This is an open access article distributed under the terms of the Creative Commons
Attribution License (Stemming from the murder of nine African-Americans in the United States by a self-proclaimed White supremacist in June 2015, political pressure mobilized to ban the Confederate battle flag on state capitol grounds in South Carolina (Clark, 2015; Jones & Barrett, 2015). This mobilization was intricately tied to reactions toward photos of Dylann Roof holding the Confederate battle flag as one of many symbols representing White supremacy (Robles, 2015). Despite a democratic vote in the year 2000 confirming the flag’s position on state capitol grounds in South Carolina, the South Carolina Senate quickly passed a bill to remove the Confederate battle flag from the grounds of the state capitol in July 2015 (Blinder, 2015). A few days later, the South Carolina House voted 94-20, approving the removal of the flag (Blinder & Fausset, 2015). In response to this removal, and due to political pressure in Florida, Florida’s Marion County voted to reinstate the Confederate battle flag to a government-building complex. Dennis Baxley, a state representative of Florida, stated, “we are all exposed to messages and symbols that may not connect for us, but we should all honor our ancestors and protect free expression” (Guardian staff and agencies, 2015). In contrast, Darrell Jackson, a black Democrat, stated, “When I see a Confederate soldier, I don't get goosebumps and get all warm and fuzzy” (Reuters, 2015). As political pressure from supporters and opponents of the Confederate battle flag extends throughout the South, whether Confederate symbols should be removed from public grounds is being discussed. One useful approach to informing this debate involves understanding the motivations behind support for the flag. We focus on two primary perspectives: First, the group-based dominance perspective that racism is the predominant driver to Confederate battle flag support, and second, the social identity perspective that Southern pride is the predominant driver.
It should be noted from the outset that our primary concern is providing empirical data to compare these two socially and politically relevant perspectives, which largely differ along a racial divide (Agiesta, 2015). The dichotomy of Southern pride versus negative racial attitudes is certainly a false dichotomy in that the reality of Confederate battle flag support may contain elements of both, or neither, and we do not deny the theoretical openness to other predictors. However, this dichotomy is used to frame the practical discussion and history of thought surrounding Confederate battle flag support. The importance of applied social and political psychology is in answering questions for the societies that pose them, not in setting up unnecessary complexity that fails to respect the current and historical thinking of real people in the Southern United States. However, we are not constrained theoretically or empirically by this historical and politically motivated dichotomy, allowing for alternative explanations such as the principled-conservatism perspective.
Historical Perspective: Southern Pride or Racism?
Some symbols can both invoke collective pride for the ingroup, and invoke shame or humiliation for the outgroup (Thornton, 1996). This is arguably the case for the Confederate battle flag in the Southern United States. Two predominant meanings have been ascribed to the Confederate battle flag. For some, the flag is a symbol of a Southern heritage predicated on heroism and sacrifice (Webster & Leib, 2001), inducing pride in a mythic South and remembrances of past glory (Clark, 1997). Often combined with this view is the position that groups in opposition to Confederate symbols are engaged in a cultural genocide and revisionist history, ultimately amounting to anti-Southern bigotry (Woliver, Ledford, & Dolan, 2001). The alternative perspective is that the Confederate battle flag is a symbol of hate, amounting to a remembrance of a history of slavery, racism, and White supremacy (Clark, 1997). Within this view, support for the Confederate battle flag is primarily driven by negative racial attitudes.
These two conflicting views are reflected in a 2015 CNN poll (Agiesta, 2015). When asked if the Confederate battle flag is a symbol of Southern pride or a symbol of racism, White respondents predominantly suggested the former, while Black respondents predominantly suggested the latter. While it is historically accurate that the Confederate battle flag has been used as a symbol of hate and racism at various points in history, including by the Ku Klux Klan throughout the civil rights era (Newman, 2007; Thornton, 1996), the current motivations for supporting the flag should be the driver of discussions related to the flags’ removal across the South. It is instructive to consider a brief historical context for the Confederate battle flag before engaging in a theoretical and empirical overview.
The Confederate battle flag, often incorrectly labeled the “Confederate flag”, played only a small role during the American Civil War. In fact, it was not a “Confederate” flag at all. Rather, it was instituted as a battle flag for the Northern Army of Virginia to reduce confusion due to the similarity between the “Stars and Bars” (i.e., the first national flag of the Confederacy) and the “Stars and Stripes” (i.e., the U.S. flag). The cross itself was adopted from Scotland’s St. Andrews Cross—a symbol of the Christian apostle and martyr. In later iterations of the national flag of the Confederacy, the battle flag emblem was inserted into the top left corner, similar to the current state flag of Mississippi. As Thornton (1996) suggests, “Today’s Confederate symbols are not artifacts of the Confederacy, but rather artifacts of post-Reconstruction Confederate revivals” (p. 237). For some, these revivals commemorated the heroism of the sons and daughters of the Confederacy (Thornton, 1996), whereas for others, specifically during the civil rights era, these revivals marked a resistance to integration and cultural racism (Newman, 2007). As Thornton (1996) summarizes, “To read the Confederacy solely in terms of slavery is to create a counter-myth to the Lost Cause, and to cast Southern history solely in the satisfying, though inaccurate, terminology of good and evil” (p. 241).
The current wave of frustration over the Confederate battle flag is not the first. It was a deep political issue during the civil rights era, in the early 1990’s in Georgia, and in South Carolina and Mississippi in the early 2000’s. It ultimately stems from divisive reactions to the Confederate battle flag among Whites and Blacks in the Southern United States (Leib, 1995; Reingold & Wike, 1998), which are linked to two general arguments regarding support for the Confederate battle flag: the Southern heritage argument and the racial ideology argument.
Southern Heritage Perspective
The Southern heritage perspective, that support for the Confederate battle flag stems from a deep pride in the culture of the Southern United States, can be discussed best in terms of social identity theory (Hornsey, 2008; Tajfel & Turner, 1986; Turner, 1982). We link Southern pride to Confederate battle flag support, independent of negative racial attitudes, by suggesting that the group’s pride is imbedded within its symbolism, and support for this symbolism is a mechanism for exhibiting one’s positive distinctiveness. Support for Confederate symbolism may also link to Southern pride through a motivation to symbolically protect the idea of a glorious history in the Southern United States, which exhibits positive representations of the ingroup (Sibley, 2010). Within this view, the removal of Confederate imagery represents an affront to this history and a threat to the positive representations of the ingroup. However, we suggest that Southern pride as related to the Confederate battle flag is not necessarily linked to outgroup animosity (Brewer, 1999), and is thus similar to the concept of patriotism (Kosterman & Feshbach, 1989). It can exist as an entity devoid of dominance and ethnocentrism.
We see “Southerness” as a reference group, as a way to orient oneself to the rest of the United States. It encapsulates a “healthy national self-concept” (Kosterman & Feshbach, 1989, p. 228) driven by affect for one's regional space, in contrast to the feeling of dominance and superiority exemplified by nationalism. Thus, our concept of Southern pride embodies both cognitive and affective space. Regional affiliation research (Reed, 2008) suggests that geography of space acts as an important frame for Southerness, and we embed the importance of geography in the life space of individuals within our concept of Southern pride, which ultimately is defined by the cognitive act of being, the affective admiration for place and entity, and the geography of space and history.
Racial Ideology Perspective
The racial ideology perspective is best grasped by a group-based dominance perspective, which suggests that negative racial attitudes should drive opposition to race-targeted social policies because of a belief in societal group-based hierarchies and a desire to preserve the group’s position within the hierarchy (Federico & Sidanius, 2002). While empirical work does suggest that negative racial attitudes can drive opposition to race-based policies such as affirmative action (Federico & Sidanius, 2002; Sidanius, Pratto, & Bobo, 1996), this can often be construed as a misrepresentation of what affirmative action is. As an example, Whites are generally opposed to racial preference programs, not affirmative action as it was originally construed as outreach to increase the pool of qualified Black applicants (Kuklinski et al., 1997; Sander & Taylor, 2012). The group-based dominance perspective suggests that negative racial attitudes should be related to Confederate battle flag support out of a desire for Whites to maintain their privileged position in the United States and reaffirm the historical hierarchy through symbolism.
The social identity perspective and the group-based dominance perspective from the social psychological literature help explain the colloquial Southern heritage perspective and racial ideology perspectives respectively. However, other politically relevant dimensions are likely to impact support for the Confederate battle flag.
Principled-Conservatism Perspective
The principled conservatism perspective suggests that individual preferences for or against race related policies, such as affirmative action, are primarily driven by political ideology (Federico & Sidanius, 2002). For example, Sniderman and Carmines (1997) suggest that opposition to affirmative action is driven by ideological beliefs about fairness and individualism. While the principled conservatism argument does not relegate racial attitudes to complete abandonment as a cause of attitudes toward various race-targeted policies, evidence does suggest that negative affect toward Blacks is only weakly related to opposition to race-targeted policies (Sniderman, Piazza, Tetlock, & Kendrick, 1991) and Whites actually do support extra effort made by universities to ensure admittance to qualified Blacks (Kuklinski et al., 1997). More recent research suggests that political ideology is the best predictor of opposition to universal health care (Study 1; Shen & LaBouff, 2016) and individuals only increase opposition to universal healthcare in response to perceptions of free-riding but not race (Study 3; Shen & LaBouff, 2016).
Empirical Studies Addressing Attitudes Toward the Confederate Battle Flag
The current study is largely driven by the theoretical positions of social identity theory, patriotism, group-based dominance, and principled-conservatism. The efforts to link these to the broader scope of social issues surrounding the Confederate battle flag is an extension of the broader work regarding race-targeted policies. While some empirical work has been conducted specifically on Confederate battle flag support, it is minimal and that which has been done suffers from methodological flaws and failure to account for the alternative theoretical perspectives.
Racial division in attitudes toward the flag is exemplified through an analysis of the 1997 referendum allowing specialty license plates with the Confederate battle flag (Webster & Leib, 2001). Representatives of districts with low African American populations overwhelmingly voted in favor, while representatives in districts with high African American populations voted overwhelmingly against. In an analysis of attitudes toward the Georgia state flag, Clark (1997) noted that ideological, racial, and traditionalist attitudes contribute. While his findings implicate racial attitudes as the strongest predictor, the use of a single item measure of racial attitudes is prone to measurement error (Diamantopoulos, Sarstedt, Fuchs, Wilczynski, & Kaiser, 2012). Additionally, the item assessing racial attitudes asked about support for social and economic aid to Blacks in America, and it is not clear that this item is optimal for accurately measuring racial attitudes in general.i In a subsequent study regarding attitudes toward the Georgia state flag, Reingold and Wike (1998) determined that the effect of racial attitudes outweighed the effect of Southern identification. Again, the measures are potentially problematic. The predictors for Southern identification and racial attitudes were not measured on a continuous scale, nor was evidence of reliabilities provided. The measure of racial attitudes only addressed attitudes toward the speed of civil rights progress. The Southern identification index was adapted from Reed (1982), who found no evidence of a relationship between Southern identification and attitudes toward segregation, and even questioned the precision of the measure.
Attitudes toward the speed of civil rights progress might be termed “symbolic” or “modern” prejudice (Jackson, 2011), but are an indirect measure of racial attitudes which may be affected by attitudes toward traditionalism or the status quo. In an effort to expand on this terse literature within the context of the Mississippi state flag, Orey (2004) attempted to distinguish between “old-fashioned” racism — racism that expresses a biological superiority of the White race — and “symbolic racism” – a newer, more subtle and indirect form of racism. No longer being defined by biological differences, symbolic racism is defined by a belief that Blacks violate social norms of work ethic and discipline (Kinder & Sears, 1981). Orey (2004) determined that attitudes toward the Mississippi state flag were primarily determined by old-fashioned racism, and to a lesser extent by symbolic racism and political ideology. However, like all studies previously, Orey (2004) failed to account for Southern pride as a possible factor in support for the flag at all.
While poor conceptualization and measurement problems plague some of the early studies regarding attitudes toward the Confederate battle flag, another, more obvious problem is the lack of comparison of White and Black attitudes. While polling shows a racial divide in these attitudes, no studies have assessed the potential moderation of race on how Southern pride, conservatism, or racial attitudes toward the other may be related to Confederate battle flag support. Cooper and Knotts (2006) present a more complex picture of the flag debate, with specific reference to the Confederate battle flag in South Carolina. In a national survey sample of both Black and White Southerners and non-Southerners, this research suggested that conservative ideology, being from the South, and being White, were highly related to supporting the position of the Confederate battle flag on South Carolina’s capitol building. Regional and race differences existed such that White Southerners were most likely to support the flag, followed by non-Southern Whites. Black Southerners expressed the most opposition to the flag followed by non-Southern Blacks. Black and White Southerners may have different interpretations of what “Southern” means and have different perceptions of the South’s historical past (Thornton, 1996; Wilson, 2007), which could lead to different attitudes toward the Confederate battle flag. As Webster and Leib (2001) note, “black and white Southerners arguably constitute two separate ‘nations’ occupying the same space” (p. 288).
While racial attitudes may be an important predictor of attitudes toward the Confederate battle flag (Clark, 1997; Cooper & Knotts, 2006; Orey, 2004; Reingold & Wike, 1998), this relationship has not been tested against the competing view (e.g., the Southern heritage view), nor have the data been compelling. The effects of Southern identification are less clear (Clark, 1997; Reingold & Wike, 1998) and whether negative racial attitudes continue to be a predictor of attitudes toward the Confederate battle flag when Southern pride is considered has yet to be tested at all, despite these two perspectives being the primary arguments for and against the position of Confederate symbols throughout the Southern United States. Do White Southerners who support the flag do so because they are proud of their Southern heritage or because they hold strong negative racial attitudes toward Blacks? What about Black Southerners who support the flag? Could this also be driven by a strong pride in the South? What about the claim that Black opposition to the flag might be due to anti-White racism (e.g., Woliver et al., 2001)? Based upon the prior literature, among Whites we expected negative racial attitudes toward Blacks and a conservative political ideology to correlate with positive attitudes toward the Confederate battle flag; however, in line with the Southern heritage argument, we also expected Southern pride to be correlated with more positive attitudes toward the Confederate battle flag, independently of racial attitudes or political conservatism.
We take a more exploratory view of Black attitudes toward the flag, to consider whether Southern pride or political conservatism also impact Black attitudes toward the flag. Additionally, we consider whether Blacks’ blatant negative racial attitudes toward Whites impacts their attitudes toward the Confederate battle flag, in line with a reverse racism or anti-Southern bigotry perspective (Woliver et al., 2001).
Control Variables
We recognize that people, especially under threat of being perceived as racist, may be inclined to temper their positive attitudes toward a symbol fraught with contention. This may occur through either a motivation to control prejudiced reactions (Dunton & Fazio, 1997) or through a desire to be viewed favorably by society (Stöber, 2001). Thus, we also hypothesize that both motivation to control prejudiced reactions and social desirability may be two constructs negatively related to support toward the Confederate battle flag and we consider these in our theoretical model.
Method
Participants
We used Amazon Mechanical Turk (Mturk) to collect data from 638 people from the Southern United States. We selected Mturk respondents based upon their accounts being registered in South Carolina, Georgia, or Mississippi, three states that have a large percentage of African Americans (approximately 40%)ii and that have had controversy over Confederate symbolism in the recent past. All participants were awarded 50 cents for approximately 15 minutes of their time. We limited data collection to Mturk workers with minimum 95% approval rates to ensure data quality (Peer, Vosgerau, & Acquisti, 2014).
Measures
All participants completed demographic questions, including race, state of birth, state of residence, and the state that the respondent considered their “home” state.
Political Ideology
Political ideology was measured by the single item: “Please select the ideology that you most align with,” rated on a scale from 1 “extremely liberal” to 7 “extremely conservative”.
Racial Attitudes
Based upon participants’ racial self-identification, each participant completed a measure of blatant racial attitudes toward either Blacks or Whites. Whites completed a measure of blatant racial attitudes toward Blacks, while Blacks completed the same measure of blatant racial attitudes toward Whites. We measured racial attitudes via 7-items (e.g., “To live in a neighborhood with Black/White people creates problems,” Blacks, α = .76; Whites, α = .85) adapted from various sources (Brigham, 1993; McConahay, Hardee, & Batts, 1981). Items were rated on a 5-point scale from “strongly disagree” to “strongly agree” and averaged to create total scores. Higher scores indicate more negative racial attitudes.
Southern Pride
We measured Southern pride via three items (e.g., “To what extent do you feel pride in being from the South,” Blacks, α = .75, Whites, α = .90) adapted from multiple sources (Kosterman & Feshbach, 1989; Reed, 2008). Items were rated on a 5-point scale and the endpoints differed for each item, which are reported in the Appendix. Scores were an average of the three items, with higher scores indicating more pride in the South.
Social Desirability
Social desirability was measured via the 16-item social desirability scale (Whites, α = .70; Blacks, α = .65; Stöber, 2001). Items are responded to on a True-False scale and total scores are the sum of True responses. Example items include: “I always admit my mistakes openly and face the potential negative consequences” and “In traffic I am always polite and considerate of others.”
Motivation to Control Prejudice
Motivation to control prejudice was measured via 17 items (Blacks, α = .75; Whites, α = .83; Dunton & Fazio, 1997) on a 5-point scale ranging from “strongly disagree” to “strongly agree”. Example items include: “In today's society it is important that one not be perceived as prejudiced in any manner” and “It's important to me that other people not think I'm prejudiced.” Item responses were averaged to create total scores with higher scores indicating a greater motivation to control prejudice.
Attitudes Toward the Flag
Attitudes toward the Confederate battle flag (ATF) were measured by two items: “To what extent do you support Southern states’ maintaining the Confederate battle flag on government premises” and “To what extent do you support Southern states’ maintaining the Confederate battle flag as a part of the official state flag (e.g., in Mississippi)”. Each was measured on a 5-point scale ranging from “strongly oppose” to “strongly support”. These items correlated well within Blacks (r = .84) and Whites (r = .90) and were combined into a single score. Higher scores indicate more positive attitudes toward the flag.
The Appendix reports full scales for racial attitudes, Southern pride, and attitudes toward the flag.
Results
We eliminated data from 8 participants for missing necessary demographic information and data from 45 participants due to incomplete responding.iii We assessed whether participants fit our criterion of “being Southern” by examining their responses to what state they were from, what state they were born in, and what state they reported currently living in. We excluded the data from 15 participants who did not report South Carolina, Mississippi, or Georgia for at least one of these questions. Following these exclusions, a total of 570 completed surveys were available for analysis.
Responses to ethnicity were then recoded into the following three groups: White, Black, or Other. Participants were considered “White” if “White” was the only option selected. Participants were considered “Black” if “Black” was selected in combination with any other group. Finally, “Other” includes all other ethnic groups in any combination (i.e., non-Black and mixed-White). Because of the small sample size of those labeled “Other” (n = 44), we do not report analyses for this group. As a result, analyses were conducted with 526 respondentsiv. This sample consisted of 417 Whites and 109 Blacks, was predominantly female (n = 380), with a mean age of 35.04 years (SD = 11.49). Political ideology leaned liberal (M = 3.85, SD = 1.65), but there is no restriction of range with political ideology. Scores are normally distributed (skew = .051; kurtosis = -.791), with 34.1% of participants scoring above the midpoint and 39.7% of participants scoring below the midpoint. Both extreme ends of the scale were represented, with 6.1% extremely conservative and 8.6% extremely liberal.
Differences in Southern Pride, Political Ideology, Racial Attitudes, and Attitudes Toward the Confederate Battle Flag as a Function of Race
A series of independent sample t-tests suggest that Whites and Blacks in the South differ significantly on political ideology, t(524) = 3.11, p = .002, g = .34v, with Whites (M = 3.96, SD = 1.68) being more conservative than Blacks (M = 3.41, SD = 1.46). Additionally, Whites and Blacks differ significantly in attitudes toward the Confederate battle flag, t(524) = 8.42, p < .001, g = .91, with Whites (M = 2.96, SD = 1.52) being more supportive of the flag than Blacks (M = 1.66, SD = 1.0). Whites and Blacks did not differ on our measure of Southern pride, t(524) = .827, p = .409 (MWhites = 3.57, SD = 1.20; MBlacks = 3.47, SD = 1.04) or on our measure of racial attitudes, t(524) = .928, p =.354 (MWhites = 1.86, SD = .754; MBlacks = 1.93, SD = .677). Because Whites completed a measure of blatant racial attitudes toward Blacks and Blacks completed the same measure of blatant racial attitudes toward Whites, this compares the blatant negative racial attitudes toward the “other” between both racial groups.
Bivariate Correlations
Bivariate correlations among all variables are presented in Table 1, reported separately for the Black and White samples.
Table 1
Bivariate Correlations, Means, and Standard Deviations Among All Variables for Both White and Black Samples
| Age | PI | SD | MAP | RA | SP | ATF | |
|---|---|---|---|---|---|---|---|
| Age | 1 | .016 | -.048 | .168 | -.091 | -.170† | -.079 |
| Political Ideology (PI) | .140** | 1 | -.001 | -.085 | .085 | .177† | -.027 |
| Social Desirability (SD) | .173*** | .096* | 1 | .254** | -.152 | .182† | -.024 |
| Motivation to Control Prejudice (MAP) | -.012 | -.182*** | .048 | 1 | -.187† | .135 | .031 |
| Racial Attitudes (RA) | .099* | .307*** | -.118* | -.336*** | 1 | -.116 | .056 |
| Southern Pride (SP) | .244*** | .450*** | .078 | -.158*** | .234*** | 1 | .190* |
| Attitude Toward the Confederate Battle Flag | .020 | .546*** | .060 | -.275*** | .427*** | .485*** | 1 |
| αWhite | .70 | .83 | .85 | .90 | .95 | ||
| αBlack | .70 | .75 | .76 | .75 | .91 | ||
| MWhite | 35.49 | 3.96 | 9.63 | 3.21 | 1.86 | 3.57 | 2.96 |
| SDWhite | 11.56 | 1.68 | 3.07 | 0.60 | 0.75 | 1.20 | 1.53 |
| MBlack | 33.31 | 3.41 | 9.69 | 3.10 | 1.93 | 3.47 | 1.66 |
| SDBlack | 11.09 | 1.46 | 3.04 | 0.54 | 0.68 | 1.04 | 1.00 |
Note. Correlations for Blacks are above the diagonal and italicized. Correlations for Whites are below the diagonal and bolded.
The reliability coefficient reported for two-item ATF is the Spearman-Brown coefficient.
†p < .08. *p < .05. **p < .01. ***p < .001.
Among the White sample, positive attitudes toward the Confederate battle flag were significantly related to more political conservatism, more Southern pride, more negative racial attitudes toward Blacks, and less motivation to control prejudice. Positive attitudes toward the Confederate battle flag were more strongly associated with both political conservatism and Southern pride than with negative racial attitudes. Among the Black sample, positive attitudes toward the Confederate battle flag were only related to more Southern pride.
Table 2
Summary of Structural Equation Model Predicting Attitudes Toward the Confederate Battle Flag (N = 526)
| White
|
Black
|
|||||||
|---|---|---|---|---|---|---|---|---|
| Variable | B | SE | β | 95% CI | B | SE | β | 95% CI |
| Age | -.020*** | .005 | -.157*** | -.238, -.077 | .000 | .010 | .001 | -.225, .227 |
| Social Desirability | .090 | .079 | .058 | -.040, .156 | -.087 | .159 | -.085 | -.367, .196 |
| Motivation to Control Prejudiced Reactions | -.234* | .114 | -.089* | -.172, -.005 | .064 | .236 | .032 | -.199, .264 |
| Political Ideology | .279*** | .038 | .325*** | .237, .412 | -.054 | .073 | -.085 | -.310, .140 |
| Racial Attitudes | .553*** | .108 | .259*** | .166, .353 | .217 | .241 | .139 | -.163, .442 |
| Southern Pride | .436*** | .067 | .332*** | .239, .425 | .279 | .191 | .258 | -.062, .579 |
| R2 | .499*** | .067 | ||||||
*p < .05. **p < .01. ***p < .001.
Partial Correlations
Among White respondents, the partial correlation between attitudes toward the Confederate battle flag and racial attitudes remained significant even when controlling for age, political ideology, Southern pride, social desirability, and motivation to control prejudice, r(407) = .285, p < .001. The partial correlation between attitudes toward the Confederate battle flag and Southern pride remained significant even when controlling for age, political ideology, racial attitudes, social desirability, and motivation to control prejudice, r(407) = .322, p < .001. Positive attitudes toward the Confederate battle flag were more strongly associated with Southern pride than with racial attitudes when accounting for these covariates. Additionally, among Black respondents, the correlation between attitudes toward the Confederate battle flag and Southern pride remained significant even when controlling for age, political ideology, racial attitudes, social desirability, and motivation to control prejudice, r = .198, p <.001.
Structural Equation Modeling
In order to compare the Southern heritage and racial ideology perspectives on support for the Confederate battle flag while controlling for the measurement error inherent in regression-based approaches, we utilized a multi-group structural equation modeling approach with parcels as the indicators of latent constructs. We selected a parceling-based approach rather than an item-level approach because our interest is in understanding the relationships between the sets of constructs rather than the structure of the items themselves (Little, Cunningham, Shahar, & Widaman, 2002). Additionally, we utilize random assignment of items to parcels, which, on average, leads to parcels with roughly equal common factor variance (Little et al., 2002). This approach minimizes random error variance, improves stability of parameter estimates, and leads to better fitting models than item-level approaches (Bandalos, 2002; Matsunaga, 2008).
We estimated our model using MPLUS v.7.4 with full information maximum likelihood estimation (FIML) and 5000 bootstrapped samples. FIML estimates a likelihood function, which includes both the complete and missing data to produce parameters with the best probability of reproducing the sample data (Baraldi & Enders, 2010). It estimates means and variances of the missing portions of data based upon the availability of all other data in the model (Wothke, 2000) and produces estimates on par with multiple imputation (Ferro, 2014; Graham, 2003). Bootstrapping was used to obtain confidence intervals for the standardized parameter estimates.
Our model accounted for age, political ideology, racial attitudes, Southern pride, motivation to control prejudice, and social desirability. Model fit was good, CFI = .951, RMSEA = .063 (95%CI: .054, .072), SRMR = .054. Hu and Bentler (1999) suggest combination metrics of the SRMR < .08 combined with either the RMSEA < .06 or the CFI > .95. The chi-square test of model fit is not appropriate because it is a measure of exact fit, determining whether the fit of the model is perfect (Kline, 2011), which always fails in large sample sizes (Steiger, 2007).
As seen in Table 2, the implications of our factors differ between Whites and Blacks. Among Blacks, no factors were significantly associated with attitudes toward the Confederate battle flag. Among Whites our model accounted for 49.9% of the variance in attitudes toward the Confederate battle flag. Being older (β = -.157) and being more motivated to control prejudice (β = -.089) were associated with reduced support for the flag. Increased support for the flag was associated with more conservatism (β = .325), more Southern pride (β = .332), and more negative racial attitudes toward Blacks (β = .259). The effects of Southern pride and political conservatism were the largest effects; however, in order to test whether these effect sizes were statistically different from each other, we compared the bootstrapped confidence intervals of the standardized effects. The confidence intervals for political ideology, Southern pride, and racial attitudes all overlapped, suggesting that the size difference between the standardized effect sizes is not statistically reliable.
Cluster Analysis
In order to further explore the complex interrelationships between our theoretical variables of interest, we conducted a cluster analysis of the White sample to determine whether there are distinct clusters of individuals who hold positive or negative attitudes toward the flag for distinct reasons (i.e., conservatism, Southern pride, negative racial attitudes). This allows for a more nuanced interpretation that moves beyond aggregated general trends. Cluster analysis groups individuals based on their similarity across the variables of interest so that members within a group are as similar as possible and members from different groups are as different as possible (Clatworthy, Buick, Hankins, Weinman, & Horne, 2005).
We conducted a two-step cluster analysis using log-likelihood and Schwarz’s Bayesian Criterion (BIC) (Satish & Bharadhwaj, 2010) in SPSS v. 24. This approach automatically selects the optimum number of clusters that maximizes the overall probability of the data given the final clusters. The variables are standardized to make them commensurable. The results suggest four unique clusters of individuals (see Figure 1). The largest group (34.8%), designated the “New Southerners”, have a more liberal orientation (M = 3.17, SD = 1.09), strong Southern pride (M = 4.05, SD = .619) and reflect positive racial attitudes (M = 1.63, SD = .430). The next largest group (29.7%), designated the “Cosmopolitans”, also have a liberal orientation (M = 2.90, SD = 1.36) and positive racial attitudes (M = 1.58, SD = .529), but have little Southern pride (M = 2.02, SD = .586). The third group (24.2%), designated “Traditionalists”, are more conservative (M = 5.91, SD = .750), have strong Southern pride (M = 4.50, SD = .561), and positive racial attitudes (M = 1.78, SD = .436). The final group (11.3%), designated the “Supremacists”, reflect conservativeness (M = 5.02, SD = 1.13), strong Southern pride (M = 4.20, SD = .815), and negative racial attitudes toward Blacks (M = 3.47, SD = .594).
Figure 1
Means for cluster variables by cluster.
In order to determine whether these distinct sub-groups of Southerners codified by their political ideology, Southern pride, and racial attitudes toward Blacks show different support for the Confederate battle flag, we conducted a univariate ANCOVA with age, sex, motivation to control prejudice, and social desirability as control variables. The results, F(7, 408) = 33.57, p < .001, suggest that there are significant differences in attitudes toward the flag based upon distinct subgroup (see Figure 2). In order to test all pairwise comparisons we adopted the Bonferroni correction for our tests of significance, resulting in a cut-off criterion of .0083. The Cosmopolitans’ attitudes toward the flag were significantly more negative than the attitudes of New Southerners, g = .60, p < .001, Traditionalists, g = 1.43, p < .001, and Supremacists, g = 1.80, p < .001. New Southerners showed significantly more negative attitudes toward the flag than did Traditionalists, g = .84, p < .001, and Supremacists, g = 1.18, p < .001. Finally, Traditionalists and Supremacists did not statistically differ in attitudes toward the flag, g = .33, p = .053.
Figure 2
Estimated marginal means for ATF by cluster. Errors bars are 95% confidence interval.
Discussion
For some in the United States, the Confederate battle flag is a symbol of the South’s mythical glory, heroism, and sacrifice (Clark, 1997; Webster & Leib, 2001) and this symbolic pride in the South drives support for the Confederate battle flag. For others, it is a symbol of hate, racism, and oppression (Clark, 1997), and support for the flag is seen as an avenue for expressing negative racial attitudes. While debates regarding the reasons for positive attitudes toward the Confederate battle flag often depict these two positions as a dichotomy, our results suggest a more complex view.
The bulk of our findings relate to White Southerners, who have been the focus of popular discussions and most of the empirical work surrounding support for the Confederate battle flag and Confederate symbols. In general, we found support for three theoretical perspectives. In line with the Southern heritage perspective, Southern pride was strongly associated with support for the Confederate battle flag even while controlling for blatant negative racial attitudes toward Blacks. While this accounted for the largest association, both political conservatism (principled-conservatism perspective) and blatant negative racial attitudes toward Blacks (racial ideology perspective) were also both associated with positive attitudes toward the flag. However, there is a complex interweaving of political ideology, Southern pride, and racial attitudes revealed by the cluster analysis.
Our cluster analysis revealed four distinct groups of White Southerners who hold distinct attitudes toward the Confederate battle flag and have these attitudes linked to distinct make-ups of political ideology, Southern pride, and negative racial attitudes. The most positive support for the Confederate battle flag came from two groups: the Supremacists and the Traditionalists. Both groups reflected a non-statistically significant difference between their Confederate battle flag support and yet their strong support was linked to different profiles. For Supremacists, it was a combination of strong conservatism, pride in the South, and blatant negative racial attitudes toward Blacks; however, for Traditionalists it was only a combination of strong conservatism and pride in the South, without blatant negative attitudes toward Blacks. Traditionalists held similar positive attitudes toward Blacks as did Cosmopolitans and New Southerners. Of the Supremacists and Traditionalists, who tend to favor maintaining the Confederate battle flag on government premises and as part of official state flags in the South, Supremacists make up the minority in our sample. The majority of supporters in our sample show positive racial attitudes toward Blacks.
This has important political, social, and legal implications. Legal arguments for the removal of Confederate symbols from government buildings and grounds rely on the assumption that Confederate symbols are used with the intention to inflame race relations deliberately in violation of the equal protection clause of the 14th amendment of the United States Constitution (Weeden, 2001). Our data suggest that, in general, supporters of the Confederate battle flag are not motivated primarily by blatant negative racial attitudes and most (i.e., Traditionalists) do not exhibit negative racial attitudes at all. Thus, an explanation of support for Confederate symbols that assumes an intention of exhibiting racism in a symbolic way via flag support may be incorrect. Most supporters of the flag are doing so because of their strong Southern pride and their conservative political views and do not hold negative racial attitudes toward Blacks.
The assumption of exclusivity inherent in the racial ideology argument, that what appears to be a Southern pride explanation for support of Confederate symbols is actually masked racism, is not supported by our results. In contrast, there is evidence that Southern pride can exclusively increase support for the Confederate battle flag, as can be seen with the comparison of New Southerners to Cosmopolitans whose only difference is in Southern pride, with equivalent levels of political liberalism and racial attitudes. These findings may help restrain inaccurate stereotypes of supporters of the Confederate battle flag as bigots, when in fact this support is more intimately tied to political conservatism and Southern pride, as revealed through the “Traditionalists” as the majority group of flag supporters. The “Supremacists”, with negative racial attitudes toward Blacks, reflect only a small percentage of Southerners and a minority of Confederate battle flag supporters within our sample.
Black Support for the Confederate Battle Flag
As previous work has indicated (Agiesta, 2015; Webster & Leib, 2001), Blacks are in overwhelming opposition to the Confederate battle flag. Our results also confirm this. Our results generally demonstrate an antagonism toward the Confederate battle flag among Black respondents and no factors examined were significant predictors of Black support for the flag. This may be due to a lack of heterogeneity in flag support among the Black respondents. Only 7.3% of the Black sample scored above the midpoint for positive attitudes toward the flag and 60.6% scored the lowest possible value.
Despite the significant bivariate and partial (controlling for all factors) correlations between Southern pride and attitudes toward the Confederate battle flag among Blacks, this association disappeared in our structural equation model, which reduces the influence of measurement error. While this is the first quantitative study we are aware of that evaluates factors possibly impacting attitudes toward the Confederate battle flag among Black Southerners, we are unable to provide any strong conclusions as to the motivations of the minority of Black supporters. However, we can suggest that anti-White racism does not seem to be involved in Black opposition to the Confederate battle flag, as is sometimes portrayed within the Southern heritage view (Woliver et al., 2001). Qualitative research on Black Southerners who support the flag may help identify more useful variables to analyze in the future.
Limitations and Future Directions
Our study has various strengths in comparison to previous work, including reduced measurement error by avoiding single item measurements of racial attitudes and utilizing latent variable modeling. We also included Southern pride as an important theoretical variable and controlled for motivation to control prejudice and social desirability. Finally, we applied cluster analysis to reveal important underlying patterns in the data otherwise invisible to researchers. This allows for a more accurate interpretation of findings beyond generalized regression coefficients. Despite these, a few limitations should be noted.
First, Mturk respondents are certainly not representative of the general population in the Southern United States. Mturk samples generally have more female respondents and tend to be more politically liberal, which is at odds with the generally conservative states in which we collected our data. This may have resulted in overestimates of the percentages of Cosmopolitans, and New Southerners and underestimates of the percentages of Traditionalists and Supremacists. While our sample is more liberal than the populations of the states we collected from, our sample had political diversity and was not restricted in its range of political attitudes. The more important finding is the relationship of the sub-groups to attitudes toward the Confederate battle flag and in the relative comparison between Traditionalists and Supremacists, not their absolute sizes in the population. Mturk offers more diverse sampling than traditional samples in the psychology literature, with a generally high degree of data quality (Buhrmester, Kwang, & Gosling, 2011; Casler, Bickel, & Hackett, 2013; Paolacci & Chandler, 2014) and Mturk samples show greater attentiveness than traditional samples (Hauser & Schwarz, 2016). Our use of only Mturk workers with minimum approval ratings of 95% acted as an additional protection on data quality (Peer et al., 2014).
Second, we collected data only from three states in the Southern United States, and this certainly limits our generalizability to other states. However, South Carolina, Mississippi, and Georgia do reflect important political contexts for our research questions. It may be the case that the impact of Southern pride, political conservatism, and racial attitudes will differ among other states. For example, some states like Texas are geographically part of the South but may reflect greater pride in a specifically Texan identity and symbols (Lang, 2010), which may alter the impact of Southern pride on attitudes toward Confederate symbols. Likewise, our categorization of Southerners into the four groups of Cosmopolitans, New Southerners, Traditionalists, and Supremacists may or may not hold within each individual state in the Southern United States.
Third, we used a blatant measure of racial prejudice based largely on items from old-fashioned racism scales (Brigham, 1993; McConahay et al., 1981). We opted for this type of measure because past research has demonstrated a stronger link between this “old-fashioned” racism and attitudes toward the Confederate battle flag than with modern prejudice (Orey, 2004). Additionally, we were interested in how attitudes toward Blacks impacted attitudes toward the Confederate battle flag, not whether attitudes toward specific policies (e.g., affirmative action, reparations, speed of reducing inequality) were related to attitudes toward the Confederate battle flag. While it will be useful in future research to include alternative measures and conceptualizations of racial attitudes, such as more subtle prejudice scales or implicit measures of racism, these measures come with their own criticisms and complexities (Carlsson & Agerström, 2016; Chambers & Schlenker, 2015; DeBell, 2016; Oswald, Mitchell, Blanton, Jaccard, & Tetlock, 2013; Popa-Roch & Delmas, 2010; Sniderman & Tetlock, 1986; van Ravenzwaaij, van der Maas, & Wagenmakers, 2011).
Fourth, our application of a general model applied to both Black and White participants limited our ability to understand the attitudes of Black Southerners. While the number of Blacks expressing support for the Confederate battle flag was low, it is this small group that may be of interest in future studies examining why some Blacks choose to support the flag. While we found no evidence in this study that Southern pride or political conservatism functioned similarly across Blacks and Whites, this may have been due to the smaller sample size among Blacks being unable to detect the effect. Plausibly, it could also be that a differential effect of Southern pride exists due to different historical experiences, memories, and attribution of meaning. What is the qualitative content of Southern pride among Whites and Blacks respectively? Additionally, the link between conservatism and attitudes toward the Confederate battle flag may be dependent on specific dimensions of conservatism (e.g., Dodd et al., 2012).
We have conducted a robust correlational study on attitudes toward the Confederate battle flag among White Southerners with the intention of clarifying the respective impacts of Southern pride, racial attitudes, and political conservatism. We provide strong support for the Southern heritage perspective that positive attitudes toward the Confederate battle flag can often be driven by pride in the South independently of blatant negative racial attitudes toward Blacks. We further demonstrate that there is a unique sub-group of Southerners that we identify as Traditionalists who reflect positive racial attitudes toward Blacks and yet still show great support for the Confederate battle flag. This sub-group of Southerners seems to constitute the majority of those supporting the flag.