Historical Mob Violence and the 2016 Presidential Election

Prior Literature

Lynching as a practice was geographically concentrated in southern states (Tolnay and Beck 1995), only one in an array of violent tactics used to oppress Black Americans (Campney 2019). Wide variation in levels of mob violence existed across the South, however, shaped by local social, demographic, and economic factors. Importantly for the current study, partisan voting patterns were linked to whether communities embraced or resisted lethal violence as a mechanism for maintaining white supremacy. Counties with greater support for the segregationist Democratic Party during the Jim Crow era experienced more lynchings, on average (Tolnay and Beck 1995) and were more likely to allow threatened mob violence to become deadly (Hagen, Makovi, and Bearman 2013).

A community’s history of racist violence persists in indicators of intergroup conflict and routinized bigotry. Historical lynchings predict elevated rates of homicide and interracial murder committed by whites (Messner, Baller, and Zevenbergen 2005) and lower probability of law enforcement compliance with federal hate crimes reporting (King, Messner, and Baller 2009). Public schools in these counties inflict more corporal punishment today (Ward et al. 2019) and have higher concentrations of white supremacist organizations (Durso and Jacobs 2013). Historical lynching “hot spots” experienced prodigious establishment of private “segregationist academies” in the mid-twentieth century (Porter, Howell, and Hempel 2014) and more Black church burnings toward its end (Howell et al. 2018).

The link between politics and white racial anxieties is clear (Knowles et al. 2013; Kuziemko and Washington 2018; N. Valentino and Sears 2005). ³ A wide array of scholarship across the social sciences identifies individual characteristics of voters supporting Trump’s candidacy (Hooghe and Dassonneville 2018; Major, Blodorn, and Blascovich 2018; Whitehead, Perry, and Baker 2018), including white voters who felt their control of political structures waning in the face of a growing nonwhite population (McVeigh and Estep 2019). Our study contributes to scholarship exploring the connection between electoral support for Trump’s candidacy and local context. For example, Trump performed well in communities with declining life expectancy (Bor 2017) and low economic mobility (Rothwell and Diego-Rosell 2016).

Extant literature from across the social sciences finds enduring associations between historical racial circumstances and individual attitudes. This finding includes links between county histories of slavery and white Americans’ political identification and racial attitudes (Acharya, Blackwell, and Sen 2016), contemporary anti-Semitism among Germans and their states’ historical support for the Nazi Party (Mocan and Raschke 2016), and current levels of trust among Africans whose ethnic groups were more and less heavily trafficked during the slave trade (Nunn and Wantchekon 2011). Local racial climates also appear be marked by historical circumstances. These findings include correlations between a U.S. county’s concentration of enslaved people and its postemancipation lynching trajectory (Acharya, Blackwell, and Sen 2016) and between Medieval pogroms in German communities and Nazi-era anti-Semitic violence (Voigtlander and Voth 2012). In the political realm, counties with higher rates of 1960s Klan activity witnessed larger growth in Republican electoral dominance through the end of the twentieth century (McVeigh, Cunningham, and Farrell 2014).

Theoretical Perspectives

Gabriel and Tolnay (2017) use analogies to Ohm’s law in the physical sciences to consider how violent historical events might shape local variation in contemporary realities. Ohm’s law posits that the transmission of a current between two points is moderated by the level of resistance it encounters. We believe that communities where threatened lynchings were prevented likely had higher levels of resistance than those where lynchings occurred. Threatened mob violence provides evidence that white supremacist sentiments prevailed in a community. Threats are unlikely to emerge in places where robust racism does not already govern social life. The collective response however—whether threatened violence is allowed to become lethal—may signal contestation of these arrangements and shape whether toxic levels of bigotry are fostered going forward.

These local experiences may affect observed voting patterns through a variety of mechanisms. Gabriel and Tolnay (2017) suggest that resistance may be facilitated by political, organizational, or population-based factors. For example, Black people, who overwhelmingly support Democrats, may have lower levels of voter participation (Williams, forthcoming) in communities that allowed mob violence to evolve into murder compared to those that suppressed a potential lynching. White racial identity may be more salient in communities that experienced a lynching (Smångs 2017) rather than thwarting one, facilitating geographic variation in white voters’ responsiveness to Trump’s message. Indeed, Richard Hogan (2018) finds that during Reconstruction, different kinds of attacks on freedmen (outrages, vigilantism, and lynching) represented distinct forms of suppression and were uniquely connected to political processes and economic structures.

Research has also shown spatial and temporal variation in lynching to be associated with racial threat (Blalock 1967; Tolnay and Beck 1995; Bailey and Tolnay 2015). This perspective suggests that as the dominant group perceives an increasing threat to economic well-being, political power, or social status from a minority group, it will act to suppress the subordinate group. More recent work (Wimmer 2013) emphasizes that salience of racial boundaries and struggles over symbolic and actual power fluctuate with political opportunities, rather than “factual” distinctions between groups. To the extent that Trump’s rhetoric signaled an opportunity for whites to mobilize political structures to neutralize perceived threats to their monopolization of power, we might anticipate his candidacy was most heartily embraced by voters from communities that allowed lynchings to occur.

Cultural practices, social norms, personal values, and familial socialization likely all play a role in the persistence of local racism (Voigtlander and Voth 2012). Social settings, particularly in formal and informal organizations, can shape perceptions of intergroup competition (Cunningham 2012). Racial violence may amplify racial hostilities (Acharya, Blackwell, and Sen 2016), and, as prior scholarship has suggested about slavery, both normalize harsh punishment and sharpen attitudes about race relations (Vandiver, Giacopassi, and Lofquist 2006). Intensified racism may also result from violence, as perpetrators frequently construct biased attitudes post hoc to explain or justify their actions (B. Valentino 2013, 30–65). To the extent that the dispositions and attitudes survive and are adopted by a broader swath of the local population, the “mean value” of racial animus would shift. Indeed, racial bias may represent, in part, “the cognitive residue of past and present” social arrangements (Payne, Vuletich, and Brown-Iannuzzi 2019, 11693).

To be clear, we do not contend that all, or perhaps any, people living in counties where mob violence was threatened or allowed to occur personally support violence as a means of racial suppression. Rather, in places where historical community norms legitimated violence as a means of protecting group interests (Suttles 1972), social networks and processes of social reproduction might render individuals sympathetic to more muscular expressions of racism (Blee 2004). If we conceptualize the push to “Make America Great Again” as a social movement, we would anticipate spatial variation in cultural receptivity to messages promoted by the Trump campaign and the policies it advanced (McVeigh, Welch, and Bjarnason 2003). Candidate Trump’s promise to reify intergroup boundaries may have been particularly attractive, and his oblique endorsement of violence as a tool to accomplish this more palatable, to people operating within these communities.

Data and Methods

Our research encompasses instances where racist mob violence resulted in someone’s death—a “lynching”—as well as those where a lynching was publicly threatened, but no one was killed—a “threat.” ⁴ We employ a definition of lynching that requires evidence that (1) a person was killed; (2) the killing was extralegal; (3) the murder was perpetrated by at least three individuals; and (4) the murder was justified with reference to honor, justice, or tradition (Waldrep 2000). Recent scholarship identifies thousands of additional threats—made so credibly and publicly that the local newspaper predicted a lynching—but where the person threatened was not killed (Beck 2015, 2019). A “threatened lynching” meets criteria for a lynching in all respects except that no one was killed. Using both lynching and “threatened lynching” in this article allows us to specify how a community’s historical willingness to allow a threat of collective violence to escalate into murder, or not, might affect resistance to the dominance of toxic white supremacy and ultimately shape its contemporary voting patterns.

Interrogating the relationship between community histories of violence and contemporary voting behavior requires two key measures: the number of lynchings and threatened lynchings that occurred in each county during the lynching era; and the distribution of votes cast for each candidate during the 2016 Republican primary and presidential elections. ⁵ Counties are our units of analysis.

Findings

We find positive relationships between historical violence and electoral support for Donald Trump in the Republican primary and the Trump-Pence ticket in the November general election. Trump performed better in counties where a larger share of incidents of potential racist mob violence targeted against Black people turned lethal. We present the main findings from our statistical models in Table 1. ¹⁰ The association between voting patterns and historical racist mob violence is notable, and that association persists despite the inclusion of metrics like the level of urbanization and the economic dominance of “working-class” industries (results are not presented but are available in an online repository), ¹¹ which are well documented to be linked to Trump’s political base (see, e.g., Muro and Liu 2016; Pew Research Center 2018). Support for Trump was negatively associated with county evangelical concentration in the primary but positively associated in the general election, reflecting initial support for other candidates among conservative Christians.

OPEN IN VIEWER

Table 1

Association between Historical Mob Violence and Share of Votes Cast for Trump and Romney

	2016 Republican Primaries		2016 Presidential Election		2012 Presidential Election
	Model 1	β	Model 2	β	Model 3	β
Ratio of lynchings to all threats of violence	1.864** (0.715)	.090	2.505** (0.789)	.063	2.591** (0.833)	.071
Percentage voting in Republican primary	0.134* (0.066)	.097
Difference in Trump and Romney votes			0.481***(0.060)	.179
N	551		652		652
F Statistic	35.80		199.78		146.62
Adjusted R-squared	.410		.771		.691

NOTE: Coefficients appear first, with standard errors in parentheses. All covariates are included in each model.

*

p < .05. **p < .01. ***p < .001.

In additional model specifications, ¹² we find that the number of lynching events where Black people were killed was not related to a county’s voting results. The number of threatened (Black) lynchings that did not result in a killing also evidenced no systematic relationship to the percentage of votes that Trump was able to cultivate. In short, the distinction between being a community that does and does not tolerate lethal mob violence against Black people is what shaped its residents’ receptivity to Donald Trump as a political candidate, rather than the tempo of that violence or the number of times an unfulfilled threat was made. We interpret this finding to mean that communities evidencing less historic resistance to racial violence are more friendly to racially hostile political rhetoric today.

We find, then, support for both of our hypotheses. A larger percentage of voters supported Trump in counties with greater percentages of “successful” lynchings of Black people, as we predicted. This expectation—that the ratio of completed to all potential (Black) lynchings would predict Trump voting—was fulfilled during both the 2016 Republican primaries and the November general election. ¹³ We also tested additional models predicting the percentage of the vote cast for Mitt Romney in 2012 and find a similar positive relationship with historical racist mob violence. ¹⁴ As the standardized coefficients presented in Table 1 demonstrate, the largest effect size is associated with the share of 2016 presidential primary votes cast for Trump. We plot predicted values for this relationship for each election in Figure 1.

OPEN IN VIEWER

Figure 1

Predicted Percentage of Votes for Trump or Romney, by Level of Lethal Violence. Panel A: 2016 Primary. Panel B: 2016 General Election. Panel C. 2012 General Election.

We depict in Figure 2 the percentage of votes cast for each candidate across the spectrum of racial violence. Panel A presents the measured results. While the slope of the line signifying the relationship between historical violence and Trump voting appears to be steeper than that for either Trump or Romney in the general election, the overall percentage is much lower, owing to the large number of candidates standing for party nomination in the primary. Panel B, then, standardizes both the distribution of violence and the percentage of votes that each candidate garnered. The standardized measure clearly demonstrates a stronger relationship between violence and votes for Trump.

OPEN IN VIEWER

Figure 2

Observed Relationship between Election Results and Historical Violence. Panel A: Percentage of Votes for Each Candidate by History of Violence, 2012 and 2016. Panel B: Violent History and Percentage of Votes Cast for Each Candidate, Standardized

The racio-political cultures persisting in communities affected by collective murder are strikingly different from those not affected. Figure 3 illustrates how the relationships between voting and other social indicators are governed by local experiences with violence. In counties with no history of lethal mob violence, represented by the solid black line, the expected relationship between the proportion of the population that is non-Hispanic white and the share of votes cast for Trump is small and somewhat negative. The positive association between Trump support and white racial isolation, represented with the dashed line, emerges only in communities where mob violence targeting Black people became deadly. This suggests racially isolated whites were most likely to vote for Trump if they lived in places that had at some point embraced violent racial terror.

OPEN IN VIEWER

Figure 3

Trump Support in the 2016 Primary, by Percentage White and Lynching History

Discussion

This article has examined whether durable traces of community histories of mob violence might be observed in the preferences of its voters nearly a century after the close of the lynching era. We identified that, controlling for a variety of alternative explanations for the level of support evinced for Donald Trump’s candidacy, larger percentages of voters supported him in places where the community allowed a greater share of mob violence threats to become lethal. Viewed in light of connections that prior scholarship has made between historical and contemporary patterns of racialized behavior, we believe that these findings support the growing body of evidence that suggests that absent meaningful resistance (Gabriel and Tolnay 2017), aspects of local racial culture endure over decades, and perhaps centuries. That the relationship is robust to including historical and contemporary variables, and when examining more generalized Republican voting patterns, gives us confidence in this assertion.

The strength of the relationship that we identify would seem to belie the claim that Trump’s 2016 campaign rhetoric was not overtly racialized or intended to galvanize racist hostility (Lopez 2016). Indeed, as Mattias Smångs’s work (2017) identifies, lynching’s specific function was frequently to solidify white racial identity. His article in this volume (2021) similarly connects community presence of an organization built on a specific vision of that identity—the Ku Klux Klan—to Trump’s electoral strength. Our supplementary analyses, linking historic violence with George Wallace’s 1968 campaign results, reach a similar conclusion. ¹⁵

To put it plainly, stronger local endorsement of Donald Trump’s candidacy in 2016 may also be associated with contemporary behaviors and organizational practices that are both tangibly understood in racial terms and more prevalent in communities with histories of mob violence: noncompliance with federal hate crimes reporting, higher levels of interracial homicide, and terrorist activities like church burnings. Importantly, because voting behavior determines the operatives whose hands guide the levers of state resources and power, the persistence of these associations also reinforces “feedback loops” and the capacity for white supremacists to monopolize governance structures and further sculpt the social landscape to reflect their racial vision.

We are unable to comment on whether the patterns that we observe are causal or merely associative. It seems more likely that historical patterns of mob violence reflected, rather than resulted from, local racial dynamics. However, it is also quite plausible that the collective experience of witnessing or participating in a lynching would launch a cascade of unanticipated consequences, and sharpened racial animosities might be one such result. Similarly, an effective community response to suppress a potential outburst of violence is probably only possible in communities with high underlying levels of social organization and strong opposition (or, in Gabriel and Tolnay’s [2017] parlance, resistance) to visceral displays of racial antagonism. Successfully thwarting a threatened episode of collective violence, however, may also make visible the strength of what, today, would be called antiracist sentiment, and foster support for greater racial equity. In any case, a systematic relationship appears to exist between a community’s history of mob violence and the willingness of its members to endorse the racially charged candidacy of Donald Trump.

Despite our own hypotheses, informed by empirical evidence and theoretical propositions, we were dismayed at what our analyses revealed. While this research contributes to the growing body of scholarship focused on local factors associated with support for Trump’s candidacy and administration, we are profoundly disappointed at what it signals about our country. That the impulses driving racist animosity and the hard edges of mob violence have survived into the twenty-first century is no real surprise. That the geographic contours of collective violence, and community resistance to that violence, a century ago can predict today’s presidential voting behavior gives us pause.