Introduction
How does exposure to Confederate commemorations affect citizens? In recent years a number of salient events have put monuments and other commemorative elements at the center of the political debate. Examples include the Black Lives Matter (BLM) movement that advocated for the removal of Confederate memorials; or far-right rallies like the one in Charlottesville 2017 that protested against such removal.
Recent studies have looked into the effect of decisions made on this type of contested symbols – typically their removal (Rozenas and Vlasenko, Reference Rozenas and Vlasenko2022; Rahnama, Reference Rahnama2025; Villamil and Balcells, Reference Villamil and Balcells2021; Charnysh and Nikiforova, Reference Charnysh and Nikiforovan.d.; Wright et al., Reference Wright, Cheung and Esses2019).Footnote 1 Most of this literature shows that removing contested objects may lead to backlash (Rozenas and Vlasenko, Reference Rozenas and Vlasenko2022; Villamil and Balcells, Reference Villamil and Balcells2021; Charnysh and Nikiforova, Reference Charnysh and Nikiforovan.d.; Wright et al., Reference Wright, Cheung and Esses2019; but see Rahnama, Reference Rahnama2025).
However, if the question we seek to investigate is whether governments should remove contested symbolic objects or not, then the answer depends on the effect of not intervening. Even if removing objects leads to backlash, the best course of action can only be decided upon understanding the cost of the status quo.
I argue that non-intervention can have detrimental effects because standing symbolic heritage can operate as a signal of the beliefs of a community. Even in the absence of official action such as protecting monuments (Britt et al., Reference Britt, Wager and Steelman2020), or removing them (Rozenas and Vlasenko, Reference Rozenas and Vlasenko2022), symbolic objects located in public, governmental grounds are used to infer the hegemonic worldview in a given location. Their presence suggests that, in debates about whether these objects are legitimate for public display, authorities – and the residents that elect them – deem that they are. As such, the meaning that the observer attributes to the symbol is likely to be transferred to the community and its members. Following this argument, I look at the effects of Confederate memorials not only on the individual who sees them, but also on their formation of second-order beliefs about the people and locations associated with the Confederate monuments.
While understanding this question is crucial for policy, empirically assessing it poses a number of challenges. On the one hand, observational studies would struggle with identification issues stemming from self-selection. On the other hand, an experimental study would have to envision a design that manipulates exposure to Confederate monuments without engaging in deception. Moreover, it is challenging to pinpoint the effect of commemorative objects themselves, rather than the effects of decisions on them as reported by the media (Wright et al., Reference Wright, Cheung and Esses2019; Britt et al., Reference Britt, Wager and Steelman2020).
I propose an empirical strategy that deals with these issues, relying on a survey experiment that administers the treatment through the display of a video. Leveraging this empirical strategy, this paper is able to provide evidence of the signaling mechanism – inferences that observers make about the town and its residents based on Confederate commemorations – and to show the impact of Confederate commemorations isolating the effect of such objects.
Confederate States of America and Its Commemorations
The legal status of slavery caused divisions between Northern and Southern states of the United States for decades, becoming a pressing concern to the integrity of the Union by the late 1850s. Following the election of President Abraham Lincoln in 1860, seven slaveholding Southern states seceded from the Union. Lincoln had opposed the expansion of slavery to new states and territories of the United States, which was seen by the Southern states as inevitably leading to the abolition of slavery. The seven seceding territories formed the Confederate States of America, and fought the Union in the 1861 American Civil War. The Confederacy was later joined by four more seceded states.
Confederate commemorations are symbolic elements aimed at celebrating the Confederacy and the events and people associated with it. There are many types of commemorations, from school names, parks, and military bases, to statues and monuments. While such commemorations have been built since the American Civil War in 1861, their proliferation took place in two main waves. The first dates from 1898 to 1918, with its peak around 1910. Importantly, the second wave came about a century after the Civil War, and coincided with the Civil Rights movement period (1954–68) (SPLC, 2019; Acharya et al., Reference Acharya, Blackwell and Sen2016), which aimed at abolishing racial segregation and discrimination. According to research by the Southern Poverty Law Center, there are at least 1,700 Confederate symbols in the United States, 800 of which are monuments (SPLC, 2019).
Given the Confederacy’s defense of white supremacy and slavery, Confederate symbols are mnemonically associated with racism (Strother et al., Reference Strother, Piston and Ogorzalek2017). The American Historical Association has stated that ‘memorials to the Confederacy were intended … to intimidate African Americans politically and isolate them from the mainstream of public life’ (Henderson et al., Reference Henderson, Powers, Claibourn, Brown-Iannuzzi and Trawalter2021, 2). Many sectors of the population, from activists to public officials, have removed – or attempted to remove – a number of Confederate commemorations. Opposing political actors have argued for the maintenance of the objects, on the grounds that they represent historical heritage tied to Southern identity and pride. Some states have issued laws to protect all symbolic heritage, which includes Confederate monuments.
Experimental Design
In order to identify the signals and effects of Confederate commemorations I rely on a survey experiment via an online survey. Respondents were recruited using Prolific, and the survey was implemented through Qualtrics’ survey tool.
Respondents were compensated for their time an average of $12.71 per hour,Footnote 2 above the country’s minimum wage ($7.25). Data collection took place from 10 June to 9 July 2021 (N 1,537). The final sample consists of 1,406 observations.Footnote 3
The Sample
Using Prolific’s pre-recorded data, the experiment was block-randomized on party ID (Democrat/Republican), race (Black/White), being from the South (or from outside), and gender (male/female). The descriptives of the sample are shown in Table 1. While acknowledging that a representative sample of the American population would have advantages, for logistic constraints the design focuses on Black people and White people only. The main controversy around Confederate commemorations lies in their relationship with slavery and white supremacy. Additionally, recent scholarship demonstrates that Confederate commemorations are positively associated with number of lynching victims (Henderson et al., Reference Henderson, Powers, Claibourn, Brown-Iannuzzi and Trawalter2021). It should be noted, however, that the choice to focus on a non-representative sample puts the emphasis on internal validity at the expense of reducing external validity. This sample allows me to understand how White and Black Americans respond to the treatment, and how those responses differ by race. This is only possible by oversampling the Black population, even if that means that the effect sizes cannot be directly extrapolated to the American population as a whole.Footnote 4
Sample characteristics
The Town and Video
The treatment was delivered within a video that gives a tour of a town. Using professional video-editing tools, I created a video that mimics a touristic video presenting a real location in the United States: Bishopville, South Carolina – although this information is never given to respondents. However, given that Confederate commemorations are far more likely to exist in the South versus other regions of the country, both treatment and control videos emphasize that the town is located in the American South (see Figure D.1, online Appendix D).
This town was chosen for three reasons. First, neither the town nor the Confederate commemorations are particularly well-known or recognizable. In support of this point, only 0.2 per cent of respondents can identify its name.Footnote 5 Second, Bishopville has several common types of Confederate remains still found in the United States: schools, parks and trails, and monuments on a Courthouse ground (SPLC, 2019). Third, the placement of commemorations in Bishopville enabled me to show them in a seemingly incidental manner. That is, the focus of the video is to display the town, and Confederate commemorations are shown as part of the scenery when visiting the main areas of the locality.
Treatment and Outcome Variables
Videos in treatment and control groups were virtually identical, except for four Confederate commemorations that appear in the video of the treatment group. Three such commemorations are places that have been named after a well-known Confederate general: Robert E. Lee. In these cases, the image in the background remains identical in both treatment and control groups, and the treatment is introduced via a label that shows the naming commemoration. For instance, while seeing the same image of a park, the control group saw the label ‘Park’, while the label of the treatment group read ‘Lee Park’ (Figures 1, 2, and 3). Such label resembles those used in documentaries or news broadcasts to give the audience context. The idea is that, on the field, observers would also learn about the Confederate legacy of an object by reading similar signs – those on the object itself.
The fourth commemoration is a monument to Confederate soldiers and the Confederate States of America, which is located in front of a courthouse. In the control group, the image shows the courthouse only, and the label reads ‘Courthouse’. With a minimal change to the angle from which the picture is taken, the treatment group sees both the courthouse and the Confederate memorial. Additionally, given that the video setting makes it difficult to read the inscriptions in the monument the way that a bystander could, a sub-label in a smaller font was added to describe the existence of the monument (Figure 4).Footnote 6
The video lasts 63 s. In 27 of these, one of the treatments is being shown. Images of the video used screenshots taken from Google Street View out of the principle of Fair Use. The pictures corresponding to the courthouse were taken by Ammodramus (2013a,b). All frames of treatment and control videos can be found in the online Appendix D. The full videos were submitted as additional material along with the manuscript.
Control and treatment 1.
Control and treatment 2.
Control and treatment 3.
Control and treatment 4.
After this treatment, respondents were asked three clusters of outcome variables. The first taps what Confederate commemorations signal, inquiring about second-order beliefs about the town and its residents. This section includes two placebo outcomes: how much residents of the town care about parks (a park appears in the video) and about having big parking spaces (does not feature in the video). Following previous studies on the topic (Wright et al., Reference Wright, Cheung and Esses2019), the second cluster measures the emotional effect of exposure to commemorations. The third cluster explores respondents’ attitudes and relationship to the town in the video. This includes a quasi-behavioral outcome to do away with social desirability bias concerns that may stem from standard survey items. Additionally, it seeks to identify potential economic costs of having Confederate symbols, as their hypothetical cost has been shown to make public officials more likely to favor their removal (Grose and Peterson, Reference Grose and Peterson2020). All outcomes and the way they are measured are summarized in Table 2.
Outcome variables
Results
The first block of the experiment examines what Confederate commemorations signal; that is, what information observers gather from their presence in the town of the video.Footnote 7 To allow for comparison, all variables in this section have been standardized.Footnote 8 The analyses in Figure 5 (and Table A.3 in the online Appendix) show that towns with Confederate commemorations signal that their history has had conflict, that its citizens support the secession of their state from the United States of America, that its citizens are proud to be from the South, and that the majority of its citizens vote for the Republican Party. Most importantly, the presence of Confederate monuments signals that the residents of the town support BLM to a lesser extent, and care about racial equality less.
The direction of the treatment effect on these signals-related outcomes is similar across respondents’ race and party ID.Footnote 9 There are, however, some differences in the size of the effects, which are larger for Black- and Democrat-identifying respondents. These additional analyses can be found in Figures A.1 and A.2 in the online Appendix. One relevant exception is that Republicans do not believe that residents care less about racial equality.
The effects are not only statistically significant; their size is also substantially meaningful. Using the non-standardized version of the variables, the effects range from 8 per cent of the baseline (increase in the belief that the town votes for the Republican Party) to 39 per cent of the baseline (increase in the belief that the town’s history is full of conflict). To take a signal that is easy to interpret: respondents in the control group believe that 39.2 per cent of the citizens in the town of the video support the BLM movement. Being exposed to Confederate commemorations reduces that number by 9.5 percentage points, which constitutes a decrease of 24 per cent when compared to the control group.
I find no effect for the placebo variable that is shown on the videos – the importance of parks – in any of the models.
It is important to note that, while respondents’ perceptions of support for BLM in the town are significantly reduced in the presence of Confederate monuments, their personal support for BLM remains unaltered. This finding is particularly revealing, as it suggests that the results are not driven by social desirability bias. If respondents were reporting the town’s lower concerns for racial discrimination because they believed that to be the social expectation, one would expect to find an overreporting of personal support for BLM alongside it.
What do Confederate commemorations signal about the town and its residents?
Note: analyses in table format may be found in Table A.7 in the online Appendix.
The second part of the experiment focuses on the effects of Confederate commemorations on individual emotions and attitudes. Exposure to a town with Confederate commemorations increases negative emotions, and reduces positive ones (Figure 6; Table A.4 in the online Appendix). Observers exposed to the video showing Confederate commemorations also trust town inhabitants less, believe that they would be less valued in such communities, think people like them who live in the town have less of a say in what the government does, and find the town a worse place to live overall.
I also find effects on a quasi-behavioral outcome included in the experiment. Respondents were offered the opportunity to participate in a raffle, and to donate their preferred amount to one of the schools in the video. I find that observers in the treatment group are less inclined to make a donation, and they would donate less money.
Once again, these constitute effects of great magnitude. The model with non-standardized variables shows effects that range from 7 per cent of the baseline (reduction in external political efficacy) to 22 per cent of the baseline (amount respondents are willing to donate to schools). To take some examples that are of easy interpretation, being exposed to Confederate commemorations reduces potential donors from 24 per cent to 20 per cent of respondents (17 per cent decrease). Additionally, respondents are on average willing to donate $2.9 less to schools in the town with Confederate commemorations, which constitutes a 22 per cent reduction from the baseline ($13.4). Finally, on a scale from 0 to 10, citizens of the town without Confederate symbols receive a score of 6.3 in trustworthiness, whereas citizens in the town with such symbols receive a score of 5.1 (19 per cent decrease).
Effects by race resemble those of the main effect, although, importantly, both the emotional and attitudinal effects are larger for Black respondents. Strikingly, the greatest heterogeneity is found across party lines. The effect for Democrats is similar to the main analysis. However, the emotional effect is null across the board for Republicans and, if anything, exposure to Confederate commemorations is positive; they believe they would be better valued in the town, and they find they would have more of a say on what the local government does.
A potential concern with the overall results is that I estimate the effect of exposure to Confederate memorials for outsiders to the town (for example, potential investors or tourists watching the promotional video, etc.), and not locals. This could pose a concern if outsiders reacted differently, and more strongly, to the memorials than locals, as then my results would constitute an upper bound. Further analyses suggest otherwise. Presumably, if the effects were different between locals and outsiders, this would be likely due to a sense of belonging to the town that holds the monuments and, therefore, to the monuments it contains. I replicate the analysis with states with a strong Southern identity and I do not find that the effect is stronger for the rest of the states. If anything, the opposite seems to be true for the effect on attitudes, which might indicate that signals given by Confederate symbols are used to a greater extent by locals. This analysis and an extended discussion on the topic are available in the online Appendix E.
Discussion
Does it matter which symbolic elements are displayed in public spaces? Using a block-randomized survey experiment in which respondents observe images of the same town with (treatment) and without (control) Confederate commemorations, I show that respondents’ beliefs, emotions, and attitudes are significantly affected by these objects.
Individuals who are exposed to Confederate commemorations believe that the town in which those commemorations are placed has a more conflictive history. They also believe that the people in the town have a greater sense of Southern pride, that they vote for the Republican Party to a greater extent, and that they are more willing to secede from the Federal Government. Most importantly, Confederate commemorations signal that town residents support BLM to a lesser extent, and that they care about racial equality less.
These signals are not inconsequential; exposure to Confederate heritage increases negative emotions, makes respondents trust the citizens in the town to a lesser extent, and makes them feel they would be less valued in that community. Respondents also believe that the local government is less responsive to demands of people like them and that, overall, the town is a worse place to live. Finally, exposure to Confederate commemorations decreases the willingness to donate to one of the schools in the town.
The greatest heterogeneity in the effect of Confederate commemorations is found across party lines. This is striking considering that race constitutes a defining feature of American politics. And even more so considering that the commemorations themselves are controversial for their association with racial discrimination, slavery, and lynching. The fact that party ID is the strongest moderator suggests that in a context of great polarization, partisanship is able to coopt topics that are directly associated with race.
In recent years, discussions of whether to remove controversial symbolic objects have come to the forefront of the political debate. Previous research has shown that their removal may generate backlash. Such findings constitute an argument against the removal of these objects. However, the findings of this paper urge caution with this conclusion. Even if removal may lead to backlash, there is an argument to remove such objects because the status quo is not neutral: keeping these objects also has negative effects on individuals that are exposed to them. Moreover, as shown in the analyses of heterogeneity, exposure to these monuments can increase the already very high polarization of American society.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0007123425100641.
Data availability statement
Replication data for this article can be found in Harvard Dataverse at: https://doi.org/10.7910/DVN/KMCZ2Z.
Acknowledgements
I am very grateful to Elias Dinas, Irene Martín Cortés, Miguel Pereira, Ellen Immergut, Vicente Valentim, Annabelle Wittels, all participants of the Sparks Colloquium, Levi Littvay, Carles Boix, and Vicky Fouka.
Financial support
This work was supported the European University Institute and the Salvador de Madariaga Programme of the Spanish Ministry of Education, Culture and Sports [Grant number : 998759].
Competing interests
None.
Ethical standards
The research was reviewed by the Ethics Committee at the European University Institute. It received clearance as complying with the rules, norms, and values of the ‘EUI- Code of Ethics in Academic Research’. A detailed discussion of the ethical considerations may be found in Appendix C.
Open access
