America experienced a violent populist backlash against the results of the 2020 presidential election when an estimated 2,000 people stormed the US Capitol on January 6, 2021. They were seeking to overturn the election’s results and were supported by many thousands more who surrounded the Capitol but did not enter.
Scholarship on the causes and consequences of this paradigmatic event and its implications for understanding violent populism is still in its early stages. To date, studies have explored both “top-down” explanations, which emphasize the role of then-President Trump and political elites who supported the “Big Lie” that the election was stolen (Arceneaux and Truex Reference Arceneaux and Truex2022), and “bottom-up” explanations, which emphasize the importance of perceived victimhood, white identity, conspiratorial thinking, and other key factors (Armaly, Buckley, and Enders Reference Armaly, Buckley and Enders2022; Armaly and Enders Reference Armaly and Enders2022; Crothers and Burgener Reference Crothers and Burgener2021; Jardina and Mickey Reference Jardina and Mickey2022; Piazza and Van Doren Reference Piazza and Van Doren2022). An important gap in the existing scholarship is that it does not analyze the insurrectionists themselves, focusing instead on public support for the Capitol insurrection and political violence after the attack occurred. This study, by contrast, uses data on the charged insurrectionists to understand the local conditions that produced insurrectionists in the first place.
The insurrectionists traveled to Washington, DC, from communities across the country that varied in their support for Donald Trump. For example, many communities that might be expected to produce insurrectionists—counties with large populations of more than 350,000 and that voted for Trump in 2020 by more than 10 points, including Tulsa, Oklahoma; Waukesha, Wisconsin; and Kern, California—produced no insurrectionists. At the same time, some communities with small populations that voted for Joe Biden by more than 40 points—counties including San Miguel, Colorado; Santa Fe, New Mexico; and Berkshire, Massachusetts—all produced at least one.
Why did some communities produce more insurrectionists than others? What are the local political, economic, and social conditions of the communities that sent insurrectionists to the US Capitol in support of Donald Trump? What does the political geography of the insurrection reveal about the factors mobilizing violent support for Trump in 2021 and tell us about the risk factors for future violent populism in America?
To answer these questions, this article systematically analyzes the political geography of the January 6, 2021, US Capitol insurrection (hereafter Jan 6), using a new dataset of the counties of origin for individuals charged by the US Department of Justice in this event. We tested two prominent theories about demographic and political structures associated with local support for electoral populism and populist leaders like Donald Trump—demographic change (Jardina Reference Jardina2019; Mutz Reference Mutz2018) and manufacturing decline (Baccini and Weymouth Reference Baccini and Weymouth2021; Broz, Frieden, and Weymouth Reference Broz, Frieden and Weymouth2021)—and whether they also explain violent populism. The goal of our structural analysis was to identify the key local community factors that separate “insurrectionist-prone” communities from others, just as structural analyses of electoral populism identify the conditions of communities most likely to generate votes for populist candidates.
Our primary independent variables are the change in the percentage of the population that is non-Hispanic white between 2010 and 2020 and the change in manufacturing share of employment between 1970 and 2020, both measured at the county level. County-level analysis is used widely in studies of community factors associated with electoral populism (Broz, Frieden, and Weymouth Reference Broz, Frieden and Weymouth2021) and political violence (Nemeth and Hansen Reference Nemeth and Hansen2021). Counties change less frequently than other political units, which is important for measuring the demographic and economic changes over decades at the core of this study.
Overall, our findings suggest that white population decline is more important than manufacturing decline as a predictor of violent populism and that counties that voted for Trump were less likely to fight for Trump. Moreover, the effect of white population decline was magnified by certain political conditions.
Overall, our findings suggest that white population decline is more important than manufacturing decline as a predictor of violent populism and that counties that voted for Trump were less likely to fight for Trump.
First, the decline in a county’s proportion of non-Hispanic white population strongly accounts for a county’s rate of insurrectionists, whereas economic conditions matter less and populist support for Trump runs in the opposite direction. Specifically, the impact of non-Hispanic white population decline was three times greater than manufacturing decline and at a higher level of statistical significance. These results held when we controlled for the effects of population size and the distance between each county and the Capitol. Furthermore, our results are robust to the exclusion of counties at the extreme ends of the population scale.
Second, we evaluated how local political conditions interacted with white population decline on insurrection proneness of counties. We found that counties represented by a member of Congress who objected to the certification of the 2020 election were associated with higher rates of insurrectionists as their counties became less white. This suggests that inter-Republican efforts to outbid their rivals through demonstrated support for Trump could magnify the effect of white population decline. Other political circumstances, including close elections and a lack of Republican representation at the local level, did not magnify the effects.
Our findings should give us pause in presuming that economic solutions to electoral populism will axiomatically ameliorate violent populism. Although economic factors do play a role, social factors are stronger, which suggests the crucial need to confront the fear among certain whites about becoming a minority and their associated status anxiety.
To be clear, our claim is not that our analysis shows that factors such as local relative white population decline cause status anxiety to produce differential insurrectionist rates. Rather, our analysis reveals evidence consistent with such an effect. It is our hope that future research will build on these results to further investigate the differences among causes of electoral versus violent populism.
STRUCTURAL PREDICTORS OF VIOLENT POPULISM IN AMERICA
Jan 6 is an example of “violent populism,” a term we coined to capture support for violence to install or maintain a populist party or leader in power. As such, it is a form of populism, commonly understood as a political movement or party emphasizing an “us-versus-them” worldview in which the “us” refers to “the people,” who are engaged in a zero-sum battle with “them,” who they perceive to be represented by powerful and corrupt elites (Berman Reference Berman2021, 72–72; Mudde and Kaltwasser Reference Mudde and Kaltwasser2017, 6). Populist goals can be achieved through either the electoral process or extra-democratic means, including violence.
Structural conditions associated with local communities have been linked to support for political violence and electoral populism. Research on political violence, including terrorism and riots, has found that violence can be predicted by the characteristics of the area where perpetrators live (Nemeth and Hansen Reference Nemeth and Hansen2021; Spilerman Reference Spilerman1970). Structural factors also drive electoral populism. As Broz, Frieden, and Weymouth (Reference Broz, Frieden and Weymouth2021, 465) demonstrated, “there are strong geographic patterns to the populist backlash, and political choices are powerfully affected by local socioeconomic conditions.”
The literature on right-wing electoral populism in America during the “Age of Trump” is divided into two competing theories about why people voted for Trump. One theory focuses on social grievances, particularly grievances among white Americans who fear the consequences of their declining power and group status (Mutz Reference Mutz2018). The other theory focuses on economic grievances, such as the steady decline in high-paying manufacturing jobs since the 1970s and the consequences of that for American “rustbelt” communities (Broz, Frieden, and Weymouth Reference Broz, Frieden and Weymouth2021). It is interesting that both theories have corollaries within the literature on political violence, with some scholars showing how racial grievances easily can escalate to violence and other scholars showing that economic grievances also can create a society primed to respond to government failures with violence.
White Population Decline, Status Anxiety, and Support for Violence
In the United States, white population decline—meaning that the percentage of the US population that is white is declining and the percentage of the US population composed of minority groups is growing—is associated with whites perceiving their status as the dominant racial group to be threatened. Past studies have shown how exposing white Americans to the prospect of future white-minority status is associated with feelings of threat, fear, and anger toward minorities (Outten et al. Reference Outten, Lee, Costa-Lopes, Schmitt and Vala2018), as well as support for tougher immigration laws (Craig and Richeson Reference Craig and Richeson2014), and even right-wing extremist groups (Bai and Federico Reference Bai and Federico2021). This status anxiety is exacerbated by political rhetoric that exploits these fears for political gain (Danbold and Huo Reference Danbold and Huo2015; Major, Blodorn, and Blascovich Reference Major, Blodorn and Blascovich2018; Mutz Reference Mutz2018; Oberhauser, Krier, and Kusow Reference Oberhauser, Krier and Kusow2019; Young Reference Young2013). Indeed, the perception of white population decline, when combined with the inflammatory rhetoric used by Trump and other right-wing personalities, has been found to increase status anxiety among whites (Feola Reference Feola2022; Jardina Reference Jardina2019; Newman et al. Reference Newman, Merolla, Shah, Lemi, Collingwood and Ramakrishnan2021).
Racial and ethnic group demographic change has long been implicated as a cause of political violence. Studies of ethnic violence, for example, find that violent conflict can occur when political power in a society is organized along ethnic or racial lines and a dominant group experiences numerical decline (Horowitz Reference Horowitz1985). Group-position theory and the related racial-threat theory suggest similar dynamics (for a review, see Craig, Rucker, and Richeson Reference Craig, Rucker and Richeson2018). For example, Blumer (Reference Blumer1958) hypothesized that when dominant groups experience a numerical decline, the salience of in-group identity increases and members are more likely to perceive the corresponding outgroup as threatening their interests (Jardina Reference Jardina2019).Footnote 1 When a dominant racial group becomes “deeply concerned with its position vis-à-vis the subordinate group,” the subordinate group is perceived as a direct threat to the dominant group’s power and privilege (Blumer Reference Blumer1958, 4). The dominant group thus is more likely to have a negative attitude toward the outgroup (Jardina Reference Jardina2019; Outten et al. Reference Outten, Schmitt, Miller and Garcia2012; Schlueter and Scheepers Reference Schlueter and Scheepers2010) and to support the status quo or past hierarchical political and social arrangements (Mutz Reference Mutz2018). Once perceptions of threat set in, members of the group are more likely to support and engage in behaviors—including violence—to protect the dominant group’s status against the encroaching subordinate group.
In these circumstances, status competition can increase politically motivated violence by creating incentives and emotions that may overwhelm normal social norms against violence. First, the perception of group decline can encourage violence intended to suppress or destroy the political power of the rising outgroup because “to lose out in competition and comparison to others who are differentiated on a birth basis is to be afflicted with an apparent permanent disability” (Horowitz Reference Horowitz1985, 147). Second, political leaders can be incentivized to exploit ethnic fears to mobilize support among ethnic constituents and increase their political power (Kaufman Reference Kaufman2015; Levy and Myers Reference Levy and Myers2021). Observable demographic changes in a community enhance the persuasiveness of demographic threat narratives and increases support for policies to address the threat, including violence (Fischer and O’Mara Reference Fischer and O’Mara2023).
In the United States today, “these threats, both real and perceived, [have led] a sizeable proportion of whites to believe that their racial group, and the benefits that group enjoys, are endangered” (Jardina Reference Jardina2019, 3–4). As an extensive literature shows, white status anxiety contributes to support for right-wing electoral populism in the US context (Engler and Weisstanner Reference Engler, Weisstanner, Careja, Emmenegger and Giger2020; Mason, Wronski, and Kane Reference Mason, Wronski and Kane2021; Mutz Reference Mutz2018; Norris and Inglehart Reference Norris and Inglehart2019). White status anxiety can be the product of perceptions by whites of threat from rising minority groups (Danbold and Huo Reference Danbold and Huo2015; Kinder and Sears Reference Kinder and Sears1981), causing greater identification with the Republican Party (Craig and Richeson Reference Craig and Richeson2014; Major, Blodorn, and Blascovich Reference Major, Blodorn and Blascovich2018; Sides, Tesler, and Vavreck Reference Sides, Tesler and Vavreck2019) and greater implicit and explicit racial bias (Craig and Richeson Reference Craig and Richeson2014).
Accordingly, white population decline within communities can exacerbate perceived threats to white group status, leading some whites within that community to support violent political action as an ameliorative to their fears. The theory is straightforward: relative white demographic decline increases racial status anxiety among whites who observe that change. This makes candidates promising to secure white status (whether explicitly or via coded “dog whistles”) more appealing and increases the likelihood of support for violence to ensure that these candidates take and retain power. When this theory is applied to Jan 6, counties experiencing greater levels of white population decline should have produced more individuals mobilized by white status anxiety to use extra-democratic and even violent means to retain Trump in power because Trump was widely perceived as defending the status of whites (Kydd Reference Kydd2021). If county decline in the white population share is associated with the number of individuals from that county who were arrested for actions taken on Jan 6, it would be consistent with the white-status-anxiety mechanism.
Manufacturing Decline, Economic Insecurity, and Support for Violence
Economic factors also are associated with support for populism and political violence in the United States. For example, differential economic growth and the decline of once-important economic sectors can lead to loss of personal welfare and support for political parties—and even extremism to achieve major political changes that would reduce economic insecurity. Scholars have found that the rustbelts of the Northeast and the Midwest played an important role in Trump’s election in 2016, which suggests that economic insecurity among blue-collar workers was largely responsible for Trump’s populist support. That is, voters were mobilized specifically by Trump’s “America First” and protectionist policies and not simply his status as the Republican candidate (Broz, Frieden, and Weymouth Reference Broz, Frieden and Weymouth2021; McQuarrie Reference McQuarrie2017).
Economic insecurity also has long been viewed as a cause of voting and violence, based on the operative assumption that these behaviors can maximize individual welfare under certain conditions. As wages in a community go down—whether due to loss of crops from global warming, political fractures in disintegrating or failed states, or technological change leading to the loss of manufacturing jobs—the loss of traditional sources of income leads some individuals to support anti-establishment political parties and political violence (Häusermann, Kemmerling, and Rueda Reference Häusermann, Kemmerling and Rueda2020). Indeed, rational-choice scholars of civil war have built on the economic model, contending that rebels will engage in political violence if this behavior is the best way for them to maximize their personal welfare—most often when they are from poor, rural communities (Castrovillari, Mineyama, and Leepipatpiboon Reference Castrovillari, Mineyama and Leepipatpiboon2023; Collier and Hoeffler Reference Collier and Hoeffler1998).
Scholarship provides evidence that long-term economic decline, particularly in the manufacturing sector, affects communities in ways that correspond to electoral support for far-right populist candidates in the United States (Baccini and Weymouth Reference Baccini and Weymouth2021; Broz, Frieden, and Weymouth Reference Broz, Frieden and Weymouth2021; Rodrik Reference Rodrik2018). As with white status anxiety due to white population decline, economic insecurity due to loss of jobs and a low standard of living are associated with local communities because it is the community and the local economy that provide economic opportunities—especially for already-low-income individuals who have few resources to move to more productive areas. Faced with the consequences of local economic decline, individuals experiencing deep anxiety about their economic future may turn to populist political candidates who promise to reverse this decline and restore an often-romanticized era marked by high wages for now unemployed or underemployed individuals.
In America today, studies comparing the 2012 and 2016 elections consistently find that Donald Trump made above-average vote gains in white, rural communities with older blue-collar workers experiencing a decline in manufacturing employment (Broz, Frieden, and Weymouth Reference Broz, Frieden and Weymouth2021; Scala and Johnson Reference Scala and Johnson2017). To be clear, the key trends affect communities: “jobs and income decline, property values fall, the local tax base erodes…after a couple of decades, the [county] is reeling from waves of economic and social shocks, affecting everything” (Broz, Frieden, and Weymouth Reference Broz, Frieden and Weymouth2021, 465).
The theory is straightforward: long-term manufacturing job loss in a community makes it more likely that some people within that county will face economic insecurity and become more likely to not only vote for populist political candidates but also to become violent supporters of them. In the case of Jan 6, the greater the local decline in manufacturing and associated loss in manufacturing income, the deeper is the commitment to Trump as the most likely leader to bring prosperity back to their communities and the greater was the willingness to use extra-democratic and even violent means to retain him in power will have been. If counties with more manufacturing decline were associated with higher rates of arrested insurrectionists, it would be consistent with the economic mechanism.
We now assess the relationship between the structural factors of (1) white population decline and (2) manufacturing decline, as well as (3) populist support for Trump in 2020 (which is included to distinguish the main independent variables from pro-Trump voting behavior) on increased rates of participation on Jan 6 across counties. Before proceeding, we emphasize that our empirical analysis does not seek to show that these structural factors have a causal effect on differential insurrectionist rates. This is because the path from community-level factors to aggregate political outcomes likely runs through multiple intermediate mechanisms for which comparable data at the county-level are not generally available. As a result, our analysis reveals evidence consistent with the effect of county-level demographic and economic conditions (and their associated mechanisms) on insurrection proneness; it is not evidence of the effect itself.
THE POLITICAL GEOGRAPHY OF JANUARY 6: A NEW DATASET
Our dependent variable is the county count of individuals charged for their role in the Jan 6 insurrection. An estimated 2,000 people illegally entered the Capitol (Reilly Reference Reilly2021) and, as of April 2023, 951 people had been charged for their involvement in the attack. Using court filings and other public records, we constructed a database of all 951 charged insurrectionists and their home counties aggregated to the county (Pape, Larson, and Ruby Reference Pape, Larson and Ruby2024). The charged insurrectionists are overwhelmingly white (93%) and male (85%), with an average age of 41 (see also Pape Reference Pape2022).
Figure 1 displays the geographic distribution of insurrectionists by county. Geographically, 458 of 3,141 counties (15%) had at least one insurrectionist, representing 46 states (including Alaska and Hawaii) and the District of Columbia.
Figure 1 Geographic Distribution of Charged Insurrectionists by County of Residence
Independent Variables
For our measure of white population decline, we calculated the difference in the percentage of each county’s non-Hispanic white population between 2010 and 2020 using the Decennial Census.Footnote 2 Similar measures have been used to study the community-level effects of white status anxiety in the past, including an examination of how anxiety related to local white demographic decline drives increased mortality rates among middle-aged whites (Young Reference Young2016).
For manufacturing employment decline, we calculated the difference in county manufacturing job share between 1970 and 2020. This measure was adapted from Broz, Frieden, and Weymouth (Reference Broz, Frieden and Weymouth2021), which we updated to 2020. Additionally, to mitigate the problem of missing data, we imputed this information for the 473 counties where this information was not available using the data for other counties in the same state.
Vote share for Trump in the 2020 presidential election does not capture a county’s populist support for Trump because it also includes those Republicans who voted for Trump not because they supported his populist policies but simply because he was the only Republican on the ballot. To capture populist Trump support, we followed Broz, Frieden, and Weymouth (Reference Broz, Frieden and Weymouth2021) and included the difference in the percentage of the vote that Trump received in 2020 compared to the county vote share that establishment Republican candidate Mitt Romney received in 2012. This assumes that a higher proportion favoring Trump proxies the strength of Trump’s populist base in each county.
Table 1 compares counties in three categories: counties that had at least one insurrectionist, counties with any populist support for Trump (i.e., counties that voted for Trump in 2020 at a higher rate than they had voted for Romney in 2012), and counties overall. The data show that counties with at least one charged insurrectionist were between two and four times more likely to be medium or large metropolitan areasFootnote 3 and with a population between five and eight times larger than either populist pro-Trump counties or counties overall. They also were marginally closer to Washington, DC, the location of the insurrection.
Table 1 Counties That Fought for Trump versus Counties That Voted for Trump
Turning to our key variables of interest, counties with at least one insurrectionist had higher rates of white population decline between 2010 and 2020, averaging 6% compared to 4% in counties that voted for Trump. They also had higher rates of manufacturing employment decline. Accordingly, there are valid reasons to investigate the role of these variables as drivers of insurrectionism.
Control Variables
We include the following covariates in our analysis to account for the most likely alternative explanations of the dependent variable. The first control is the distance from the county centroid to Washington, DC. One of the counties that was most overrepresented in our data is Washington, DC, which is not surprising: people who live in Washington, DC, incurred no travel costs to participate beyond what they already would have for typical everyday tasks. By contrast, coming to Washington, DC, from the West Coast or the Midwest would require significantly greater costs, in both time and money. To control for the difficulty of travel, we included a simple distance to Washington, DC, variable.
The second control is population size, which is commonly controlled for in county-level analyses because there are many small counties with populations in the low thousands and relatively few enormous counties with populations in the millions. We expect that small counties, with populations of a few thousand, would produce zero insurrectionists on average, whereas large counties with populations of a few million would produce more, based simply on the number of people in a given county. Accordingly, the single-most important predictor of insurrectionist participation is expected to be population. To account for the effect of population size, we included county population as a control in all models.
To assess the robustness to concerns that our findings are unduly influenced by the wide variation in county population size in general and population outliers such as Orange County and Los Angeles County in particular, we also run the primary model on three subsets of counties: (1) the 420 counties between 0 and 1 standard deviation of mean county population; (2) the 1,562 counties above the median for county population; and (3) the 3,071 counties below 3 standard deviations of the population mean.
Finally, we also included covariates for whether a county is urban (vs rural) and the share of county population that is non-Hispanic white in 2020.
MODELS AND RESULTS
We analyzed the relationship among the count of charged insurrectionists for a given county, our dependent variable, and independent variables of interest using negative binomial regression appropriate for modeling counts. The negative binomial is preferable to Poisson regression when the variance of the dependent variable is greater than the mean (Hilbe Reference Hilbe2007), as it was in our case (i.e., variance=1.17, mean=0.30). We conducted two analyses. First, we tested whether white population decline or manufacturing decline predicted county-level insurrectionists. Second, we analyzed the potential moderating effect of county-level political circumstances.
White Population Decline versus Manufacturing Loss
To assess whether white population decline or manufacturing loss predicted county-level insurrectionists, we ran seven separate models, examining the effect of each of our independent variables of interest independently (models 1–3), all in the same model (model 4), and restricted by population (models 5–7).
Table 2 presents the results as incidence rate ratios (IRRs) for ease of interpretation. IRRs represent the relative difference in the rate of charged insurrectionists attributable to a one-unit increase of a given factor, controlling for other factors in the model. An IRR of 1 means no effect, greater than 1 means a positive effect, and between 0 and 1 means a negative effect. All covariates with the exception of the binary indicator for Urban Center have been standardized (mean = 0, standard deviation = 1) to simplify interpretation as the effect of a one standard deviation change in the covariate on the dependent variable.
Table 2 Modeling Insurrection Propensity of Counties
Notes: Results from negative binomial regression with robust standard errors in parentheses and coefficients reported as IRRs where values greater than 1 indicate an increased rate, less than 1 indicate a decreased rate, and 1 indicates no difference/effect. All non-binary covariates are standardized. Models 3–7 omit Alaska because Alaska does not report vote totals by county. *p<0.05, **p<0.01, ***p<0.001.
Overall, our analysis finds that white population decline was a strong predictor of insurrectionist-prone counties, whereas manufacturing employment decline mattered far less and populist support for Trump ran in the opposite direction.Footnote 4 The results are statistically significant controlling for county population, white share of population, distance to Washington, DC, and being an urban versus rural county.Footnote 5
First, although both increased white population decline and manufacturing decline were associated with an increased number of insurrectionists, the impact of local white population decline was significantly greater than local manufacturing employment decline across all seven model specifications. Using full model 4 as a reference because it includes both factors, for every 1-standard-deviation decline in a county’s white population share, the rate of insurrectionists from that county was expected to increase by 37% versus 12% for a 1-standard-deviation decline in manufacturing employment. It is important to note that the effect of white population decline was robust even when we restricted the county sample to adjust for the wide disparities in county population (see models 5–7).Footnote 6
Figure 2 visualizes the relationship across the range of the standardized independent variables (i.e., ±3 standard deviations) on predicted county count of insurrectionists holding all other variables in the model at their mean. Panels (a), (b), and (c) in figure 2 compare results from the models with each factor alone (see models 1–3) to the full model (see model 4) with all included.
Figure 2 Effect on Predicted County Count of Insurrectionists
Notes: This figure presents the predicted count of insurrectionists from models 1–4, holding all other covariates at their mean. The distribution of each standardized independent variable is indicated by the underlying histogram (right y-axis).
When we compare figures 2(a) and 2(b), the positive relationship between white population decline and manufacturing loss and the predicted count of insurrectionists is clear, as is the steeper slope for white population decline. Counties that are 3 standard deviations above the mean for white population decline (i.e., an approximate 13% decline in white population share between 2010 and 2020) had an expected count of insurrectionists of approximately 0.4, compared to an expected count of 0.2 for 3 standard deviations of manufacturing employment decline (i.e., an approximate 38% decline in manufacturing employment share in 50 years).Footnote 7 In each case, the results were almost identical for the individual and full models.
Third, as shown in figure 2(c), the higher populist support for Trump, the lower the rate at which counties are expected to produce insurrectionists. We also examine general support for Trump by substituting Trump’s county vote share in 2020 for the populist support measure and finds similar effects (see model 1, table A2, in the online appendix). This evidence indicates that counties that voted for Trump were less likely to fight for Trump.
The Impact of Political Circumstances
Did the effect of white population decline on the insurrection-proneness of counties depend on local political conditions? Scholars have shown that electoral competition, elite out-bidding, and voter frustration matter for political violence and contentious politics in other contexts. There are valid reasons to believe that these logics may amplify the effect of white population decline.
Electoral Competition
Scholars have found that close electoral races can create the conditions in which political violence occurs (Wilkinson Reference Wilkinson2004). Thus, the effect of localized demographic change on participation in the Jan 6 insurrection may have been accentuated in counties with close electoral races. In these locations, Republican politicians—incentivized to draw on grievances such as white status anxiety made salient by demographic change—may have sought to mobilize Trump’s base for their electoral advantage. This would have increased the salience of white population decline as a political grievance while simultaneously deepening support for Trump, whose candidacy was linked to defending white interests.
Elite Outbidding
Elite outbidding—i.e., spirals of competition between elites from the same party or ideology—has also been found to exacerbate political extremism and violence in important contexts (Bloom Reference Bloom2004). Accordingly, white population decline may have had greater effects on participation in the Jan 6 insurrection in counties where Republicans are politically dominant. In the absence of meaningful Democratic competition, the outcome of political races is determined in primary contests between two or more Republican candidates. Under such conditions, candidates are incentivized to adopt extreme positions to outbid Republican competitors. These positions may have included those related to white status anxiety and manifested as extreme support for Trump—for example, actively promoting the conspiracy theory that Democrats stole the 2020 election and objecting to the certification of Joe Biden as president—as did at least 139 Republican members of the House of Representatives. Indeed, a New York Times analysis found that 10 of the 12 counties in which whites became the minority during the past 30 years were represented in the House by an objector (Keller and Kirkpatrick Reference Keller and Kirkpatrick2022).
Conservative Voter Frustration
Third, a lack of Republican representation at the local level could have led to a greater likelihood of participation in the Jan 6 insurrection out of frustration. Citizens of a county under solid Democratic control might have felt especially frustrated by areas of political concern—including demographic change—that they could not address through local electoral means.
To evaluate these three scenarios, we replicated our previous analysis for white population decline (see model 1) but with an interaction term between white population decline and political competition, outbidding, or frustration variables in order to capture whether the effect of white population decline was conditional on these political circumstances. For electoral competition, we used a binary to indicate counties for which the margin of victory in the 2020 House congressional vote was within 5%. For elite outbidding, we used a binary that indicated counties that were represented by members of Congress who refused to ratify the 2020 election results. For voter frustration, we used a binary variable that captured counties that voted for Biden by a 20% margin or more. Table 3 presents results of the analysis.
Table 3 Effect of Political Circumstances and White Population Decline
Notes: These models are based on model 1. All continuous covariates are standardized. Negative binomial regression with robust standard errors, controlling for % non-Hispanic white, urban center, distance from Washington, DC, and county population. Reporting IRRs where values greater than 1 indicate a higher number of predicted insurrectionists, less than 1 indicates fewer, and 1 indicates no effect. Robust standard errors in parentheses. *p<0.05, **p<0.01, ***p<0.001.
Of the three political conditions, the only one supported by the evidence was elite outbidding: counties represented by one of the 139 Republican House Representatives who objected to the 2020 election results had a higher predicted rate of insurrectionists as their counties became less white. In locations with higher white population decline, having a local political elite refuse to certify the 2020 presidential election was an additional catalyst for the Jan 6 insurrection (see model 9). By contrast, contested counties do not appear to have produced insurrectionists at higher rates as the white population declined (see model 8). The IRR for frustration (see model 10) was statistically significant but negative, indicating that solidly Democratic counties in which Republicans are unlikely to achieve political power produced insurrectionists at a lower rate than other counties with higher levels of white population decline.
Of the three political conditions, the only one supported by the evidence was elite outbidding: counties represented by one of the 139 Republican House Representatives who objected to the 2020 election results had a higher predicted rate of insurrectionists as their counties became less white. In locations with higher white population decline, having a local political elite refuse to certify the 2020 presidential election was an additional catalyst for the Jan 6 insurrection.
It is important to note that these three political conditions do not appear to have had much effect in determining which counties were more insurrectionist prone on their own. It is also notable that the effect of white population decline remains significant when all three political factors are absent (at 0). This is evidence that the effect of white population decline on insurrection proneness is not explained by the three political conditions but rather more likely creates the incentive to exploit it.Footnote 8
CONCLUSIONS
Our analysis finds that a demographically declining white majority is a stronger contributor than manufacturing decline in local support for violent populism, at least in the case of the Jan 6 insurrection. We further find that the effect of white population decline is greater in counties whose US House Representative rejected the results of the 2020 election. Although it is not direct evidence that the mechanism of white status anxiety matters more than economic anxiety in mobilizing violent populism, these findings are consistent with such an effect.
Our findings contribute to existing scholarship on white status anxiety and its role in the Jan 6 insurrection and anti-democratic political violence (Armaly, Buckley, and Enders Reference Armaly, Buckley and Enders2022; Armaly and Enders Reference Armaly and Enders2022; Jardina and Mickey Reference Jardina and Mickey2022; Piazza and Van Doren Reference Piazza and Van Doren2022). Our approach of focusing on the political geography of the charged Jan 6 insurrectionists provides new evidence of the effects of white demographic change on a concrete instance of violent populism. Furthermore, the finding that communities with a House Representative who refused to certify the 2020 election were more likely to produce insurrectionists in the context of white population decline suggests that elites are important moderators of violent populism. Similar to Kalmoe and Mason (Reference Kalmoe and Mason2022, 136), our findings show that white population decline can provoke “violent reactions among partisans even in the absence of enflaming leadership.”
One limitation of our analysis is its inability to test mechanisms directly because it does not measure white status or economic anxiety directly. Another limitation is that county-level heterogeneity—particularly large variation in population size— impedes our ability to make inferences about mechanisms from observed effects. However, our analysis seeks to minimize this problem through robustness checks that focus on similarly sized counties. A future project could examine the relationship between white population decline and status anxiety by conducting surveys across comparable geographic units, which then could test the mechanism directly.
To further understand the relationship between demographic and economic changes on white status anxiety and other sociopolitical concerns about the risk of violent populism, future research should examine the scope and determinants of violent populist sentiments in nationally representative surveys; expand the range of violent populist outcomes to include risks of lethal attacks against minorities; examine the role of social networks in mobilization; and investigate the causal relationship among extremist rhetoric by militias, political leaders, and prominent media figures in amplifying white-status concerns and promoting violent populist outcomes.
Although more research is necessary to understand the specific causal processes connecting white status anxiety to violent populism, especially at the individual level, two implications about violent populism follow from our analysis. First, our results suggest that solutions to the problem of violent populism that focus solely on economic issues such as manufacturing decline may not be successful, at least in the American context. To diminish the problem of violent populism, we must confront the fear among certain whites of becoming a minority and their associated status anxiety, as well as the role that our politicians and media figures play in activating and exploiting this fear. The need to confront these fears and temper associated violent behaviors is especially great in areas where the white proportion of the local population is declining the most.
…our results suggest that solutions to the problem of violent populism in America that focus solely on economic issues such as manufacturing decline may not be successful. To diminish the problem of violent populism in America, we must confront the fear among certain whites of becoming a minority and their associated status anxiety, as well as the role that our politicians and media figures play in activating and exploiting this fear. The need to confront these fears and temper associated violent behaviors is especially great in areas where the white share of local population is declining the most.
Second, our findings increase concerns about the risks of violent populism in the United States and perhaps other liberal democracies around the world. For many years, scholars have shown that concern among dominant groups that they are losing privileged social and political status is playing at least a partial role for political support of populist and extremist candidates in the United States and Europe (Engler and Weisstanner Reference Engler, Weisstanner, Careja, Emmenegger and Giger2020, Reference Engler and Weisstanner2021). These same factors may portend the emergence of violent populism as well.
ACKNOWLEDGMENTS
The authors gratefully acknowledge funding provided by the University of Chicago, the Pritzker Military Foundation on behalf of the Pritzker Military Museum & Library, and the Hopewell Fund. We also thank the Workshop on International Politics at the University of Chicago and the anonymous reviewers for helpful comments.
DATA AVAILABILITY STATEMENT
Research documentation and data that support the findings of this study are openly available at the PS: Political Science & Politics Harvard Dataverse at https://doi.org/10.7910/DVN/KOOIRH.
CONFLICTS OF INTEREST
The authors declare that there are no ethical issues or conflicts of interest in this research.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit http://doi.org/10.1017/S1049096524000040.
Open access
