SENSING SOCIAL ORDER: POLITICAL SYMBOLS, COALITIONAL COGNITION, AND PERFORMATIVE VIOLENCE

Political symbols and performative political violence share an important function: both are tools for inferring and influencing the social order. I define the social order as a more or less shared mental representation of group boundaries and status hierarchies. Who belongs and who is excluded? Who is dominant and who is subordinate? Both symbols and violence “solve” fundamental problems of the social order: how to “sense” it and how to shape it. The social order is not directly observable in the context of large-scale modern societies characterized by imagined communities (Anderson Reference Anderson2006) and is always inferred. Moreover, the social order is nearly always dynamic and contested to some degree because actors want to influence it to achieve their own aims.

How do we build such a mental representation given that the social order is not directly observable? Evidence from evolutionary and cognitive psychology suggests that we possess a specialized cognitive apparatus to make inferences about the social order and social alliances: a kind of coalitional cognition (Cikara Reference Cikara and Gawronski2021; Kurzban, Tooby, and Cosmides Reference Kurzban, Tooby and Cosmides2001; Lopez, McDermott, and Petersen Reference Lopez, McDermott and Petersen2011; Pietraszewski Reference Pietraszewski2022; Pietraszewski, Cosmides, and Tooby Reference Pietraszewski, Cosmides and Tooby2014). I argue that political symbols activate this coalitional cognition and allow us to track status hierarchies and mark group boundaries. Coalitional cognition is likely triggered and sustained by emotions having to do with status and belonging: envy, humiliation, shame, anger, pride, and resentment (Barnhart Reference Barnhart2021; McClendon Reference McClendon2018; McDermott Reference McDermott2020; Mendelberg Reference Mendelberg2022; Sznycer et al. Reference Sznycer, Tooby, Cosmides, Porat, Shalvi and Halperin2016; Tooby and Cosmides Reference Tooby, Cosmides, Lewis, Haviland-Jones and Berrett2008). Coalitional cognition relies upon common knowledge heuristics: mental shortcuts that inform our assumptions about what is widely known and what is not (De Freitas et al. Reference De Freitas, Thomas, DeScioli and Pinker2019). I argue that political symbols are an essential input of coalitional cognition: enabling us to both infer and influence the social order.

Political symbols can be understood as the tip of an iceberg.Footnote ¹ We can see what is present above the surface (i.e., a public monument) and use that to infer the size, shape, and location of what is underneath (the social order, power relations). The cognitive challenge is that we cannot directly access what is under the surface, we must infer it. Because it is impossible to attend to the full extent of possible indicators of the social order, political symbols serve as a valuable heuristic. That “filling in” may be based on previous experience, education, and knowledge. I do not argue that monuments are the sole source of information about the social order, only that they are critical cues.

Drawing upon and synthesizing a large interdisciplinary literature on symbols and politics, I define political symbols as common knowledge heuristics with content about the social order. Political symbols have to do with common knowledge because they generate intersubjective or mutualistic knowledge. Their main function is informing us what other people know. They are heuristic because we make these intersubjective assumptions automatically rather than consciously. They are about the social order because they communicate information about collective identity and hierarchy (Abdelal et al. Reference Abdelal, Herrera, Johnston and McDermott2006; Mendelberg Reference Mendelberg2022). In particular, political symbols help define who we are and what is our relationship to other groups.

Examples of symbols include monuments, flags, rituals, parades, religious/nationalist sites and events. A large and diverse body of scholarship has investigated the causal effects of symbols on a range of political outcomes, although there is no widely agreed-upon definition (Butz Reference Butz2009; De Juan et al. Reference De Juan, Haass, Koos, Riaz and Tichelbaecker2024; Dinas, Martinez, and Valentim Reference Dinas, Martinez and Valentim2024; Kaufman Reference Kaufman2019; Kertzer Reference Kertzer1988; Reference Kertzer1996; Manekin, Grossman, and Mitts Reference Manekin, Grossman and Mitts2019; Rahnama Reference Rahnama2024; Wedeen Reference Wedeen1999). Existing literature has conceptualized symbols in five main ways: as meaning-makers (Wedeen Reference Wedeen2002; Geertz Reference Geertz1973), political resources (Cohen Reference Cohen1974; Edelman Reference Edelman1964; Laitin Reference Laitin1986), deep-seated attitudes (Cobb and Elder Reference Cobb and Elder1973; Kaufman Reference Kaufman2001; Mendelberg Reference Mendelberg2022; Sears Reference Sears, Iyengar and McGuire1993), a special type of information (Chwe Reference Chwe2001; O’Neill Reference O’Neill2001), and as identity markers (Elgenius Reference Elgenius2018; Lamont and Molnár Reference Lamont and Molnár2002; Mach Reference Mach1993; Ross Reference Ross2007).

I contribute to this literature by introducing the heuristic nature of political symbols and connecting it to the problems of inferring and influencing the social order. By way of illustration, consider the Berlin Wall as a political symbol which signified the division of Germany into two polities, as well as the broader ideological struggle of the Cold War. An observer looking at the Wall was able to infer the prevailing social order. However, as soon as the Wall came down, the Germans who tore it apart challenged the old social order, communicating the beginning of reunification, and heralding the end of the Cold War. Alternatively, the old bridge in the city of Mostar in Bosnia and Herzegovina was destroyed during the ethnic civil war that convulsed the former Yugoslavia in the 1990s. Its reconstruction signified the postwar order of reconciliation and inter-ethnic amity. Similarly, the monument at the pearl roundabout in Manama, Bahrain, which served as the locus of Arab Spring demonstrations in 2011, was subsequently destroyed by the country’s government and turned into a traffic intersection. The demolition meant to illustrate that the relatively more open era that birthed the protests was over. Finally, following the fall of Syrian President’s Bashar Al-Assad regime in 2024, ordinary Syrians spontaneously took to the streets to remove and deface statues of Al-Assad and his father, Hafez, marking the end of that family’s long rule. These are four examples among many.

Considering political symbols in this light has important and under-explored implications for the likelihood of political violence, especially status- or honor-motivated violence. One of the main drivers of political violence is status (Renshon Reference Renshon2017). Scholars of interstate wars have found that states frequently fight due to inconsistency between their perception of their deserved status and their actual status (Dafoe, Renshon, and Huth Reference Dafoe, Renshon and Huth2014; Duque Reference Duque2018; Paul, Wohlforth, and Larson Reference Paul, Wohlforth and Larson2014; Renshon Reference Renshon2016). Similarly, Barnhart (Reference Barnhart2021) and Petersen (Reference Petersen2002) have shown how humiliation and resentment, especially when associated with actual or perceived reversals in power, frequently lead to violence as groups attempt to regain lost position in a hierarchy. However, the status literature does not address whether lost status leads to particular kinds of violence, and how actors know that status has been regained. This is especially challenging since status is by definition something that is conferred by others (Renshon Reference Renshon2017). Which patterns of political violence do status motivations lead to (Gutiérrez-Sanín and Wood Reference Gutiérrez-Sanín and Wood2017)?

I argue that status injuries are likely to lead to performative, rather than merely coercive, violence. While violence can be purely instrumental, for instance when used to kill a would-be assailant, it can also be used to “send a message,” for instance in a terrorist attack or forms of extra-lethal violence (i.e., the mutilation of a dead body). Violence is frequently motivated by “moral” considerations: people carry out violent actions not only because they believe that violence will lead to just outcomes but also because they believe that the violence itself is just (Fiske and Rai Reference Fiske and Rai2014; Slovic et al. Reference Slovic, Mertz, Markowitz, Quist and Västfjäll2020). Violence is seen as a way of “ratifying” desired social relationships and political orders, and existing evidence suggests that participating in violence in and of itself strengthens group identification (Cohen Reference Cohen2016; Fiske and Rai Reference Fiske and Rai2014; Kalyvas, Shapiro, and Masoud Reference Kalyvas, Shapiro, Masoud, Shapiro, Kalyvas and Masoud2008; Littman Reference Littman2018). Performative and public violence is used to clarify group boundaries and status hierarchies: “When actors put violence on display, they are bringing to life ideas about how the world should be and more specifically, how it should be ordered—who should have power and who should be included and on what basis people should claim belonging” (Fujii, Finnemore, and Wood Reference Fujii, Finnemore and Wood2021, 3). In sum, performative violence serves a similar function to political symbols: to communicate information about the social order.

The overlapping function of political symbols and political violence raises the possibility that symbols may act as a substitute for performative violence. In that sense, political symbols whose meaning is aligned with the aims of status-based violence are likely to reduce rather than increase political violence. Political symbols shape perceptions of the status quo and therefore affect the motivation to carry out status-based violence. If the social order has been restored to its “rightful” state, then violence that “fixes” it is unnecessary. Paradoxically, it is precisely when symbols reify the social order that violence is meant to affirm that they reduce the likelihood of future violence by attenuating the motivation to carry it out. Rather than creating a permissive environment for violence, political symbols show that the social order is already secure, making the violence less necessary. Conversely, when political symbols are removed or altered, we should expect the motivation to carry out political violence to increase, as actors seek to protect an existing social order that is now perceived to be under threat.

Put differently, political symbols enable individuals to evaluate and shape the balance of power among social groups. In doing so, they directly influence perceptions of group power and threat—factors long recognized as causes of violence and social control, particularly in the context of U.S. race relations (Blalock Reference Blalock1967; Key Reference Key1949). Precisely because the social order is impossible to directly observe, is inherently contested, and may involve contradictory signals, political symbols serve as powerful heuristics for hierarchy and group belonging.

THE POST-RECONSTRUCTION UNITED STATES SOUTH, 1877–1928

I test this theory by looking at the effect of Confederate monument presence and removal on anti-Black performative violence (lynching/public executions and hate crimes) in the US. The aftermath of the American Civil War and the Reconstruction period represent an important case of contestation over the post civil war order (Byman Reference Byman2021). The Union victory in the civil war destroyed the antebellum white supremacist social order, which was defined by the institution of chattel slavery. The defense of slavery was the prime motivator of Southern nationalism, secession, and mobilization for war. Following the war, Southerners were contesting the contours of the new social order in the 11 former Confederate states.Footnote ² For the formerly enslaved, two of the most important post-war goals were guaranteeing the franchise and equal protection under the law. They had also played a critical role in securing the defeat of the Confederacy and their own freedom. By 1865, 10% of all Union troops were Black and, equally important, Black Southerners provided valuable intelligence to Union troops throughout the war (Litwack Reference Litwack1979).

For most white Southerners, the end of the war represented an almost unfathomable humiliation. White Southerners faced a reality in which a large proportion of military-aged white Southern males died during the war, Union soldiers were still occupying the South (including eighty thousand Black Union troops [Litwack Reference Litwack1979]), they had suffered enormous economic losses due to the destruction of slavery, and were now forced to treat their former chattel as social and political equals. In addition, former Confederate combatants were not allowed to vote or carry weapons until they re-pledged allegiance to the Union. The post-war situation was described in apocalyptic terms by white Southerners sympathetic to the Confederacy. In her 1914 textbook, future UDC Historian-General Laura Rose Martin, describes the Reconstruction period as far worse than the actual civil war, calling Reconstruction “a situation fraught with more terrors than the war itself” and as a “bondage worse than death” (Rose Reference Rose1914, 16-7).

At the risk of oversimplifying a complex historical period, I conceptualize the period between the end of the Civil War and the end of Reconstruction (1865–77) as a protracted contest between a vision of an unranked, more egalitarian political order and a ranked, hierarchical one (Horowitz Reference Horowitz2001).Footnote ³ On the one hand, the Reconstruction period represented an attempt at creating an unranked ethnic system in the form of a multiracial democracy, what Foner (Reference Foner2019) has called the Second Founding (Du Bois Reference Du Bois and Gates2014; Foner Reference Foner2002).Footnote ⁴ On the other hand, we can characterize “redemption” policies—which involved the use of violence, electoral intimidation, and attempts to legally restrict democratic participation—as an attempt by Southern whites to restore central elements of a social order defined by white superiority and impose a system of legal segregation (Byman Reference Byman2021; Du Bois Reference Du Bois and Gates2014). We now know that the ranked vision won the day, establishing authoritarian enclaves in the former Confederacy until the Civil Rights Movement of the mid-twentieth century, and over time, forcing many who objected to flee northward (Mickey Reference Mickey2015; Tolnay and Beck Reference Tolnay and Beck1992). Du Bois writes that over this time period “a determined psychology of caste was built up. In every possible way it was impressed and advertised that the white was superior and the Negro an inferior race. This inferiority must be publicly acknowledged and submitted to” (Du Bois Reference Du Bois and Gates2014, 695). I argue that white Southerners used the construction of Confederate monuments and the gruesome performative violence of lynching—alongside public executions—to build this social order.

Lynching and Performative Violence

The post-Reconstruction period in the former Confederacy was characterized by pervasive racial terrorism. There were 3,823 lynching victims between 1877 and 1928, and likely far more whose deaths were never recorded. By 1928, nearly 70% of counties in the 11 former Confederate states experienced at least one lynching event. This figure was over 90% in Mississippi, Louisiana, and Alabama. See Figure 1 for a spatial distribution and Table A5 in Section F.8 of the Supplementary Material (SM) for a full breakdown by state. Even shortly after the time period, Black thinkers understood lynching’s broader purpose as far more than mere vigilantism: “Lynchings serve the indispensable social function of providing whites with the means of periodically reaffirming their collective sentiment of white dominance” (Cox Reference Cox1945, 581). The practice of lynching was a “routine and systematic effort to subjugate the African-American minority” (Tolnay and Beck Reference Tolnay and Beck1995, 17) through the use of extralegal, extra-lethal,Footnote ⁵ quasi-ritualistic violence (Cox Reference Cox1945; Wells-Barnett Reference Wells-Barnett2002; Wood Reference Wood2009; Patterson Reference Patterson1998). This kind of lynching was distinct from other forms of vigilante violence in U.S. history, with scholars suggesting that this period represents a unique “lynching regime” governed by the logic of ethnic terrorism (Seguin and Rigby Reference Seguin and Rigby2019; Staggs Reference Staggs2021).

Figure 1. Spatial Distribution of Lynching Events in the Former Confederacy by 1928

Note: The map shows the spatial distribution of lynching events by 1928 in the counties of the 11 former Confederate states (AL, AR, FL, GA, LA, MS, NC, SC, TN, TX, VA). Counties highlighted in gray experienced at least one lynching event between 1870 and 1928. Counties highlighted in white did not.

One of the core functions of lynching was to serve “as a symbolic manifestation of the unity of white supremacy” (Tolnay and Beck Reference Tolnay and Beck1995, 50). These extralegal activities were often carried out openly and proudly, with grisly depictions of the violence shared via postcards, stories, and newspapers afterwards (Wood Reference Wood2018, 759). This was, in many ways, the whole point. Lynchings were often festival-like gatherings with large crowds, which reflected widespread public sentiment, and in which it was not unusual for prominent members of a community to participate in the mob (Douglass Reference Douglass1892; Tolnay and Beck Reference Tolnay and Beck1995; Wells-Barnett Reference Wells-Barnett2002).

Contrary to common perception, lynchings were likely not a result of action by organized groups such as the Ku Klux Klan (KKK). The lynchings of the post-Reconstruction period peaked in the early 1890s, during years in which the KKK was likely inactive (see Figure 2). Although the use of KKK discourse and ideology persisted throughout the late nineteenth and early twentieth century, the first KKK was disbanded by 1872, five years prior to Reconstruction’s end, and was only reconstituted as the second KKK in 1915 (Parsons Reference Parsons2015). This understanding of lynching is consistent with contemporary scholarship about vigilantism which recognizes its influence on public discourse and social hierarchies (Bateson Reference Bateson2021). Indeed, recent work shows that lynchings were more likely in counties that had been occupied by Black union troops during the Civil War and Reconstruction (Byun and Kwon Reference Byun and Kwon2025), in counties in which Democrats lost the popular vote (Testa and Williams Reference Testa and Williams2025), and in counties that had a Freedman’s Bureau office (Chyn, Haggag, and Stuart Reference Chyn, Haggag and Stuart2024). This evidence conforms with the notion that lynching violence was rooted in a desire to reverse perceived status loss and shape power relations.

Figure 2. Lynching Events, Lynching Victims, and Confederate Monument Dedications in the Former Confederate States, 1870–1928

Note: The plot summarizes the number of lynching events, lynching victims, and Confederate monument dedications in the former Confederate states between 1870 and 1928. The solid black line represents the number of lynching events. The two-dashed gray line represents the number of lynching victims. The dashed black line represents the number of Confederate monuments dedicated.

Lynchings proliferated despite the fact that the Southern legal system remained unmistakably hostile to Black Americans (Litwack Reference Litwack1979). In fact, both lynchings and public executions were violent spectacles that frequently had similar features. Lynchings emulated and took on the characteristics of the public executions of the time period: “The boisterous behavior at lynching rituals, the taking of photographs and the collecting of souvenirs, were carryovers from public hangings” (Wood Reference Wood2018, 773). While political scientists routinely make the analytical distinction between legal, state-led violence and vigilantism, I propose a different kind of distinction here. In the case of the postbellum South, where legal and extralegal violence worked in tandem to enforce racial hierarchy, the relevant distinction is between public, performative violence and purely coercive violence. I argue that public hangings and lynchings are conceptually similar. To be clear, this is an ideal-type distinction and many forms of violence might include both performative and coercive dimensions. There is no doubt that lynching and public executions were coercive in addition to being performative.

While instances of lynching continued throughout the twentieth century, their frequency declined substantially from their peak in the early 1890s, as can be seen in Figure 2. Scholars are divided about the causes of lynching’s decline: variously attributing it to changes in the economic structure of the Southern states, fluctuations in cotton prices, an increase in state violence against Blacks, as well as greater normative pressure from anti-lynching activists and the Northern public (Bailey and Tolnay Reference Bailey and Tolnay2015; Clarke Reference Clarke1998; Smångs Reference Smångs2016a; Reference Smångs2016b; Tolnay and Beck Reference Tolnay and Beck1995; Weaver Reference Weaver2019). While these explanations are plausible they are also incomplete. Economic explanations struggle to account for lynching’s performative nature: it is not immediately obvious why a change in prices would lead someone to hang or mutilate their neighbor in a public spectacle. The argument that state-led violence—in the form of legal executions—acted as a substitute to lynchings (see Clarke Reference Clarke1998) does not address why legal violence against Blacks, which was a readily available option throughout the peak of the lynching era, was suddenly deemed preferable to the extralegal violence represented by lynching. Moreover, this argument does not distinguish between lynch-like public executions by hanging and private executions. The latter were carried out via the electric chair without an audience, within the confines of a penitentiary, and can therefore be considered more strictly coercive. Public executions follow similar patterns to lynching while private executions do not (as can be seen in Figure A10 in Section B.3 of the SM).

The contention that normative pressure—facilitated by the development of railroads and applied through the work of anti-lynching activists—was effective in reducing lynching (Weaver Reference Weaver2019) is appealing but there are reasons to believe that its effects were partial. Lynching was condemned nationally and even locally within the South, but the practice persisted and was a central focus of civil rights organizations well into the mid-twentieth century. Chronologically, the NAACP was not founded until 1909, well past lynching’s peak, and struggled to build effective chapters in the South from its headquarters in New York. For instance, in a 1925 letter from the NAACP’s Director of Branches to a Mr. William Porter of Montgomery, Alabama, the Director laments that “…looking over our records, I find that Alabama has made an exceedingly poor showing, most of the branches of the state having done very little for the Association.”Footnote ⁶ Additionally, the political power of the Southern states, especially in the U.S. Senate, was robust enough to resist federal anti-lynching legislation well into the 1950s.

By the 1910s, mainstream historical understanding of the Civil War and Reconstruction was overwhelmingly sympathetic to the South or at least to the idea of sectional reconciliation. This trend was exemplified by the screening of the film Birth of a Nation in 1915 in Woodrow Wilson’s White House, as well as the 1914 dedication of a Confederate monument at Arlington National Cemetery. And as Ang (Reference Ang2023) has shown, screenings of the movie increased rates of violence and KKK membership. In addition, Bazzi et al. (Reference Bazzi, Ferrara, Fiszbein, Pearson and Testa2023) provide evidence that as Confederates and their descendants migrated away from the former Confederacy, they brought Confederate culture along with them. Moreover, during the 1920s, the revived, second KKK was a powerful national organization with millions of members.

Descriptive patterns suggest that the logic of substitution described above is plausible, although of course, I am not suggesting that monument construction is the sole cause of change in performative violence. Figure 2 shows that as the rate of lynching declined, the rate of monument dedications increased (see also Figures A10 and A11 in Section B.3 of the SM). Notably, this figure does not show continued monument presence in a given year, only the dedication of new monuments in a given year.

Confederate Monuments

Drawing on the theoretical account of political symbols, I propose an under-appreciated cause of the decline in lynchings and anti-Black performative violence: the proliferation of Confederate monuments throughout the post-Reconstruction South. The primary purpose of Confederate monument construction was the glorification of the Lost Cause and absolution for Confederate soldiers and their leaders (Cox Reference Cox2004; Brundage Reference Brundage2008; Wilson Reference Wilson1980). This was not only an ideological battle. Understandably, the descendants of Confederate soldiers did not want their own children to view their fathers, uncles, and grandfathers as traitors to the US. As one UDC Secretary-General put it to the organization’s annual convention in 1917, its members have been “working with tireless zeal… to teach our children that not one drop of the blood of traitors flowed through the veins of their Confederate ancestors!” (Odenheimer and Merchant Reference Odenheimer and Merchant1918, 8).

Monument construction was a central pillar in a broader set of activities meant to rehabilitate the reputation of the defeated Confederacy and transmute the perceived shame and humiliation of Reconstruction into a source of pride. Although the architects of this strategy were a range of Confederate heritage organizations engaged in monument construction (including the United Sons of Confederate Veterans, United Confederate Veterans, and Ladies Memorial Associations), it was the women of the UDC who were the most effective builders of Confederate monuments. They launched highly effective campaigns to build monuments, introduced textbooks teaching their preferred version of history into schools, wrote catechisms to instruct children in the history of the Confederacy, and established rituals to memorialize the Confederate dead (such as marking Confederate Memorial Day). Figure 3 shows the spatial distribution of UDC chapters and Confederate monuments.Footnote ⁷ While the two are not perfectly correlated, it is evident that most counties with monuments also had a UDC chapter. Equally important, it is worthwhile noting that not all counties with UDC chapters also had a monument, reflecting the fact that other organizations were also responsible for monument construction and that not all UDC chapters were equally effective.

Figure 3. Spatial Distribution of United Daughters of the Confederacy (UDC) Chapters and Confederate Monuments

Note: The map shows the spatial distribution of UDC chapters and Confederate monuments by 1928 in the counties of the 11 former Confederate states (AL, AR, FL, GA, LA, MS, NC, SC, TN, TX, VA). Counties highlighted in dark gray have a UDC chapter present by 1919. Counties highlighted in light gray do not. Black dots indicate the presence of a Confederate monument in a county by 1928.

The dedication ceremonies of Confederate monuments reflected both the importance of this intergenerational mission, as well as the white supremacist message they conveyed. The dedications of monuments were often grand, community-wide affairs that included parades and speeches, frequently involving white children unveiling the monuments as if they were gifts to be opened (Cox Reference Cox2004). By the end of the nineteenth century, UDC-funded Confederate monuments were often built in public and conspicuous spaces, such as courthouses and town squares precisely so that children could routinely observe them and internalize the status quo they communicated (Brundage Reference Brundage2008; Cox Reference Cox2004). This shift in location also coincided with a shift in the content of the inscriptions on the monuments, which became far more focused on Lost Cause ideology, as opposed to individual memorializing of the fallen (O’Connell and Forrest Reference O’Connell and Forrest2020). Dedication ceremonies also included speeches by prominent public figures or Confederate veterans.Footnote ⁸ The speeches are quite explicit about the purpose of the monuments: celebrating the reversal of the unjust social order of the Reconstruction era and the affirmation of the re-emerging white supremacist order. The following excerpt from a speech given at the 1913 dedication of a monument on the campus of the University of North Carolina in Chapel Hill is illustrative.

The present generation, I am persuaded, scarcely takes note of what the Confederate soldier meant to the welfare of the Anglo Saxon race during the four years immediately succeeding the war, when the facts are, that their courage and steadfastness saved the very life of the Anglo Saxon race in the South when “the bottom rail was on top” all over the Southern states, and to-day, as a consequence the purest strain of the Anglo Saxon is to be found in the 13 Southern States. Praise God. (Carr Reference Carr1913)

The passage shows the desire to transmit coalitional information about racial hierarchy across generations. The phrase “the bottom rail was on top” was used to describe the perceived status inversion that white Southerners resented during the Reconstruction era. Confederate heritage organizations actively fostered this sentiment. For instance, the initiation ritual of the USCV included a definition of the Confederacy as the last “white man’s government that the world ever saw.”Footnote ⁹

While they were relatively small, Confederate heritage organizations were effective at marshaling people and resources to help achieve their goals. For instance, a company operating a casino and dancing pavilion in Montgomery, AL, provided the free use of its facilities for a 1903 fundraiser to build a monument.Footnote ¹⁰ In other instances, railroad companies provided subsidized transport to unveiling ceremonies in order to increase attendance.Footnote ¹¹

RESEARCH DESIGN AND EMPIRICAL STRATEGY

To test these hypotheses, I employ a difference-in-differences design with staggered treatment timing to estimate the causal effect of Confederate monuments on the likelihood of lynching between 1877 and 1928 in the 11 former Confederate states. I construct a panel dataset of Confederate monuments and lynchings between the years 1877 and 1928, which takes the county and 3-year bins as its primary units of analysis (county-bin). The county is the standard unit of analysis used in previous work (Seguin and Rigby Reference Seguin and Rigby2019; Tolnay and Beck Reference Tolnay and Beck1995). I use 3-year bins as the time unit for two primary reasons. First, staggered treatment estimators do not perform well and risk losing observations if there are small treatment cohorts in specific years. While binning sacrifices some granularity, it provides more reliable treatment effect estimates, and most important, uses all the available observations. Second, binning by three years allows me to more easily rule out the possibility of treatment anticipation, a critical identifying assumption for difference-in-differences. While there was variation in the time it took to build monuments, 3 years is a reasonable duration to account for fundraising, design, and construction of a monument (alternative 1- and 10-year bins are presented in Section E of the SM).

I use both lynching count and binary variables (i.e., whether any lynching event took place) as the primary outcomes. While multiple lynchings in the same county in the same year were quite rare, it is more reasonable to test for the quantity of lynching events when using 3-year bins. In addition, throughout the manuscript, I also present results for performative violence outcomes, combining lynching and public executions together since they conform with the central idea that Confederate monuments substitute for performative violence.

I restrict the timeframe to 1877–1928 for both conceptual and practical reasons. Conceptually, the end of Reconstruction in 1877 marked the beginning of the period of “redemption,” while the 1928 onset of the Great Depression fundamentally altered the country. Practically, the years between 1865 and 1877 were marked by upheaval and highly unreliable reporting on lynching, although it is likely that the rate of anti-Black violence was very high (Litwack Reference Litwack1979, 276–7). Following standard practice in the difference-in-difference literature, I conceptualize the treatment as binary (Roth et al. Reference Roth, Sant’Anna, Bilinski and Poe2023). Once a county has a monument dedicated, I consider treatment “on” in that county. Most counties with monuments only have one monument, and no county had a monument removed in the time period.Footnote ¹²

I use staggered difference-in-differences estimators that are designed to accommodate cases in which there are meaningful departures from the classic two-way fixed effects setup (Callaway and Sant’Anna Reference Callaway and Sant’Anna2021; Roth et al. Reference Roth, Sant’Anna, Bilinski and Poe2023; Roth and Sant’Anna Reference Roth and Pedro H. C.2023; Sun and Abraham Reference Sun and Abraham2021). In the case of Confederate monument dedications (the treatment or independent variable), there is staggered rather than uniform treatment timing (monuments were built in different counties in different years), multiple rather than two time periods (a 52-year time period), and the possibility of heterogeneous treatment effects (monuments may have affected the likelihood of lynching differently across counties and time periods). In addition, a causal interpretation of the classic two-way fixed effects estimator relies on several identifying assumptions that can be implausible in most real-world settings. Staggered estimators allow for some violations of these identifying assumptions without losing the ability to make causal claims.

Estimation

The primary estimator I use is called the group-time average treatment effect (Callaway and Sant’Anna Reference Callaway and Sant’Anna2021).Footnote ¹⁴ The intuition behind this estimator is that in cases with staggered treatment timing, counties should first be divided into groups (g) based on treatment date. The parameter estimates the average treatment effect for units in a specific group (g) at a particular time period (t). This approach is meant to accommodate treatment effect heterogeneity. The logic is that groups treated during the same time may be more similar to one another compared with groups treated later or earlier. This feature is especially appealing in settings with relatively longer time frames, as is the case with Confederate monuments. For instance, it is quite possible that counties in which monuments were dedicated in 1910 (call them the 1910 cohort) are more similar to each other as opposed to counties that are part of the 1882 cohort.

Equation 1 puts this approach in potential outcomes notation, where g is the specific group/cohort in which monuments were dedicated, t is a particular time (in my case, 3-year period), Y is the outcome (lynching/performative violence), G is the more general notation for group/cohort, and C is the control group (those counties without monuments either because they are never built or because they have yet to be built in time period t):

⁽¹⁾ $$ \begin{array}{r}ATT(g,t)=E[{Y}_t-{Y}_{g-1}|G=g]-E[{Y}_t-{Y}_{g-1}|C=1].\end{array} $$

The group-time average treatment effect estimator calculates these values for all cohorts across the relevant time frame. The comparisons to years prior to each cohort being “treated” are used to form the pre-treatment trend estimates. I then aggregate these individual parameters into an overall average treatment effect in one of two ways. First, a simple weighted average across the entire timeframe, which is weighted according to the number of observations per cohort, so larger cohorts have greater weight. Second, as event study parameters, which estimate the treatment effect of monuments e periods after a monument is dedicated. Equation 2 presents the event-study estimator, where w is the weight, and $ g+e $ is used instead of t. I report results for both types of aggregation:

(2) $$ \begin{array}{rl}{ATT}_w^e={\displaystyle \sum_g}{w}_gATT(g,g+e).& \end{array} $$

The group-time average treatment effect estimator can also have a causal interpretation if we relax the parallel trends assumption such that it only holds conditional on relevant pre-treatment covariates. In this case, a doubly-robust estimation method is used (see Callaway and Sant’Anna Reference Callaway and Sant’Anna2021; Sant’Anna and Zhao Reference Sant’Anna and Zhao2020). As a robustness check, I also condition the parallel trends assumption on covariates (see Section F.2.2 of the SM) and report alternative difference-in-differences estimators (see Section E.1 of the SM).

As another plausibility check, I also report more standard two-way fixed effects estimation, presented in Equation 3. $ \delta $ is the parameter of interest where $ MonumentCounty $ / $ MonumentRemoval $ is a binary indicator of whether that county-bin has a monument built or removed. $ \alpha $ is a unit fixed effect and $ \gamma $ is the time fixed effect. The subscript i is a unit indicator, in this case a county. The subscript t is a time indicator, in this case 3-year bins. I also use the two-way fixed effects estimators (TWFE) and the same type of specification to test the effect of Confederate monument removal on anti-Black hate crimes:

(3) $$ \begin{array}{rl}\begin{array}{l}PerformativeViolenc{e}_{it}={\alpha}_i+{\gamma}_t\\ {}\hskip11em +MonumentCounty{\delta}_{it}+{u}_{it,}\\ {}AntiBlack{HateCrimes}_{it}={\alpha}_i+{\gamma}_t\\ {}\hskip9.5em +MonumentRemoval{\delta}_{it}+{u}_{it.}\end{array}& \end{array} $$

DISCUSSION, LIMITATIONS, AND FUTURE DIRECTIONS

Taken together, the findings suggest that in the post-Reconstruction era (1877–1928), the dedication of Confederate monuments throughout the former Confederate states reduced the likelihood of performative political violence. I argue that monuments played this role because they were able to substitute for some of the performative, communicative functions played by anti-Black lynching and public executions. This finding is best explained by conceptualizing Confederate monuments as political symbols and taking into account the performative nature and order-shaping purpose of lynching and public executions. I also find that monuments had greater effect in those areas that are most susceptible to racial threat, suggesting that political symbols might be especially salient in such contexts. The results help us better understand the scope conditions of political symbols and the possible mechanisms through which they influence political violence.

The effect of Confederate monuments did not decay, rather, it strengthened slowly with time. This pattern suggests that the impact of political symbols like monuments may be slower to take root, but could be more durable compared with one-off mass rituals, however extravagant and well-attended (Ang Reference Ang2023). Durably shifting perceptions of the prevailing social order—especially in a region scarred by slavery, a brutal civil war, and a tumultuous and violent Reconstruction period—is unlikely to happen quickly and probably requires repeated “exposure.” This logic is consistent with existing studies that show the persistent and long-term legacies of slavery and mass atrocities more broadly on contemporary politics (Acharya, Sen, and Blackwell Reference Acharya, Sen and Blackwell2018; Charnysh and Finkel Reference Charnysh and Finkel2017; Williams Reference Williams2022). We should expect changes in entrenched mass perceptions to take time. That monuments have an effect at all only underscores the importance of better understanding political symbols.

The central findings may also speak to a difference in strength between more temporary symbols (i.e., rituals, flags, or posters) and more permanent physical structures such as monuments. The results regarding Confederate memorial days reported in Section F.4 of the SM lend further credence to this interpretation. Where monuments differ from other “symbolic” activity is that they are highly visible, costly public markers. Recent evidence from Switzerland, for instance, shows that visible mosques (as opposed to inconspicuous ones) lead to an increase in support for far-right parties and anti-migrant initiatives (Valli et al. Reference Valli, Gravelle, Nai, Medeiros, Murri and Eugster2025).

The slow-building nature of the treatment effect might also point to the possibility of a generational change underway, which would be consistent with the mission adopted by the UDC and their ilk. As more white children grew up in the shadow of monuments, where they felt secure in their dominant position and farther away from the searing experience of perceived lower status brought about by the Civil War and Reconstruction, the less motivation there was to engage in the kind of public violence meant to restore lost status. Taylor (Reference Taylor2025) reports some evidence consistent with this interpretation, finding that counties with monuments are likelier to have a greater proportion of children named after Confederate leaders.

Of course, just because the rate of lynchings and public executions declined does not mean that they disappeared altogether. As Cox (Reference Cox1945) has pointed out, it was not just lynching itself that enforced white dominance, it was also the threat of lynching. That is the point of performative violence: to send a message. In this sense, it is possible to understand Confederate monuments in the post-Reconstruction South as communicating the credible threat of public violence. This interpretation also helps explain the effect of monument dedications on Black out-migration presented in Section F.9 of the SM and in Ferlenga (Reference Ferlenga2023). Du Bois, for instance, understood the meaning of monuments quite well. In a 1931 article in The Crisis about a trip to the South, he wrote quite clearly (albeit acerbically) about how he understood monuments:

To many Southern whites, on the other hand, Confederate monuments represented an affirmation of justly-restored dominance. Depending on who you were, Confederate monuments’ central message was intimidating or reassuring. But the reason monuments worked in creating a “psychology of caste” is that both Black and white populations were mutually aware of each other’s receiving of this same message. I argue that the intersubjective dimension is an essential part of monuments’ ability to attenuate racial threat among whites.

While previous research has shown that perceived status reversal or changes can lead to greater violence (Byun and Kwon Reference Byun and Kwon2025; Cikara, Fouka, and Tabellini Reference Cikara, Fouka and Tabellini2022; Petersen Reference Petersen2002; Testa and Williams Reference Testa and Williams2025), to my knowledge, far fewer studies have examined the reverse: whether status-affirmation attenuates violence. In the psychology literature, affirmation interventions have largely been studied as a means of encouraging a sense of social belonging among marginalized individuals (Tay and Cohen Reference Tay and Geoffrey L.2025). I am unaware of any studies that directly test affirmation strategies among dominant groups and their effect on support for violence. A recent study about perceived victimhood finds that it mediates support for violence in the US (Armaly, Buckley, and Enders Reference Armaly, Buckley and Enders2022). Future research should explore the specific ways that the substitution mechanism works psychologically, likely requiring longitudinal designs.

In addition, although it is difficult to directly test, it is possible that the mechanisms of reduced white motivation and Black out-migration reinforced one another. The presumed logic is that as monuments went up, Black citizens fled, while those who remained may have adopted a myriad of behaviors to avoid violence. While this interpretation is speculative, it seems plausible that Confederate monument dedications altered the behavior of Black Americans, leading them to be more deferential and circumspect around their white neighbors, as they too internalized the reification of a white supremacist social order.Footnote ¹⁹ Whites who observed these types of avoidant behaviors by their Black neighbors likely interpreted them as further confirmation of the entrenched racial hierarchy in which they were dominant. Future research could investigate these possible mechanisms, along with the intergenerational transmission mechanism, further.

A limitation of the present study is that it only focuses on Confederate monuments and does not explicitly take into account other political symbols or other activities by Confederate heritage organizations, such as attempts to influence school curricula, the display of flags, and the naming of roads and public buildings after Confederate leaders. If the theory presented here is correct, then these activities should amplify the effects of Confederate monuments. In this sense, focusing solely on monuments likely represents a hard test for the theory of political symbols. To extend the iceberg metaphor used earlier, we should expect that the more of the iceberg is exposed, the more it should change our behavior. In other words, monuments—and political symbols more broadly—are an important heuristic people use to construct mental models of their own structural conditions. In addition, the study focuses on performative, even spectacular, violence as an outcome, and does not account for more quotidian forms of violence or the wide range of indignities and forms of “structural” violence that were integral to the Jim Crow era. Future research could investigate these other important potential causes and outcomes.

I also report results that the removal of Confederate monuments in the present day increases the likelihood of anti-Black hate crimes. The Removal Hypothesis offers evidence that removing Confederate monuments signals that the present social order is under threat, thereby increasing the motivation to engage in anti-Black hate crimes in order to protect the status quo. This finding shows how the same political symbols may have different effects on political violence depending on the time period. At the same time, it should be noted that the results indicate a relatively short-lived backlash. It is possible that a brief violent backlash is an acceptable cost for longer-term changes to the social order. In addition, there is some evidence that monument removal is not associated with increased state repression and several scholars have written about the value of “reclaiming” public and symbolic space, via monument removals, legislation, as well as the construction of alternative spaces for memorialization (Davies Reference Davies2024; Larrabee Reference Larrabee2022; Turner and Crabtree Reference Turner and Crabtree2021).

This also leads us to consider the implications of political symbols for the second Trump administration. In its short time in office, the second Trump administration’s actions appear squarely aimed at reshaping public conceptions of the social order in the US. These include the glorification of Confederate figures through gestures such as renaming military bases after Confederate generals, concerted efforts to control historical narratives by banning books and reshaping cultural institutions, and threats to withhold federal funds in order to compel the removal of symbols like Black Lives Matter Plaza in Washington DC that contradict this new order. In addition, new proposals to build monuments, such as carving President Trump into Mount Rushmore, and renaming Dulles Airport after him, have emerged among the President’s supporters (Jacobs and Svirnovskiy Reference Jacobs and Svirnovskiy2025).

Most striking, the March 27, 2025 executive order on “Restoring Truth and Sanity to American History” mandates the reinstatement of monuments and condemns what it calls “a concerted and widespread effort to rewrite our Nation’s history, replacing objective facts with a distorted narrative driven by ideology rather than truth” (United States Government 2025). These actions suggest that the administration understands, and cares about, the value of public acknowledgment and submission, to echo Du Bois’ aforementioned comments about the “psychology of caste.” It appears that a core part of the administration’s agenda is the generation of common knowledge about who belongs and who is excluded, who is dominant and who is subordinate. All of this is consistent with extant findings that highlight group status threat as a core pillar of electoral support for President Trump (Mutz Reference Mutz2018; Parker and Lavine Reference Parker and Lavine2025).

What this means for future political violence is uncertain. However, persistent contestation over political symbols could lead to highly unstable perceptions of status and belonging, which could in turn lead to increases in performative violence. It is conceivable that repeated failures to reshape the social order via symbolic means might be replaced with the use of performative violence to achieve the same end. The fact that cabinet-level officials, such as Homeland Security Secretary Kristi Noem, have already shown a willingness to engage in displays of performative cruelty in order to demonstrate their view of who does not belong in the US bodes ill (Brandon and Santana Reference Brandon and Santana2025).

While it is beyond the scope of this article to test the theory in other cases, future research could expand beyond the US. Some scholars are already pushing in this direction. For instance, Balcells and Voytas (Reference Balcells and Voytas2025) show the limited effects of visits to a museum exhibit in Northern Ireland. Another promising future direction is to apply the theory to religious sites, which often serve as political symbols and potential flashpoints for violence. For example, the 1992 razing of the Babri mosque in India was followed by sustained Hindu–Muslim riots, while the recent construction of a Hindu Temple on top of the former mosque has transpired without any violence. This discrepancy can potentially be explained by the solidification of a Hindu nationalist status quo under Prime Minister Narendra Modi. Similarly, Jerusalem’s Temple Mount/Al-Aqsa Mosque has long been a trigger for Israeli-Palestinian violence, especially during times when the set of informal yet delicate religious arrangements that govern the site (called the status quo) are perceived as threatened or changing (as they are at the time of writing). More generally, while the specific meanings of political symbols are context-specific and important to understand, my theory suggests that their underlying messages are parsimonious: they concern status hierarchies or group boundaries. The approach offers a flexible conceptual framework to the study of political symbols across cultures and opens the door to future research on the relationship between political symbols and political violence.