Racial Sympathy and its Limits

Racial Sympathy and its Limits Abstract Reversing course from a long tradition of studying racial antipathy, I argue that racial sympathy, defined as white distress over black misfortune, shapes public opinion among a surprisingly sizable subset of white Americans. Using an original measure – the racial sympathy index – I find that sympathy motivates support for policies perceived to benefit, as well as opposition to policies to perceived to harm, African Americans. Racial sympathy is distinct from a more general sympathy, as it does not shape opinion related to other groups. The concept is foremost a racial attitude; as evidence of this, I find that sympathy is activated when a policy draws attention to its black beneficiaries. I probe the limits of this effect and find that even exposure to negative stereotypes of blacks does not extinguish the influence of sympathy. The consistent results across multiple national studies suggest that racial sympathy represents a distinct dimension of American racial attitudes. 1 “Negroes should be given social equality, any job they are qualified for; should be able to live in any neighborhood, and so on. ...We aren’t unified and we don’t know what we’re fighting for and the discrimination is at the root of it....The discrimination toward Negroes is because they aren’t understood and because they are physically different.” --Larry, The Authoritarian Personality, 1950 In 1950, social psychologists introduced scholars to two students: Mack, “a man high on ethnocentrism” and Larry, “a man low on ethnocentrism.” In their seminal study of The Authoritarian Personality, Adorno and his colleagues examined how Mack’s negative views of the out-group lead to a “prejudiced outlook of the world” (224). Since the publication of The Authoritarian Personality, decades of social science research have studied individuals like Mack. In the American politics subfield, scholars have found that racial prejudice towards blacks is an important determinant of white opinion on public policy (Kinder & Sanders 1996; Sniderman & Carmines 1997) and in at least some cases, may also influence vote choice in elections with black candidates (Kinder & Dale-Riddle 2012; Tesler & Sears 2010). But what about Larry? Although an extensive literature examines the antecedents and consequences of out-group antipathy (see Huddy & Feldman 2009; Hutchings & Valentino 2004 for reviews), scholars have rarely considered the other side of the coin: that is, the possibility that some whites carry sympathy towards black Americans and that this unique attitude shapes public opinion. Given the documented prevalence and power of racial prejudice, scholars’ concentration on prejudice is understandable. Still, there have been occasional but salient instances in which white Americans have endorsed political measures seen to advance African American interests. For example, some political scientists argue that a nontrivial subsection of whites favored Obama because of, rather than in spite of, his race (Kinder & Dale Riddle 2012, Tesler & Sears 2010, Tesler 2016). A 2016 study by Pew found that 36% of whites say that racial discrimination is a “major reason” why African Americans have a harder time getting ahead. Recent white participation in the Black Lives Matter movement and the Charlottesville counter protests suggests that some white Americans are 2 openly remorseful of blacks’ situations and, in some cases, critical of their own group’s role in perpetuating racial strife.1 Moreover, what I refer to as racial sympathy, defined as white distress over black suffering, is not a new phenomenon in American politics. White Americans have long engaged in efforts to both obstruct and promote the political advancement of African Americans. For example, some whites participated in the Abolitionist Movement, walked alongside blacks during the March on Washington, and spearheaded campaigns to elect Blacks to local office in the 1970s (Sonenshein 1993). These events have often left indelible marks on American politics, yet political scientists cannot use existing theories of racial attitudes to explain why some whites are compelled to engage in these types of activities. This research attempts to fill that void. In this article, I examine the role of racial sympathy in contemporary American politics by conducting a series of analyses exploring the relationship between sympathy and support for racialized policies: those policies that implicitly or explicitly reference race. Employing observational and experimental approaches across multiple independent national surveys, I unearth a clear and consistent pattern: racial sympathy matters in American politics. Crucially, racial sympathy is not merely the opposite of racial prejudice. Rather, I demonstrate that it is a unique racial attitude, lying on its own conceptual dimension. I begin the article by defining the concept of racial sympathy and introducing an original measure, the racial sympathy index. Using this measure, I explore the relationship between racial sympathy and support for racialized public policies. I uncover a strong and robust connection between this racial attitude and public opinion. Racially sympathetic whites are consistently more 1 For example: https://www.washingtonpost.com/posteverything/wp/2015/08/06/this-is-what- white-people-can-do-to-support-blacklivesmatter/ 3 inclined to support policies perceived to benefit African Americans. Importantly, I demonstrate that this relationship does not erode in the presence of multiple plausible alternative explanations. I also conduct a series of analyses that distinguish racial sympathy from other types of social sympathy, demonstrating that racial sympathy is foremost a racial attitude. I then consider the conditions that give rise to racial sympathy in American politics. Using a national survey experiment, I find that racial sympathy is activated when a policy draws attention to its black beneficiaries. I also examine how sympathy responds to cues that typically prime resentment. Viewing negative stereotypes of African Americans does not cause racially sympathetic whites to revoke their support for “pro-Black” measures.2 Instead, distress over black misfortune enables these whites to withstand the activation of resentment and its consequences. The results of the study suggest that sympathy is an attitude distinct from prejudice. Without it, white public opinion would be far less supportive of efforts to advance black interests. I conclude the article by discussing the implications, limitations and extensions of the results. I argue that by including racial sympathy in future studies, scholars can gain insight into the diversity of ways in which racial attitudes shape American politics. Racial Sympathy: What We Do and Do Not Know In his seminal work on public opinion, Converse argued that most of the mass public did not think about politics in ideological terms. Instead, most citizens conceived of politics in terms of 2 I use Craemer’s (2008) term “pro-Black” in quotes, throughout the text to refer to those policies that may be perceived to benefit blacks. Craemer writes that these policy areas, “are of a general nature and represent standard items used to measure opinions on the role of government in alleviating racial inequalities” (416). That said, some argue that these policies do not, in practice, benefit blacks. See Sander (2004) for a discussion of affirmative action. 4 salient social groups (1964). His essay spawned a rich line of inquiry on the “group-centric” foundations of public opinion, much of which has focused on racial groups (Nelson & Kinder 1996). Though Converse did not consider the possibility of positive racial attitudes in detail, he did propose a survey question intended to reveal whites’ multiple attitudes toward African Americans: “Are you sympathetic to Negroes as a group, are you indifferent to them, or do you dislike them?” (Converse 1964, 235). By presenting two alternatives to out-group animus, Converse’s question acknowledged that non-prejudicial attitudes could take several forms, including sympathy. Despite this possibility, political science research has narrowed its focus to studying those whites who dislike blacks and it has done so with gusto – indeed, the term “racial attitudes” has often become synonymous with racial prejudice.3 Research on other racial attitudes, in contrast, is scant. To be sure, some scholars have referenced the “racially sympathetic” (Kinder and Dale-Riddle 2012, Tesler 2016), whites who “hold blacks in esteem” (Sniderman & Stiglitz 2008) and other work has found evidence of “pro-Black” public opinion (Craemer 2008). Despite the contributions this work has made, it suffers from two important limitations. First, much of this work considers the potential role of pro-Black attitudes on evaluations of a single Black public figure: Barack Obama, who is, in many ways, unique.4 It is unclear whether positive attitudes about African Americans impact other domains of public opinion. 3 For example, in Kuklinski and colleagues’ article “Racial Attitudes and the “New South” (1997), the authors examine levels of racial prejudice in the South. Similarly, when Valentino and colleagues (2002) examine “how political ads prime racial attitudes during campaigns,” they are examining how ads prime three attitudes all related to animus: racial resentment, laissez-faire racism, and subscription to the notion that blacks have too much influence (86). 4 See Hutchings 2009, p. 919. 5 Second, and more importantly, much of the existing scholarship has not developed strong theoretical foundations for these attitudes. Consequently, we cannot answer basic questions about racial sympathy, such as: what is it? Who are the racially sympathetic? Is sympathy merely the opposite of prejudice? And why and when does sympathy matter in politics? To the extent that the existing literature offers responses to these questions, it does so by employing theories and measures of prejudice, mostly forgoing a theoretical account of sympathy. Instead of using the framework of prejudice, I draw on work from political theory (Nussbaum 2001, 2013) and psychology (Batson 1987, 2011) to define racial sympathy as an in-group member’s distress over out-group misfortune. In the United States, racial sympathy refers to white distress over black suffering.5 It is best conceptualized on a spectrum: those high in racial sympathy experience severe distress over black suffering, which they perceive to be prevalent, while those low in racial sympathy are more indifferent to black suffering, which they perceive to be negligible. Between these two extremes lie most white Americans. “Racial sympathy,” which I occasionally refer to in shorthand as “sympathy,” is a racial attitude. This account is consistent with a variety of research traditions in political science which 5 A racially sympathetic white American could also hold sympathetic feelings for Latinos, Asian and Native Americans and these attitudes could influence opinion on immigration or access to the Dakota Access Pipeline, for example. However, since the black/white divide is the United States’ most salient racial cleavage (Hutchings & Valentino 2004), it is the focus of this article. Additionally, it is possible that racial sympathy exists in other countries with histories of racial division. Work on reconciliation in post-Holocaust Germany (Olick & Levy 1997) and in post-Apartheid South Africa (Gibson & Gouws 2000) has found that in-groups can deliver restitution to mistreated out-groups, perhaps due to in-group distress over out-group suffering. 6 argue that attitudes toward social groups have strong emotional components. For example, Banks and Valentino (2012) suggest that racial prejudice contains the “emotional substrates” of anger and disgust. I argue that when some whites think about blacks, sympathy surfaces in an enduring and meaningful way. Racial sympathy can affect white opinion and behavior in many domains. In American politics, I argue that racial sympathy can influence white support for public policies perceived to benefit blacks. Similarly, racial sympathy can also influence white opposition for public policies perceived to harm blacks. Although there are many reasons why a white person might support a racialized policy – perhaps it is championed by his party, or corresponds with her preferences for government intervention, or aligns with his values – I suggest that, for some whites, feeling sympathy for African Americans provides an additional and significant boost on top of the many forces we already know to shape public opinion. Most white Americans do not interact with black Americans on a day-to-day basis, and when they do, these interactions are often brief and superficial (Sigelman et al. 1996). Politics is thus one of the few venues in which white Americans can exercise their sympathy. In this respect, the political effects of whites’ sympathy are rooted not in white citizens’ self-interest, but instead in more symbolic concerns. Political symbols, like African Americans, “rivet attention” (Sears 2001) thereby occupying a “central” role in American politics (Hutchings & Valentino 2004). Accordingly, racial sympathy has the potential to map onto a number of political outcomes, including those that feature blacks prominently as well as those that are associated with African Americans only implicitly. Although the attachment is symbolic, the political consequences are tangible. For some racially sympathetic whites, the association between black suffering and personal distress is so fixed that it withstands the impact of other racial attitudes, like prejudice. For example, as the introduction 7 suggests, a non-trivial percentage of whites think that blacks face racial discrimination. Some of these individuals recognize discrimination only when it is made blatant and unambiguous. However, white Americans at the highest levels of racial sympathy acknowledge and bemoan racial inequality even in the absence of these explicit attributes. For these whites, black suffering that is inconspicuous or subtle can trigger sympathy. Further, their sympathy does not hinge on evaluations of deservingness; indeed, black suffering that could be construed as somehow warranted may nonetheless evoke racial sympathy among those whites at the highest end of the spectrum. This is because sympathy is not merely the absence of prejudice, it is the presence of white distress over black suffering. I extend on this theory in the next section by highlighting the relationship between sympathy and a better known racial attitude, racial prejudice. Prejudice and Sympathy Racial sympathy is an attitude that is related to but distinct from racial animus. Like animus, sympathy is rooted in racial group-based assessments. Additionally, the two attitudes contain affective and cognitive responses to African Americans. There is reason to suspect that both attitudes develop early in life, and have shared sources including parents, teachers, community members, and peers (see Allport 1954, Bigler 1999, Hraba & Grant 1970). However, sympathy parts ways with prejudice in its focus on black suffering and the corresponding distress it evokes. Unlike prejudice, sympathy does not take up the “faulty and inflexible generalizations” that Allport (1954) observed as characteristics of prejudice. Racial sympathy is, first and foremost, an attitude that acknowledges and laments black misfortune. Since racial sympathy is independent from racial prejudice it is possible that a white individual could possess both attitudes simultaneously. For example, a white person could feel distress over black suffering while still considering some group members lazy. Or a white person could consider blacks to be hardworking but not feel sorry about the conditions the group faces. Low prejudice does not 8 guarantee racial sympathy. However, by labeling those whites who score low on resentment as “sympathetic,” scholars preclude this possibility. Indeed, although some of these whites may in fact be sympathetic to the plight of African Americans, they might also be characterized by their “utter indifference” (Kinder 2013; Pettigrew 1982) toward racial inequality. Research on white racial apathy suggests that some whites fail to register any reaction – positive or negative – when they reflect on black suffering (Forman & Lewis 2006). Under conditions of an American “apartheid” (Kinder & Mendelberg 2000, Massey & Denton 1993), white detachment from blacks’ circumstances amounts to tacit approval of racial inequality. Otherwise put, lack of prejudice does not create sympathy over African American suffering (Forman 2004; Kawakami et al. 2009). For these reasons, to measure racial sympathy, it is not sufficient to invert racial prejudice measures like racial resentment, which were born out of a need to understand white opposition to racialized public policies. Instead, we must develop and employ a new measure specifically designed to calibrate white distress over black suffering, a task to which I now turn. Measurement Previous attempts to operationalize racial sympathy have followed two approaches. As discussed, one approach has been to invert measures designed to calibrate racial animus. However, since I theorize that racial sympathy is based on feelings of distress in response to black suffering and is therefore substantively distinct from (low) resentment, existing measures of animus are inadequate for measuring sympathy. Others have attempted to measure sympathy as a distinct concept, mostly using self-report questions (Batson 1987; Iyer et al 2003). For example, in a study conducted by Iyer and colleagues, the authors asked respondents to complete the sentence “When I think about racial discrimination by white people I feel…” and were then presented with a list of adjectives. If subjects selected the 9 words sympathetic, compassionate, and/or empathetic, these responses were grouped into a single index to measure “sympathy” and included as the main independent variable in models intended to predict pro-social opinion. Combined into a single index, these emotions showed properties of high internal consistency.6 Still, it is not clear what this measure captures. Some research in social psychology suggests that sympathy, compassion, and empathy represent conceptually separate emotions (Wispe 1986). As such, collapsing distinct concepts into a single term could muddle important differences between empathy and sympathy, for example. It is certainly possible that some whites empathize with blacks, which is to say they vicariously experience (Hoffman 1981), the feelings of black Americans and that this powerful connection maps onto their politics. However, white perspective-taking does not seem to be the primary reaction to black suffering. This observation is based on information collected during preliminary research. Throughout 2013, I gathered initial background for this project by embedding myself in a series of events hosted by a public Midwestern university. These events were part of a broader series of programming aimed with the purpose of understanding race and ranged from semi-structured discussions following a museum exhibit to a casual dialogue after a play with racial themes. The events often drew white people – both students and community members – who I suspected to be racially sympathetic, so I attended to hear what they had to say. Consistently, white participants referenced feelings of sadness over racial inequality. They did not articulate their attitudes in empathetic terms; that is, very rarely did they express being able to relate to the experiences of blacks. Rather, their comments tended to focus more on their distress 6 I analyzed this index for a related study hosted on YouGov-Polimetrix in 2014. The Cronbach’s alpha was 0.93. 10 over black Americans’ circumstances. For example, during one semi-structured discussion, the moderator asked the group why racial prejudice persisted. One white woman observed: “If you look at the city, and you know, you’re saying the schools are bad and everyone there is a certain race, then you’re putting a bunch of people who, you know, have been socially misplaced or downgraded, and giving them the worst of things, and then blaming it on the color of their skin…people don’t understand that blacks’ skin color doesn’t cause the bad schools. It’s really distressing.” 7 Reflecting on this exploratory research, I concluded that although racial sympathy might have multiple emotional components, sympathy seemed to be the most central.8 Furthermore, this research also suggested that sympathetic whites experienced sympathy in reaction to tangible, 7 Furthermore, similar comments were found among whites in an online study conducted on Amazon’s Mechanical Turk (MTurk) in the summer of 2012. In this study, subjects were given an opportunity to provide open-ended responses to accompany their answers to multiple-choice questions about race. Of the 95 whites that completed the study, approximately 21 respondents consistently referenced feeling sorry for blacks and often in terms that emphasized blacks’ misfortune (for example: I feel bad for the way black people have been treated in this country.”). In contrast, fewer white subjects (approximately 5) expressed their attitudes in empathetic terms (“I have been the target of social injustice and it doesn’t feel good”). Similar results were found on a related MTurk study, conducted in the spring of 2013. See Web Appendix Table 1 for an overview of all supplemental studies referenced in this manuscript. 8 Additionally, I have examined the relationship between racial sympathy and the Mind in the Eyes (MIE) test, a measure that has been used by some political scientists to approximate empathy (Feldman et al 2015). I fielded a convenience study with both measures on MTurk in July 2017. Among whites in the sample, the correlation between MIE and the racial sympathy index was 0.11. 11 personal misfortune experienced by blacks, which is to say, the primary facets of their sympathy were not abstract and principled notions of equality.9 Using the language and themes collected during this preliminary research, I formed a measure of racial sympathy by composing a series of vignettes, each depicting an instance of black suffering. After reading each vignette, subjects are asked to report how much sympathy they feel toward the black character(s) described. The answer choices ranged from “I do not feel any sympathy” to “A great deal.”10 I combine responses to all four questions to form a racial sympathy index, presented in Table 1.11 The index captures the extent to which white Americans experience distress over black suffering across a range of contemporary contexts. Table 1 about here 9 This corresponds with Walsh’s (2008) discussion of “practical politics” (p. 7-8). 10 I refer to this question as “Question 1.” To reduce question error and increase response variation, I included an additional question, “Question 2.” The content of Question 2 differed slightly among the four scenarios: for example, respondents were asked if they liked, or would be friends with, the target of discrimination, or occasionally, with the white perpetrator of discrimination (in this case, reverse coded). I found that this eight-item index performed very similarly to a shortened, four-item index that only used the responses from Question 1 (correlation 0.95 among whites). All analyses present here, therefore, use this four-item index (that is, an index that combines four responses to Question 1), though the results are robust across specifications that use the eight-item racial sympathy index. 11 The distribution of each vignette appears in Appendix Table 1. 12 The benefit of vignettes is that they enable subjects to react directly to specific stimuli rather than to abstract notions of racial inequality. Through the vignettes, subjects can “easily identify the broader set of issues to which this particular one apparently belongs.” (Schwarz 1994 135). Data The racial sympathy index was administered on a module of the 2013 Cooperative Congressional Election Study (CCES) fielded in November 2013 by YouGov/Polimetrix. The 2013 CCES is a national Internet sample and the racial sympathy vignettes were distributed to 1,000 respondents, 751 of whom identified as white. To approximate national representativeness, the CCES uses a two-stage selection and weighting scheme based on Census estimates and propensityscore weighting. Respondents are identified using "sample matching," which means that individuals on the panel are selected for specific studies based on their demographic characteristics so that the invited sample will match the population's characteristics. Studies have found that the CCES sample performs similarly to ANES samples on important variables, such as vote choice (Ansolabehere & Rivers 2013; Vavreck & Rivers 2008).12 The Properties of the Racial Sympathy Index 12 Additionally, where possible, I replicate the results from the CCES analyses using a serviceable measure of racial sympathy which appeared in the 2008 and 2012 American National Election Studies (ANES) Time Series and in the 1994 General Social Survey (GSS). This question asks respondents “How often do you have sympathy for blacks?” Unlike the CCES, ANES and GSS respondents are selected through address-based sampling, the more traditional method of drawing representative samples. Both the ANES and the GSS are conducted using face-to-face interviews, mitigating concerns related to coverage and response effects (Krysan et al 1994). 13 Before exploring the political consequences of racial sympathy, it is important to establish the validity of the racial sympathy index, especially given the long-standing debates about the measurement of racial attitudes (see Huddy & Feldman 2009). The purpose of this section is to demonstrate the validity of the racial sympathy index and to begin to identify the ways in which it is distinct from prejudice. Since racial sympathy is defined as white distress over black suffering, all analyses in the article restrict the sample to non-Hispanic whites. To start, substantial proportions of respondents report feeling sympathy for the black individual(s) described in the vignettes. Indeed, among whites in the CCES sample, the average level of racial sympathy was 0.61 on a scale from 0 (least sympathetic) to 1 (most sympathetic); Appendix Although the targets of discrimination in the vignettes are varied, white respondents expressed relatively consistent levels of sympathy. This is reflected in the index’s Cronbach’s alpha (.74 for the 4-item index), suggesting that the index displays adequate internal consistency. Furthermore, examining the vignettes’ inter-rest correlations, that is, the correlation between an item and the scale that is formed by all other items, suggests that the responses are highly correlated with each other (Nunnally & Bernstein 1994). Factor analysis confirms this point. As Table 2a demonstrates, using promax oblique rotation, responses to the four vignettes load heavily and roughly onto a single factor. Table 2a about here Further, Figure 1 displays the racial attitudes of the white respondents in the CCES by their scores on the racial sympathy index as well as their scores on the racial resentment index.13 If racial sympathy were simply the opposite of racial resentment, we might expect for all whites who score high in sympathy to be concentrated in the top left corner of the chart, where resentment is lowest. 13 In the 2013 CCES, the Cronbach’s alpha of racial resentment index was 0.87. 14 Instead, the dispersion of whites’ sympathy and resentment scores demonstrate that many whites hold both sympathetic and resentful attitudes about African Americans. This confirms Schuman and Harding’s (1964) observation that sympathy: “…cannot be thought of as simply equivalent to what is usually measured under the term “prejudice.” The two types of measure are clearly related, but not so much so as to consider one a close substitute for the other.” (238). Figure 1 about here This point is reinforced through the modest correlation of racial resentment and racial sympathy– the raw correlation is -0.45 — which suggests that the concepts are negatively related but not interchangeable.14 Additionally, in examining the correlates of racial sympathy and resentment, as displayed in Appendix Table 3, I find that party identification, education, and limited government are all associated with racial sympathy and racial resentment. However, the strength of these associations is consistently stronger in the resentment model, reflecting the politicized nature of the concept (Huddy & Feldman 2009), a result to which I will later return. Since some scholars have argued that racial resentment captures sympathy (Kinder and DaleRiddle 2011), I conducted an additional factor analysis in which I combined the racial sympathy index with the questions from the racial resentment scale. This analysis allows me to examine whether the two concepts lie on a single dimension or, instead, are independent. Although two of the racial resentment questions draw attention to African American suffering,15 the results presented 14 The correlation between resentment and sympathy varies somewhat between samples. Among whites in MTurk samples, the correlation has been as low as -0.19 and as high as -0.4. 15 These questions ask respondents to agree/disagree with the following statements: 1) “Over the past few years, blacks have gotten less than they deserve” and 2) “Generations of slavery and 15 in Table 2b suggest that the racial sympathy index loads well onto one factor and the racial resentment index loads well onto another. Notably, the “sympathetic” items of the racial resentment scale do not appear to load heavily onto Factor 2, suggesting that if these components do capture a form of racial sympathy, it is substantively different from the type of sympathy represented in the racial sympathy index. Indeed, when I combine the two “sympathetic” items of the racial resentment scale into an abridged 2-item index and calculate the correlation of these “sympathetic” items with the racial sympathy index, I find that they are only moderately correlated (raw correlation is 0.44). Table 2b about here These exercises in concept validation provide compelling initial evidence that racial sympathy is distinct from low-end resentment. But when it comes to understanding why whites support pro-Black policies, does racial sympathy provide unique explanatory power that low animus cannot? In the next section, I take up this question by examining the relationship between racial sympathy and public opinion. Racial Sympathy and Public Opinion In this section, I conduct a series of regressions examining the association between racial sympathy and policy opinion. I start by examining white support for “government aid to blacks,” a broad policy area that has appeared on the ANES since the 1970s. The question asks respondents to place themselves on a 7-point scale that ranges from “Blacks Should Help Themselves” to “Government Should Help Blacks.” Previous research on this question has found that racial animus leads some whites to oppose government aid to African Americans (Hutchings 2009). This analysis discrimination have created conditions that make it difficult for blacks to work their way out of the lower class.” 16 reexamines the sources of opinion in this area. Table 3 about here As the first column of Table 3 displays, a preference for limited government shapes white opinion on government aid to blacks, a result that is expected, given previous research. The analysis also confirms that resentment is a powerful component of white opinion in this policy domain (Kinder & Sanders 1996).16 But does prejudice, as measured here by racial resentment, capture the full extent of racial attitudes that shape public opinion in this domain? Based on the racial sympathy coefficient displayed in the top row, it appears that the answer to this question is no. The racial sympathy coefficient has a substantively meaningful magnitude, representing one-tenth of the scale. Its significant association with policy opinion suggests that the racial sympathy index is capturing unique dimensions of racial attitudes that low animus cannot. Based on these results, it seems that racial sympathy is significant associated with support for government aid to blacks. But is it limited to this policy area? Fortunately, the CCES provides 16 It is worth noting that the racial resentment coefficient is larger in magnitude than sympathy’s. Why might this be the case? Some scholars have argued that racial resentment, as a measure of animus, is a politicized concept, containing non-racial elements (e.g., see Huddy & Feldman 2009), and the correlates of racial resentment, presented in Appendix Table 2, certainly support that interpretation. In contrast, the racial sympathy index’s questions probe subjects’ sympathetic reactions to scenarios seemingly distant from political life. For this reason, racial sympathy has a low correlation with political variables and therefore may not be as influential, relative to racial resentment, on policy opinion variables. Indeed, this is confirmed in subsequent discriminant validity analyses. 17 questions related to five other racialized policies, including: support for subsidies for black businesses, scholarships for qualified black students, funding for schools in black neighborhoods, affirmative action, as well as welfare (Gilens 1999). I replicate the analysis conducted in the first column of Table 3 across these policy areas and display them in the right-hand columns. As the table demonstrates, in four out of five cases, I find that racial sympathy provides a unique and significant contribution to explaining racialized policy support.17 Racial Sympathy and Sympathy for Other Groups Scholars of prejudice have often disagreed about the extent to which concepts like racial prejudice represent racial attitudes or whether they more accurately reflect non-racial values. One critique often levied against racial resentment, for example, is that the scale’s questions evoke ideological principles (Feldman & Huddy 2005). Similarly, white Americans could support policies like government aid to blacks not because they experience distress over black suffering, but instead, because they prefer government intervention on racial matters. Indeed, by using terms like “racial liberals” (Tesler and Sears 2010), “racial liberalism” (Zaller 1992), or “racially liberal whites” (Banks 17 Further, the 2012 ANES, the 2008 ANES and the 1994 GSS provide three additional opportunities to examine the relationship between sympathy and opinion. These surveys also provide a wealth of variables, which permit me to examine a range of alternative hypotheses. To do so, I subject the model presented in Table 3 to a series robustness checks to see if other factors, including stereotypes, implicit attitudes, and interracial closeness might absorb the significant relationship between sympathy and opinion. In each case I find that the relationship between racial sympathy and opinion remains significant, suggesting that the concept is capturing a distinct dimension of racial attitude. These results are presented in the Appendix in Tables 3-4 and in the Web Appendix Tables 2-4. 18 2014), scholars seem to acknowledge a role for ideology. The results in the preceding tables include measures of limited government as a control variable18 and Appendix Table 4 shows that the relationship between racial sympathy and opinion is also robust to models that include a measure of egalitarianism, a value defined as one’s “commitment to equality” (Feldman 1988 424).19 Generally, these analyses suggest that racial sympathy is not equivalent to broad orientations like egalitarianism or liberalism, typically defined as encompassing “the values of equality, social justice, and widespread involvement in the political system” (Feldman 2013). Instead, and consistent with the theory of racial sympathy, support for these measures is rooted, at least partially, in specific feelings about blacks. The preceding section has demonstrated that is not equivalent to prejudice. In this section, I attempt to distinguish racial sympathy from a more general sympathy for marginalized groups. To do so, I conduct a series of analyses related to gender. Following the approach of racial sympathy vignettes, I included two measures of “gender sympathy,” intended to represent distress over women’s suffering, on the 2013 CCES. In each of these vignettes, a woman is described to be experiencing a discriminatory situation and subjects are asked to indicate how much sympathy they 18 These results are also robust to analyses that include self-reported ideology. 19 Feldman and Steenbergen (2001) argue that support for policies such as welfare can be explained by considering humanitarianism. Using a convenience sample, I conducted an analysis which I replicate the analysis presented in Appendix Table 4 using the humanitarianism index in place of the egalitarianism and found similar results. I have also conducted analyses where I consider the role of personality traits (openness and agreeableness). Even with these stringent controls, I find that racial sympathy retains a significant relationship with opinion. See Web Appendix Table 4. 19 have for her, mirroring the format of the racial sympathy index.20 If racial sympathy is primarily a racial attitude, then I expect that it should not influence support for policies perceived to benefit women. The results are displayed in Table 4. The table is divided into three large columns, each representing a different women’s policy area. Within each policy area, I analyze two different model specifications: 1) the relationship between racial sympathy and policy opinion and 2) the relationship between gender sympathy and opinion. Table 4 about here Through these analyses, I find that the gender sympathy index leads some whites to support policies that conceivably benefit women. The substantive effect is large; in the case of women’s affirmative action, the magnitude of the coefficient represents almost one-fifth of the scale. In contrast, the racial sympathy index is not generally associated with these gendered public policies, even though they might seem to address the “durable inequalities” faced by women (Burns et al 2016). The results presented in Table 4 suggest that other groups can benefit from group-specific sympathy; however, racial sympathy is not a broadly transferrable attitude.21 In sum, these analyses 20 The raw correlation of the gender sympathy index and racial sympathy index was .3 for the entire sample (n=1000) and .28 for whites alone (n=751). For white women, the raw correlation of these two indices was .31 and for white men it was .3. The correlates of the gender sympathy index appear in Appendix Table 5. 21 Using the CCES, I’ve conducted additional discriminant validity exercises where I regress racial sympathy on policies that do not implicitly or explicitly reference African Americans. The results of these analyses are displayed in Appendix Table 6. Here I observe that racial sympathy is not significantly associated with policy opinion generally. Importantly, racial sympathy does not 20 reveal that racial sympathy has narrow relevance and influence on matters of race – especially on those matters involving African Americans, perhaps reflecting the uniquely “rigid color line” (Sears & Savalei 2006) that separates white America from black. Activating Racial Sympathy If, as I’ve argued, racial sympathy exists and shapes public opinion, then I should be able to activate it and observe its political consequences. More specifically, I am interested in examining the extent to which exposure to African Americans heightens the effect of racial sympathy on political outcomes by leading some whites to endorse “pro-Black” policies and oppose other policies seen to hinder the group. In a survey wording experiment conducted on the 2013 CCES, I found that racial sympathy mapped on support for public policies when they were described to benefit blacks, but not when these same policies were described to benefit the poor. These results, displayed in Appendix Table 7,22 complement earlier work that has found that support for a policy is sensitive to the group identity of the policy’s beneficiaries (Kinder & Sanders 1996; Sniderman & Carmines 1997). influence opinion on policies that involve a different marginalized group: immigrants. Further, I exploit the wide availability of policy questions in the 2012 ANES to repeat this exercise with the ANES Question. These results appear in the Appendix Figure 1 and reveal that racial sympathy does not consistently map onto support for non-racial public policies. In contrast, racial resentment does appear to be tapping into a broader orientation, a result consistent with Kinder and Sanders (154, 1996). 22 The predicted probability plots for this experiment are displayed in Web Appendix Figure 1. The manipulations themselves appear in the Web Appendix. 21 Subsequent work has found that attaching negative stereotypes to blacks further depresses support for “pro-Black” political measures among whites high in racial animus. For example, Mendelberg’s work (2001) argues that implicitly racial messages powerfully prime racial attitudes during campaigns, influencing white vote choice. Valentino and colleagues (2002) expanded on this line of research, finding that resentment is cued when subtle racial stereotypes are paired with images of black Americans. Given the well-documented link between racial primes and the activation of prejudice, it is important to examine whether negative racial cues influence sympathy. Therefore, I conducted an experiment with two goals. The first is to explore the circumstances under which sympathy is maintained in the presence of cues that might prime prejudice, thus underscoring the importance of a dedicated measure and theory of racial sympathy. The second goal of the experiment is to expand the scope of racialized policies to consider a timely domain of racialized public policy – criminal justice. Previous research suggests that negative stereotypes of African Americans undergird support for punitive policies such as the death penalty and extended prison sentences (Gilliam & Iyengar 2000, Peffley & Hurwitz 2002). Whereas the other policy areas examined in this article consider those policies that bestow benefits to African Americans, this experiment considers opinion on an area in which African Americans could be considered victims of state punishment. As such, I expect racially sympathetic whites to oppose harsher punishments for African Americans. To address these goals, I conducted a two-wave survey experiment through YouGov. Wave 1, which featured the racial sympathy and racial resentment indices, was administered in August 2016 while Wave 2, which featured the experimental stimulus, was fielded in September. Given the time elapsed during the two waves, it is unlikely that subjects’ responses to the racial attitudes questions contaminated their reaction to the stimulus. 210 white Americans completed both waves of the experiment. 22 By presenting a neighborhood profile featuring a “crime blotter,” the stimulus attempted to mimic a document that citizens might encounter in their day-to-day lives.23 The profile depicted a fictitious black neighborhood which had recently been the target of graffiti. In one condition, the crime blotter indicated that a white man had admitted to painting graffiti on a historic church and in the other condition, the blotter indicated that a black man had admitted to the same offense. The only difference between the two conditions was the race of the culprit, Gavin Tannis, as manipulated by a photograph. Subjects were randomly assigned to one of these two conditions and were subsequently asked to provide their opinions on the appropriate punishment for Tannis. The stimulus and all questions appear in the Web Appendix. I begin my analysis by examining opinion as it relates to punishment of Tannis. Specifically, I consider responses to a question which asked subjects how many hours of community service Tannis ought to serve for the vandalism. I present the results in the form of a predicted probability plot, displayed in Figure 2, with the full model appearing in Table 5. The blue line represents those whites that viewed a white photograph of Tannis (“White Culprit”) and the red line represents those whites that instead viewed a black photograph (“Black Culprit”). Figure 2 about here The charts suggest that racially sympathetic whites issue different responses to the crime, depending on the race of the culprit. The predicted probability of supporting punishment for a black criminal is .87 for those whites who are low in racial sympathy and .46 for those who are high, a difference of forty-one percentage points. On the other hand, when the criminal is white, the predicted probability of issuing a harsh sentence for the criminal is .64 for those who are low in 23 The profile format mirrored a manipulation that appeared in the 1986 GSS. 23 racial sympathy and .86 for those who are high. Otherwise put, whites high in racial sympathy assign roughly half the amount of community service to a black culprit than they do a white culprit for the same offense. Racially sympathetic whites do not hesitate to punish a culprit– which is to say they are not generally anti-punitive – however, they are significantly less likely to inflict a harsh punishment on a black person who commits an identical offense. These results are displayed in Table 5. Table 5 about here A further consideration is whether priming the negative stereotype of black criminal activity simultaneously activates racial prejudice, thus cancelling the “pro-Black” influence of low resentment. In contrast, since sympathy is based primarily on white distress over black suffering, negative stereotypes of black Americans should not dampen its effect. To consider this possibility, I conduct an analysis in which I regress the interaction of racial sympathy and experimental condition and the interaction of racial resentment and experimental condition. The results are presented in the second column of Table 5, labeled “Model 2.” They suggest that, once activated, racial sympathy retains a significant influence on policy opinion even accounting for the activation of resentment.24 On the other hand, the effect of low prejudice is eliminated. When racially sympathetic whites encounter narratives that associate black Americans with crime, thus priming negative stereotypes, sympathy remains an important force in their political decision-making. Indeed, rather than merely lacking prejudice, because these whites possess sympathy they can resist the cues that ordinarily activate animus. 24 The results for the fine punishment vary slightly. Once I add the interaction of racial resentment and experimental condition, the coefficient on the racial sympathy interaction fails to meet standard criteria for statistical significance (p < 0.11, two-tailed test). However, it is in the expected direction. 24 Discussion In general, I observe the largest and most consistent effect for racial sympathy on those policies that explicitly name blacks as beneficiaries, thus facilitating tight interstitial linkage between the policy area and beneficiary (Converse 1964, Nelson & Kinder 1996). The association between racial sympathy and support for the government aid to blacks, an area that explicitly references African Americans is uniformly significant, regardless of model specification. Similarly, the results displayed in Appendix Table 3 reveal a consistent and significant relationship between racial sympathy and affirmative action in hiring.25 This result stands in contrast to the null result between racial sympathy and affirmative action observed in Table 3. Here it is possible that the type of affirmative action might matter. The CCES affirmative action question asks subjects to report their opinion for programs that “give preference to racial minorities” in employment and college admissions. The ANES affirmative action question differs in that it refers to blacks specifically, thus clarifying the relationship black suffering and policy support. Additionally, the relationship between welfare and racial sympathy is somewhat inconsistent, perhaps because welfare is implicitly, rather than explicitly, associated with race. My argument that racial sympathy can motivate whites to express support for African Americans does not imply that efforts to activate racial sympathy will always be successful. In an experiment conducted by Harvey and colleagues (2000), attempts to emphasize black suffering backfired, instead increasing white opposition to programs intended to benefit African Americans. Future research should take up how sympathy over black misfortune responds to circumstances that threaten white group identity. These projects should consider the extent to which racial sympathy is 25 See also Web Appendix Table 2. 25 rooted in whites’ feelings about their own group.26 This project contributed to a rich research tradition that has focused on white attitudes of African Americans. However, it is also important to consider the implications of this work as it relates to other racial and ethnic groups. As the United States becomes more demographically complex and diverse, how and when will whites support those with whom they do not share an ethnic background? The results displayed in Table 4 demonstrate that other group-specific vignettes can predict support for group-specific policies, in this case women. Examining sympathy toward other racial and ethnic groups, such as Latino and Asian Americans, requires theories and measurements that are specific to white perceptions of those groups, which may involve concepts like nativism and foreignness (see Aoki 1996). Therefore, although the vignette approach could be used to study the political consequences of sympathy across a range of targets, the vignettes themselves would need to present instances of discrimination specifically applicable to those groups. Conclusion This paper introduces a new approach to the study of race in American politics. By concentrating on racial prejudice, social scientists have developed only a partial understanding of how racial attitudes affect outcomes. Put another way, the conceptual framework that current theories offer, the measures they test, and even the survey instruments they use are all informed by an effort to examine individuals like The Authoritarian Personality’s Mack, those who are high in racial 26 In other work, my colleagues and I found that racial sympathy is distinct from collective white guilt (Chudy, Shipper & Piston 2018). Collective guilt is oriented toward group responsibility in racial inequality and corresponds with desire for recompense. Guilt is less common and empirically distinct from sympathy. In factor analysis, guilt and sympathy load onto two separate factors. Guilt and sympathy also demonstrate independent effects in models used to predict policy choice. 26 resentment, racial conservatism, or anti-black affect. These tools, I argue, are not sufficient to uncover the complexity and consequences of public opinion that seeks to advance rather than restrict black Americans. My approach to examining the antecedents of “pro-Black” political behavior attempts to consider racial sympathy on its own terms. To do so, I create an original measure of racial sympathy based on fictional vignettes. I find that white responses to these vignettes, most of which take place far from the realm of politics, consistently map onto political preferences. Through a series of analyses using different samples, models, and measures, I uncover a significant relationship between sympathy and opinion on a host of implicitly and explicitly racialized policy areas. The reach of racial sympathy can be found across diverse policy domains, suggesting that some whites are persistently committed to “pro-Black” political measures. The pioneering work on racial prejudice explored opinion on busing school children, establishing the role of racial intolerance in policy opinion formation (Sears et al 1979). Similarly, the current paper limits its scope to examining the relationship between racial sympathy and support for racialized public policies. However, since many scholars have linked racial sympathy to the white vote for Obama, it is possible that the concept may also map onto candidate choice. I have conducted supplemental analyses to examine how racial sympathy shapes support for the country’s first black president.27 Future research should take up how or whether this relationship continues to 27 Using the ANES, I observe a significant relationship between racial sympathy and support for Obama in 2008. These results are presented in Web Appendix Table 5. As the table displays, sympathy is especially associated with affective evaluations of Obama more so than vote choice. Additionally, according to the 2008-09 ANES Panel, I find that the relationship between racial sympathy and partisanship intensified over the course of the 2008 campaign; as Obama emerged as 27 shape white partisan attachments. As the parties become increasingly racially polarized (Tesler 2016), it is possible that those whites who remain in the Democratic Party do so because they carry sympathetic attitudes toward African Americans, not because they are apathetic about race. Overall, then, this research builds on a vast, but ultimately narrow, group attitudes literature. As Krysan (2000) observes, “sociologists are almost always more interested in those individuals... that create some social problem…we focus almost all our efforts on understanding conservative racial policy attitudes” (160). My work does not dismiss the influence of racial antipathy but rather adds an original dimension to our understanding of racial attitudes. As long as race remains a “divide without peer” (Kinder & Sanders 1996) in American politics, it will behoove researchers to understand racial sympathy and its consequences. the Democratic Party’s nominee, the relationship between racial sympathy and Democratic Party identification strengthened. Additionally, there is some evidence that racial sympathy shapes support for Obama’s signature policy, the Affordable Care Act. Although racial sympathy is not related to support for health care in the 1994 GSS, the attitude is significantly associated with support for national health care in the 2008 and 2012 ANES, after Obama made the policy his own. This result complements work done by Tesler (2016). 28