3 statements about conditions in the USA justifying force or violence (Table 3): to “protect American democracy” if “elected leaders will not” [9.7% (95% CI 8.8%, 10.5%) in 2023, a decrease of 8.8%], and to save “our American way of life,” which is “disappearing” [12.1% (95% CI 11.1%, 13.0%) in 2023, a decrease of 3.6%].

Table 3 Beliefs concerning the potential need for violence in the USA

Strong or very strong agreement with the proposition that “in the next few years, there will be civil war in the United States” also declined [5.7% (95% CI 5.1%, 6.4%) in 2023, a 7.7% decrease] (Table 3).

American Society and Institutions

Four items explored beliefs on race and ethnicity and the great replacement assertion (Table 4), and 3 suggested an increased prevalence of racist beliefs. Strong or very strong agreement with the statement that “white people benefit from advantages in society that Black people do not have” decreased from 2022 to 2023 [35.6% (95% CI 34.2%, 36.9%) in 2023, a 3.9% decrease], as did strong or very strong agreement with the statement that “having more Black Americans, Latinos, and Asian Americans is good for the country” [39.6% (95% CI 38.2%, 40.9%) in 2023, a 5.0% decrease]. Disagreement with the assertion that “discrimination against whites is as big a problem as discrimination against Blacks and other minorities” fell slightly [48.1% (95% CI 46.7%, 49.5%) in 2023, a 2.2% decrease). An item addressing the “great replacement” belief (Table 4) and 3 items addressing the central elements of QAnon mythology and end-time Christianity (Table 5) showed little or no change.

Table 4 Beliefs concerning race and ethnicity and American society

Table 5 Beliefs concerning QAnon and biblical “end times”

Political Violence

The view that political violence is usually or always justified “in general” remained uncommon in 2023 (2.2%, 95% CI 1.7%, 2.7%) and did not change significantly from 2022 (Table 6). The proportion of respondents who considered violence to be usually or always justified to advance at least 1 political objective fell to 25.3% (95% CI 24.7%, 26.5%), a 6.8% decrease. Among those objectives considered individually (Tables 6 and 7), the proportion of respondents who considered violence to be usually or always justified decreased in 5 cases: “to preserve an American way of life I believe in,” “to oppose the government when it tries to take private land for public purposes,” “to stop voter intimidation,” “to stop police violence,” “to reinforce the police,” and “to keep borders open.” The largest decrease was for violence to reinforce the police [11.0% (95% CI 10.2%, 11.9%) in 2023, a 7.8% decrease]. There were no increases.

Table 6 Justification for political violence, in general and for 9 specific objectives

Table 7 Justification for political violence for 8 additional specific objectives*

The proportion of respondents who were not asked questions about their personal willingness to use force or violence rose by 7.8% (95% CI 6.3%, 9.2%) (Table 8); this reflects the increase in respondents who did not consider political violence to be at least sometimes justified for any of the 17 specified objectives. Among those asked, the proportions very or completely willing to use force or violence in 2023 remained low and without significant change from 2022: “to damage property,” 3.0% (95% CI 2.5%, 3.6%); “to threaten or intimidate a person,” 2.4% (95% CI 2.0%, 3.0%); “to injure a person,” 2.2% (95% CI 1.8%, 2.7%); “to kill a person,” 1.8% (95% CI 1.5%, 2.3%) (Fig. 1).

Table 8 Personal willingness to engage in political violence, by type of violence

Fig. 1

Mean difference in prevalence of willingness to engage in political violence and for firearm involvement. *Items 1–4: Personal willingness to use violence to achieve a political objective (very or completely willing). Items 5–8: Likelihood of using a gun in the future to achieve a political objective (very or extremely likely)

There were also no significant changes from 2022 to 2023 in the proportions of respondents willing to use force or violence against categories of people defined by their occupations, personal beliefs, or race and ethnicity (Table 9).

Table 9 Personal willingness to engage in political violence, by target of violence

However, all predictions that future use of a firearm “in a situation where you think force or violence is justified to advance an important political objective” increased in 2023 (Table 10; Fig. 1); 9.0% (95% CI 8.3%, 9.8%) thought it very or extremely likely that “I will be armed with a gun” (a 2.2% increase), 4.4% (95% CI 3.8%, 5.0%) that “I will carry a gun openly, so that people know I am armed” (a 0.9% increase), 1.6% (95% CI 1.2%, 2.1%) that “I will threaten someone with a gun” (a 0.6% increase), and 1.8% (95% CI 1.4%, 2.2%) that “I will shoot someone with a gun” (a 0.6% increase). Among respondents who considered violence usually or always justified to advance at least 1 specified political objective, about 1 in 20 also thought it very or extremely likely that they would threaten someone with a gun (5.4%, 95% CI 4.0%, 7.0%) or shoot someone (5.7%, 95% CI 4.3%, 7.1%) to advance such an objective (Table S2).

Table 10 Future likelihood of firearm possession and use in a situation where political violence is perceived as justified

Discussion

Many of the findings from this second wave of a nationally representative cohort survey in the USA can be viewed as improvements. There were decreases from 2022 to 2023 in perception of a “serious threat” to democracy in that country, in preference for a “strong leader” over a democracy, in expectation that civil war was imminent, in support for general statements of the potential need for violence to address social concerns, and in support for violence to advance specified political objectives. As in 2022, most respondents repeatedly rejected political violence, and most respondents who considered it justified were unwilling to participate in it themselves.

Perhaps criminal convictions and guilty pleas by hundreds of participants in the January 6, 2021 assault on the Capitol (Feuer et al. 2024)—a clear demonstration that such acts can have adverse consequences for the actors—helped decrease support for political violence. Other surveys from 2023 (States United Action 2023; Public Religion Research Institute 2023), which measured political violence differently, have found comparably low levels of support. In 1 case, 51% of respondents considered political violence to be “a major problem.” (States United Action 2023).

This good news comes with a caveat, however. National elections were held in 2022, but not in 2023. Support for political violence may vary with the election cycle, though not all studies have found this to be the case (States United Action 2023). If it does, support for political violence and other measures of polarization will likely increase in 2024.

Other differences in our findings are cause for concern. Support did not decrease for the “Big Lie” that the 2020 election was stolen and for the QAnon delusion. Endorsement of racist beliefs increased slightly. Although support for political violence decreased, willingness to engage in violence among the remaining supporters did not. Personal expectations of firearm use in political violence increased, and in 2023, 5.7% of respondents who believed such violence was usually or always justified to advance at least 1 political objective thought it very likely that they would shoot someone to achieve a political objective.

This good news notwithstanding, public safety, public health, and clinical health professionals will need to collaborate on efforts to prepare for and prevent violent events of a scale that could disrupt critical infrastructure and the 2024 elections and exceed the capabilities of many healthcare delivery systems.

More broadly, there is an urgent need for general public awareness of the threats posed by political violence and for the deployment of preventive interventions beyond those available only to law enforcement agencies. As a recent review of the research literature concluded, it will be important to focus on structural reform and behavior change; intervening on underlying attitudes and beliefs has disappointingly little effect (Kleinfeld 2023). Thoughtful lists of recommendations for policy and social change have been developed (Tisler and Norden 2024; Clapman 2024; Carey et al. 2023; Morales-Doyle et al. 2023). To these should be added, “if you see something, say something”; many prevention measures depend on critical information about threatened violence getting to those in a position to do something about the threat (National Counterterrorism Center 2021). In making that recommendation, we acknowledge that potential reporters may realistically fear that they will face arrest or retaliation, or that those whose conduct they are reporting will be harmed.

Limitations

Several technical limitations exist. The findings are subject to sampling error and nonresponse bias. Arguably, nonresponse was most important in Wave 1; the 84% response rate for Wave 2 was high. A few outcomes are uncommon, with response counts < 100 and weighted prevalences below 5%. The large study sample and small prevalence estimates result in relatively narrow confidence intervals in these cases, but the estimates remain vulnerable to bias from sources such as inattentive or strategic responses. Because 2022 results presented here are for all Wave 1 respondents who responded in Wave 2, and not just those who were included in the Wave 1 main sample, results for 2022 in this report do not replicate those in our initial study (Wintemute et al. 2023).

External events (or their absence) may have affected our findings. In 2022, widely publicized mass shootings occurred in Buffalo, NY and Uvalde, TX while the survey was in the field; there were no comparable events during the fielding of the 2023 survey. The Buffalo shooting is understood to have been a race-related hate crime motivated by great replacement thinking and may have affected respondents’ views on race, violence, and that particular belief. In 2023, the survey closed just before the federal criminal indictment of Donald Trump was handed down; support for violence to return him to the White House increased immediately thereafter (Pape 2023). In both years, Russia’s war against Ukraine may have influenced responses on violence and democracy.

Conclusions

Findings from this large, nationally representative cohort survey indicate that while support for political violence is common, it is susceptible to change. Planned additional analyses will seek to identify characteristics and life events associated with decreases (and increases) in support for political violence. Increases in expectations of firearm use in political violence are of particular concern. The findings of this analysis will be useful in designing prevention efforts.