“That’s when I was like, ‘I’m black,’ you know?” These are the words of Jorge, a university student living in Recife, a coastal city in northeastern Brazil. Like many others I met, Jorge tells me that he is classified as white on his birth certificate but that today he self-identifies as black. Recounting the details of his personal transformation, Jorge reports that while growing up he didn’t often think about himself in racial terms, per se, but over time he came to understand his past experiences as profoundly racialized. In Jorge’s words, he came to “discover [him]self” as black.
Tiago, also a university student in Recife, tells a similar story. Like Jorge, Tiago reports that he is classified as white on his birth certificate but today identifies himself as black. He explains that his racial transformation began when his friends from university inadvertently led him to a black movement event. There, Tiago heard the personal anecdotes of racism and discrimination shared by activists. He reports that he was surprised by how much their stories resonated with him because at the time he had not yet considered himself black. But identifying with these stories led Tiago to ask himself “how had I not realized this before?” He said, “I looked back said ‘jeez, that all happened to me because I was black. Because I am black.’ It was really just like that. It was a discovery.”
In many ways, there is nothing remarkable about the racial trajectories of these two young men in Brazil, a so-called “racial democracy” that often serves as a point of contrast to the rigid and institutionalized racial boundaries of the United States or South Africa. This context of fluid and ambiguous racial boundaries has long enabled individuals to cross racial boundaries and change their racial self-identifications – that is, to reclassify.Footnote 1 At the same time, however, this history of racial boundary crossing is also what makes these accounts significant. For although the Brazilian case is known for racial fluidity, it is also known for profound racial inequalities and veiled racism, which have long been said to incentivize reclassification toward whiteness, when possible.
Not so since the early 2000s. In more recent years, many Brazilians such as Jorge and Tiago have come to demonstrate a marked and growing tendency to reclassify instead toward blackness. Figure 1.1 shows the racial composition of Brazil from 1992 to 2019 as determined by the census bureau and illustrates this reversal of the status quo. Indeed, between the 2000 and 2010 censuses, the Brazilian population unexpectedly flipped from majority- to minority-white – a sudden structural shift that, as we will see, cannot be explained by intergroup differences in demographic trends or changes in census enumeration practices. What has instead become clear is that Brazilians are increasingly adopting the stigmatized labels of blackness.

Figure 1.1 Racial composition of Brazil, 1992–2019
This book analyzes this sudden reversal in patterns of racial reclassification, what I term the reclassification reversal, to shed new light on the processes of mass identity change and politicization. Empirically, its purpose is to explain why Brazilians are increasingly adopting nonwhite identities, why this tendency has increased so suddenly, and why they appear to be defying the conventional wisdom that expects reclassification toward whiteness. In seeking to provide satisfying answers to these questions, this book contributes to the broader theoretical agenda of understanding the “identity-to-politics link,” that is, the processes that translate social categories into politicized identities and into bases of group or collective politics. I focus particular attention on understanding one element of these processes, the formation of a group consciousness that shapes one’s perceptions and understandings of power – what I refer to as political identity. I leverage the puzzling variation evident in the reclassification reversal to distill empirically verifiable insights into the causes, mechanisms, and consequences of identity formation and politicization. The argument I develop directs attention toward the ways in which state-led efforts at educational expansion have reshaped individuals’ subjective self-understandings, led individuals to cross social boundaries they previously recognized, and imbued newfound identities with political meaning.
The Argument
I argue that the reclassification reversal is the consequence of expanded access to education, which has unintentionally led many Brazilians to develop racialized political identities. State-led efforts to better include lower-class sectors of the citizenry through social policy expansion have unleashed unprecedented waves of upward mobility for members of the lower classes, many of whom have options in their racial identifications and who are traditionally susceptible to practices known as whitening. Unprecedented access to secondary and university education, in particular, has increased the exposure of newly mobile citizens to information, social networks, and the labor market. In turn, this increased exposure has brought many face-to-face with racial hierarchies and inequalities in their pursuits of upward mobility, altering the personal experiences that inform their racial identifications and their political identities. The increasing adoption of nonwhite – and in particular black – identities can be understood as an articulation of newfound and racialized political identities.
From a macro perspective, the ostensibly sudden onset of the reclassification reversal can be explained by the timing of institutional reforms that expanded the nature and accessibility of social benefits allocated to citizens by state. Prior to redemocratization in the 1980s, literacy requirements for voting rights excluded large segments of the poor from the franchise, and social benefits were allocated on a corporatist basis, often accessible only to formal labor sectors and state-sponsored unions. The new and progressive 1988 constitution lifted literacy requirements for political citizenship and, moreover, codified universal social rights for all citizens, including the right to education. In the context of a newly expanded franchise, politicians and parties on the left and right faced pressure to compete for the votes of the poor masses, generating the political will and incentives to create, reform, and fund universal and targeted social policies. The federal government increased spending at all levels of education, restructured incentives for administrators, and altered pathways of resource delivery to circumvent political manipulation by subnational governments. Specifically at the university level, state and federal governments vastly increased enrollments at new and preexisting public universities, ensured greater inclusion through means and race-targeted affirmative action policies, and created financial programs to support students wishing to attend private universities.
The net result has been remarkable improvements in quality and access to education, even at the lowest levels of Brazil’s income structure. Household survey data indicate that among those in the bottom income decile, primary school completion rates increased from 10 percent in 1992 to over 65 percent by 2014; over this same period, high school completion increased tenfold in this decile, from 3 to 30 percent; and between 2000 and 2010 alone, overall university completion rates increased more than 65 percent among the adult population (25 years or older), from less than 7 to 11 percent. By 2010, university access more generally reached historically unprecedented levels, with an additional 25 percent of the adult population completing at least some university education.Footnote 2 Alongside these trends in educational expansion also came the stunning shift in patterns of racial reclassification. Analysis of reclassification between the 2000 and 2010 censuses estimates that the brown and black categories were, respectively, 10 and 30 percent larger than anticipated, and that these discrepancies were indeed due to mass reclassification (Miranda Reference Miranda2015).Footnote 3 It is no coincidence that these developments played out in tandem.
For decades, the Brazilian case has served as the perennial paradox in the comparative study of racial politics. Despite structural conditions of deep and durable racialized inequalities and the persistence of widespread discrimination, Brazil has often stood apart from similar cases for the absence of politicized racial differences or cleavages. From a strictly structural perspective, the politicization of racial differences in Brazil has long been overdetermined. But such expectations overlook the important role that citizenship institutions play in translating material conditions or group-based discrimination into political worldviews and action, into identity politics. Broadly construed, the argument I develop in this book highlights the ways in which citizenship institutions (the accessibility of education) and social structures (racial hierarchies and inequalities) interact to shape the microlevel processes of identity change and politicization.
Contributions
A Policy Feedback Account of Identity Politicization
This book contributes to theoretical debates on when, why, and how ethnic and other identities become politicized. In recent years, scholarship in the comparative ethnic politics literature has coalesced around instrumental explanations for the politicization or political salience of social identities, with a central focus on institutional incentives as the primary determinants of the identities articulated in the political arena.Footnote 4 These arguments encompass electoral rules that incentivize particular “minimum-winning coalitions” (Huber Reference Huber2017; Posner Reference Posner2005), norms or traditions inherited from colonial states (Laitin Reference Laitin1986), or incentive structures that simply render identities a convenient political means to some material end (Bates Reference Bates1974; Chandra Reference Chandra2004; Hoddie Reference Hoddie2006; Laitin Reference Laitin1998; Nagel Reference Nagel1996). But as we will see in the case of Brazil, it is not always safe to assume that identities are given, ready-made, and available for mobilization from above or below, or for the pursuit of material interests. Without consideration for the informal social hierarchies that can stigmatize social categories, identities that appear electorally or politically strategic might not in fact be viable bases for mobilization. Just as Madrid (Reference Madrid2012) argues regarding other contexts in Latin America, ethnoracial hierarchies and fluid boundaries have long deterred identification with devalued social categories, such as indigenous or black. In such a context, the willingness of individuals to adopt and articulate stigmatized identities in the political arena – whether these identities are at all available for mobilization – is also a function of the social disincentives perpetuated through institutional racism, disincentives that must be overcome.
My account highlights that social citizenship institutions can play a critical role in this regard. I build on the oft-cited tenet of policy feedback scholarship that “new policies generate new politics.” In this case, the policies I emphasize are what Marshall (Reference Marshall1950) classically termed social citizenship – the right to live as a full member of society according to prevailing standards, which encompass health, welfare, and not least of all education.Footnote 5 Specifically, I identify the ways in which educational policy reforms altered the quality and experiences of citizenship among Brazil’s poor masses, encouraging and empowering them to contest social hierarchies and articulate stigmatized identities in the political arena. At first glance, it may seem that this argument comports with instrumental explanations for identity salience, which point to state institutions as causal factors. But close examination of longitudinal patterns in the Brazilian case makes clear that citizenship institutions matter not by generating new incentive structures for black identification, per se, but by altering the personal and subjective experiences of citizens. In this case, reforms to educational policies and priorities expanded access to primary, secondary, and university education, unleashed waves of upward mobility for lower-class sectors, and set these citizens on new personal and professional trajectories. In the process, citizens gained greater exposure to information, social networks and movements, and the labor market, all of which altered the logics, discourses, and subjectivities that impact individuals’ self-understandings. In short, reforms to social citizenship institutions altered the personal and subjective experiences that inform citizens’ racial identifications and identities. In turn, these identities have “fed back” into the political process by impacting their willingness to articulate racial identities in the political arena. So conceived, the politicization of stigmatized identities and the reclassification reversal itself can be understood as consequences of the punctuated extension of (social) citizenship rights and benefits to formerly excluded sectors of society.
An Empirical and Systematic Analysis of the Identity-to-Politics Link
Beyond the comparative ethnic politics literature, this book also contributes to the interdisciplinary study of identity and group politics by providing a systematic and rigorous empirical analysis of identity formation processes. In recent years, prominent critics have urged scholars to avoid the analytical pitfalls of groupism and to disaggregate the processes of identity formation and politicization. Most famously, Brubaker and Cooper (Reference Brubaker and Cooper2000) criticize what they see as abuses of constructivism in the rapidly growing identity politics literature, citing scholars’ tendency to conflate categories with groups, and categories of analysis with categories of practice (also see Lee Reference Lee2008; Smith Reference Smith2004). Instead, scholars ought to more thoroughly scrutinize whether social categories qualify as “identities,” that is, categories of self-understanding “used by ‘lay’ actors … to make sense of themselves, of their activities, of what they share with, and how they differ from, others” (Brubaker and Cooper Reference Brubaker and Cooper2000, 4).Footnote 6 And, scholars ought to focus attention on understanding “the ways in which self-understandings may harden, congeal, and crystallize” (Brubaker and Cooper Reference Brubaker and Cooper2000, 1), and problematize, rather than assume, that microlevel identities scale up into fully fledged “groups” (Brubaker Reference Brubaker2004), or what Bartolini and Mair (Reference Bartolini and Mair1990) call “cleavages.” Lee (Reference Lee2008) echoes these calls, urging scholars to focus on what he terms the “identity-to-politics link,” the disaggregated sets of processes that lead from social categories to politicized identities, to group or collective politics.
Despite these prominent calls from scholars, we still possess relatively few empirical and systematic analyses of the microlevel processes of identity formation and politicization.Footnote 7 To be sure, this is a tall empirical order, as not just any context can offer empirical leverage in the way that Brazil’s reclassification reversal can. Fewer still pair this kind of phenomenon with a wealth of data sources that can be mined to shed light on these processes. Thus, one simple, but no less important, contribution of this study is the close, empirical examination of these distinct identity processes. By integrating and analyzing a wealth of data from multiple sources – including original qualitative data and survey experiments, longitudinal analysis of microlevel and municipal census data, national public opinion surveys, and a panel dataset of university students – we gain a comprehensive and empirical account of the identity-to-politics link.
A Reassessment of Racial Politics in Brazil
Finally, this book also contributes to the comparative racial politics literature, especially interdisciplinary debates on the significance and political relevance of race in the Brazilian and Latin American contexts. As Clealand (Reference Clealand2022) argues, while the study of race in comparative politics remains woefully underdeveloped, Latin American contexts provide rich and ample opportunity to probe, theorize, and test new theoretical perspectives on racial politics, especially regarding the consequences of fluidity and identification. In this regard, one contribution of this book is the empirical documentation of the reclassification reversal, which runs counter to scholarly wisdom. A long line of research in anthropology and sociology, in particular, established the idea that racial fluidity and black stigmatization produce whitening, especially among the upwardly mobile (Cardoso and Ianni Reference Cardoso and Ianni1960; Degler Reference Degler1971; Harris Reference Harris and Wagley1952; Silva Reference Silva1994; Wagley Reference Wagley and Heath1965). This study is not the first to take notice of and scrutinize the reclassification reversal (Jesus and Hoffmann Reference Jesus and Rodolfo2020; Miranda Reference Miranda2015; Soares Reference Soares, Theodoro, Jaccoud, Osorio and Soares2008). But it is the first to document these patterns and provide a theoretical and empirically tested account of its causes, mechanisms, and consequences for Brazilian (racial) politics. More so than the studies that precede it, this book probes and triangulates multiple types and sources of data to make sense of this new era in Brazilian racial politics. The central finding in this regard is that upward mobility does not inevitably produce whitening and has instead come to produce darkening in recent decades. What’s more, the coincidence of the reclassification reversal and unprecedented waves of upward mobility for lower-class Brazilians also suggests a parallel with cases like the United States, often discussed in juxtaposition to Brazil’s racial politics. To be sure, black-identified Brazilians do not exhibit the level of group cohesion witnessed in the United States. But just as the experiences of upward mobility were shown to deepen blacks’ racial consciousness in the wake of the Civil Rights Movement (Dawson Reference Dawson1995; Hochschild Reference Hochschild1995), so also have Brazilians exhibited greater racial consciousness in Brazil’s era of social inclusion.
This should not be taken to mean that Brazil’s racial politics now resembles, or will resemble, the hyper-politicization of race evident in the United States or South Africa. But falling short of the extreme outcomes in these canonical cases also should not overshadow this period of flux in Brazilian racial politics. In particular, this book’s findings regarding patterns of racial identification, the political content of racial identities, and the consequences for political behavior signal two important changes in this case. The first is growing heterogeneity within racial categories in racial subjectivities, that is, the logics and discourses individuals employ to rationalize their way into or out of racial categories. The traditional interpretation of Brazilian racial subjectivity emphasizes colorism, or racial classification based on fine, phenotypical distinctions made between individuals. But increasingly, and especially among black identifiers, colorist logic is being replaced (though not fully supplanted) by a political understanding of blackness based on shared experiences of racism and discrimination and the desire to contest racial hierarchies. Black movements in Brazil have long sought to promote this way of thinking to encourage race-based mobilization among the masses. But until recently, these efforts were met with resistance from the very communities they aimed to represent (Burdick Reference Burdick1998b; Hanchard Reference Hanchard1994). Yet more now than ever before, Brazilians are exhibiting a willingness to claim and politicize blackness.
Second and relatedly, there are clear signs that these race-conscious Brazilians carry these identities into the political arena and incorporate them into their political calculations. Political scientists have long remarked on the absence of racial politics in Brazil’s electoral arena, in particular. Despite profound structural inequalities, Brazil did not see the emergence of race-based political parties, nonwhite candidates struggled to win elections, and racial differences did not map onto partisan or electoral preferences (Mainwaring Reference Mainwaring1999; Mainwaring et al. Reference Mainwaring, Meneguello, Power and Middlebrook2000; Samuels and Zucco Reference Samuels and Zucco2018). But as I show in this book, the reclassification reversal is not a purely sociological phenomenon. Alongside this sea change in racial subjectivity has emerged a new, if overlooked, electoral constituency of highly educated black identifiers who prove themselves committed leftist voters. This is a pattern that emerges after the electoral realignment that followed the 2005 mensalão corruption scandal and is one that holds through the polarization of the 2018 election of Jair Bolsonaro. Taken together, this growing leftist constituency and new understandings of blackness reveal cracks in Brazil’s status as the perennial paradox in the comparative study of racial politics, a context decidedly lacking politicized racial differences. Simply put, it is no longer tenable to dismiss race as politically irrelevant in contemporary Brazil.
Alternative Explanations
The argument I advance in this book departs from dominant theories in the comparative ethnic politics literature to explain identity politicization. My focus is on social citizenship institutions, their impacts on microlevel subjectivities, and their consequences for identification and political behavior. This explanation is more compelling, I submit, because it better accounts for the timing of the reclassification reversal and the longevity of newly adopted racial identifications evident in the census. Though I dedicate significant attention to the effects of education on reclassification, let me also clarify that I do not intend to advance a monocausal theory of racial reclassification or identification. My argument is probabilistic and non-exhaustive; I do not argue that educational expansion explains reclassification in all cases nor that education supplants all other factors already known to impact racial identification. Indeed, a large interdisciplinary literature has well established that myriad factors beyond phenotype impact racial identification, including family socialization, social movements, and the media. I do not deny or dismiss the impact of these factors but argue instead that the effect of education matters separate and apart from such factors. Beyond these sociological arguments, my argument must also be situated against prevailing explanations from the literature, which face limitations in explaining this case of identity change and politicization.
Affirmative Action
Among those familiar with Brazil, the likely knee-jerk explanation for the reclassification reversal is the advent of affirmative action policies. Indeed, since the early 2000s, the federal and state governments in Brazil have begun to experiment with such race-targeted policies, predominantly in the form of quotas in university admissions. Scholars hypothesize two ways in which affirmative action can impact racial identification. The first comports with an axiom in the identity politics literature that institutions incentivize identification and identity salience (Chandra Reference Chandra and Chandra2012). In this view, identification and salience are the simple product of means-ends calculations in contexts of resource scarcity. Affirmative action policies have featured explicitly in this literature. Nagel (Reference Nagel1996), for example, argues that affirmative action incentivized Native American identification in the United States; Hoddie (Reference Hoddie2006) argues that preferential policies incentivized ethnic identities in China and Australia; and Chandra (2005) argues that in India, these policies incentivized excluded groups to mobilize and demand inclusion as policy targets. In the Brazilian case, journalists and public intellectuals echo this logic, often crying foul of the incentives for blackness and so-called fraud these policies are said to generate (e.g., Fry Reference Fry2007; Fry and Maggie Reference Fry and Maggie2004). But as we will see in Chapters 5 and 6, such crude instrumental motivations struggle to account for the long-term identity change in this context of racial fluidity and stigmatized blackness. In other words, such perspectives fail to fully consider the conditions and social forces that have long disincentivized blackness in the first place. Considered in its full context, affirmative action offers only short-term and risky benefits to manipulating one’s racial identification, and these simply cannot explain long-run behaviors that outlast short-term payoffs.
A second set of explanations based on affirmative action focuses on how states “make race” by naturalizing or making salient social boundaries and differences (Bourdieu Reference Bourdieu1985; Marx Reference Marx1998). In one vein, scholars have focused on official censuses as sites where states institutionalize and actively shape boundaries and identities (Kertzer and Arel Reference Kertzer, Arel, Kertzer and Arel2002; Lieberman and Singh Reference Lieberman and Singh2017; Omi and Winant Reference Omi and Winant1994). This line of argument finds a seemingly clear parallel in the Brazilian case, where scholars have pointed to the census as a major explanation for racial fluidity and weak racial consciousness (Loveman Reference Loveman2014; Nascimento Reference Nascimento2016; Nobles Reference Nobles2000). However, the state’s classification scheme and enumeration practices have remained unchanged in recent decades. Thus, these factors simply cannot account for the dramatic shifts evident over this period.
Yet another vein of state-centered arguments emphasizes the symbolic dimension of state policies and institutions and their impact on racial subjectivities. From this perspective, affirmative action policies are a prominent piece of a broader shift in symbolic state institutions and the state’s posture toward the racial question, which lend legitimacy to alternative racial discourses or make salient altered racial boundaries and identities. There is merit to this argument. The state’s racial stance has indeed shifted over the course of the twentieth century, from one of eugenic race science at the turn of the twentieth century, to the oft-touted colorblindness encapsulated in the idea of “racial democracy,” to a recognition of racial difference and discrimination perhaps best exemplified by affirmative action. The reclassification reversal coincides with this color conscious era, which also included other symbolic changes such as establishing a national racial consciousness holiday, mandating the teaching of African and Afro-Brazilian teaching in public school curricula, and founding a new governmental agency tasked with promoting racial equality. These are meaningful changes to the state. But while they might have contributed to a greater acceptance of blackness broadly, the major shortcoming of this symbolic institutional explanation is that diffuse, national-level factors are analytically too blunt to account for individual-level variation in reclassification; they simply cannot explain microlevel variation in reclassification. Though we should not deny their symbolic significance, we also must acknowledge that symbolic institutional explanations leave more to be said about who reclassifies and why they do so.
Of the alternative explanations outlined in this introduction, I devote most attention to those related to affirmative action in the chapters that follow. In Chapter 5, I analyze the hypothesized effects of educational expansion prior to the passage of the 2012 federal affirmative action law, and show that these findings are not contingent on the presence of state-level affirmative action policies. In Chapter 6, I employ a variety of empirical strategies to directly test the effects of affirmative action, separate and apart from the effects of educational expansion. My analyses provide mixed support, at best. While there is evidence that affirmative action amplifies the effects of educational expansion, affirmative action policies are neither necessary nor sufficient conditions for the reclassification reversal.
Mobilization from Above
A prominent line of research in identity politics, and ethnic politics in particular, attributes the politicization of identities to political elites. Particularly in contexts where elites seek political office to gain access to patronage or state resources, scholars argue that politicians politicize or incentivize identification through top-down electoral mobilization. In this view, rent-seeking politicians mobilize social cleavages that are sizable enough to win elections and maximize distributive payoffs for voters and elites (Huber Reference Huber2017; Posner Reference Posner2005; also see Chandra Reference Chandra2004). In these instrumental accounts, political elites are the central agents, and voters comply in order to receive their own post-electoral payoffs. These theories assume stability in the size and boundaries of social groups that are problematically unstable in the Brazilian case. But the implication nonetheless is that elites may have incentivized nonwhite identification by promising group-targeted benefits in electoral campaigns.
These explanations, however, fall short in this context. As noted earlier, scholars of Brazilian electoral and party politics have traditionally argued that there are few social bases in Brazil’s fragmented electoral system, let alone racial ones. In more recent years, scholars have identified class and region as salient electoral differences (Handlin Reference Handlin2013; Hunter and Power Reference Hunter and Power2007). But mainstream studies conducted in the decade following redemocratization identified few sociodemographic correlates of electoral preferences, leading scholars to conclude that politicians have not courted votes along group lines (Mainwaring Reference Mainwaring1999; Mainwaring et al. Reference Mainwaring, Meneguello, Power and Middlebrook2000; Samuels Reference Samuels2006). Moreover, scholars argue that the persistence of traditional, clientelistic politics has disarticulated social differences (Hagopian Reference Hagopian1996). As I discuss in Chapter 7, racial politics scholars doubt some of these claims from the party and electoral systems literatures and have identified salient racial differences in electoral behavior even at the height of state’s embrace of colorblind racial democracy discourse (Castro Reference Castro1993; Soares and Silva Reference Soares and Silva1987; Souza Reference Souza1971; Valente Reference Valente1986). But even these scholars agree that, while race has been too quickly dismissed by political scientists, there is little evidence to suggest race forms the basis of top-down electoral strategy.Footnote 8 Indeed, nonwhite politicians are woefully underrepresented among officeholders in Brazil (Janusz Reference Janusz2018; Johnson III Reference Johnson III1998, Reference Johnson III, Heringer and Johnson III.2015), and those who win office do so by avoiding explicit racial appeals in their campaigns (Mitchell Reference Mitchell2009; Oliveira Reference Oliveira2007; Valente Reference Valente1986). From the electorate’s perspective, it remains unclear that voters prefer candidates of their race or color (Aguilar et al. Reference Aguilar, Cunow, Desposato and Sangali Barone2015; Bueno and Dunning Reference Bueno and Dunning2017), and evidence from Brazil and elsewhere in Latin America suggests all voters simply prefer lighter-skinned candidates (Contreras Reference Contreras2016; Janusz Reference Janusz2018).Footnote 9 Recent research by Janusz (Reference Janusz2021, Reference Janusz2023) indicates that elites do respond strategically to racial considerations by changing their own racial identifications. But, Janusz argues, the causal direction is reversed: demographic structures impact candidates’ racial identifications, not the other way around. But above all, there simply has been no significant or recent pattern of explicit, top-down electoral mobilization of blackness in Brazil. Electoral mobilization from above finds little traction on this question.
Mobilization from Below
Another alternative explanation attributes identity politicization to social movements and bottom-up mobilization. Of particular note is McAdam’s (Reference McAdam1982) seminal study of black organizing and political cohesion in the U.S. Civil Rights Movement. For McAdam, one important function movements provide is cognitive liberation, or the ability for aggrieved or oppressed populations to “define their situations as unjust and subject to change through group action” (McAdam Reference McAdam1982, 51). Thus, social movement organization entails the activation and cohering of identities for the purpose of pursuing political action. Other scholars agree, arguing that social movement participation itself can play a role in fomenting political consciousness and in leading individuals to develop and articulate interests and claims in the political arena (Klandermans Reference Klandermans, Morris and Mueller1992; Roberts Reference Roberts1998; Stokes Reference Stokes1995).Footnote 10 Other prominent social movement scholarship has promoted the view that collective and politicized identities are both causes (Friedman and McAdam Reference Friedman, McAdam, Morris and Mueller1992; Klandermans Reference Klandermans2002) and consequences (Escobar and Alvarez Reference Escobar and Alvarez1992) of mobilization.Footnote 11 As I detail in Chapters 3 and 4, social movements and new social networks via the pursuit of education forms part of the exposure-via-education argument. And there is no denying that many racially conscious Brazilians seek out social movements as venues in which to articulate racialized political identities. But the role of social movements is, in and of itself, an unsatisfactory causal explanation for two primary reasons.
First, I argue that the educational services the black movement organizations provide to low-income households play an important role in these processes, and research discussed earlier has shown that social movements are certainly capable of fomenting consciousness and inspiring identity change. But it remains unclear that the Brazilian black movement could or would have had such a large effect on the mass public in the absence of broader educational reforms that expanded access to primary and secondary education and created demand for university access. As I detail in later chapters, several of my interviewees discussed their contact with the black movement, reporting that this raised questions and doubts about themselves, altered their ways of thinking about blackness, and fomented a racialized view of Brazilian society. But many interviewees also explained that their first contact with the movement came inadvertently, often through new social contacts from university or in pursuit of low-cost services, like preparatory courses for the university entrance exam. Without the prior reforms that increased enrollments in and completion of primary and secondary schooling, demand for higher education and the altering of social networks would certainly have been less, and consequently so also this inadvertent exposure to the black movement. Educational reforms were crucial to generating bottom-up demand for universities, and this – indirectly – increased exposure to social movements. This should not be read as a dismissal of the important role that black movements play in the reclassification reversal. But for the reason outlined here, I theorize social networks and movements as one pathway through which education exerts a causal effect rather than as an independent cause that operates distinctly from education, or that would have succeeded in bringing about the reversal absent broader educational reforms.
Second, the challenges that Brazil’s black movement has historically faced in mobilizing the masses also raise doubts about the movement as an independent cause of sudden mass change. No doubt, social movements broadly have been key actors in Brazil’s major political developments and institutional reforms, including democratization and the constitutional convention in the 1980s (Alvarez Reference Alvarez1990; Garay Reference Garay2016; Keck Reference Keck1992). Specifically, the black movement has more recently exerted significant influence by shaping policy at the highest levels of the Brazilian government, pressuring the state to acknowledge racial discrimination and adopt affirmative action, and occupying key offices in federal government agencies since the 2000s (Htun Reference Htun2004; Paschel Reference Paschel2016a). But while successful in influencing elites and policy at the national level, the black movement has historically struggled to mobilize the masses at the grassroots (Hanchard Reference Hanchard1994; Marx Reference Marx1998). Scholars disagree on why, but one compelling explanation advanced by Burdick (Reference Burdick1998b, Reference Burdick1998a) is that the movement’s leadership and rank-and-file tend to be populated by middle-class professionals, who face resistance from the masses to being mobilized along racial lines. This is in part, Burdick argues, because the movement’s prerequisite that participants adopt politicized black identities is a hurdle too great for Brazil’s darker-skinned masses, who prefer to distance themselves from blackness – a point echoed more recently by Alves (Reference Alves2018).Footnote 12 While race-based activism and mobilization occur in some interest arenas (Caldwell Reference Caldwell2007; Perry Reference Perry2013; Smith Reference Smith2016), it remains unclear how widespread black movement participation has become among the mass public. It is thus important to draw a distinction between influencing elites and mobilizing the masses. Social movements have recently made great strides, but the black movement in particular has long struggled to mobilize and connect with the mass constituencies it aims to represent. What has changed in recent times, I argue, is the receptiveness of certain sectors of the public to the black movement’s message when they encounter it, a change brought about by upward mobility and educational expansion. For this reason too, social movements are better conceived as one possible (but not exclusive) pathway of exposure-via-education, rather than an independent or competing cause of wide-scale reclassification.
Cleavage Structure
Finally, structural theories attribute identity salience to cleavage structure, or the degree to which multiple group memberships coincide with or cut across one another. Most prominently, Horowitz (Reference Horowitz1985) argues that conflict between groups is more likely when groups are organized vertically, in some form of hierarchy (also see Dunning and Harrison Reference Dunning and Harrison2010; Rogowski Reference Rogowski1990). Cederman et al. (Reference Cederman, Gleditsch and Buhaug2013) similarly argue that civil war onset is more likely when political and economic inequalities map onto group lines. The logic here is relatively simple: objective structural conditions of group-based inequality produce identity politics.Footnote 13 Indeed, Brazil’s income distribution is among the most unequal in the world (Hagopian Reference Hagopian and Ames2018; Lustig Reference Lustig2015; Morley 2001), and class inequalities have historically mapped cleanly onto racial hierarchies (Andrews Reference Andrews1991; Hasenbalg Reference Hasenbalg1979; Silva Reference Silva1978; Telles Reference Telles2004). Moreover, like the United States and South Africa, Brazil continues to contend with the legacies of colonialism and the practice of slavery. Again, the politicization of racial differences in Brazil has long been overdetermined.
Yet, prior to the 2000s, scholars remarked on the weak politicization of race in Brazil, despite these conditions (Hanchard Reference Hanchard1994; Lieberman Reference Lieberman2003, Reference Lieberman2009; Marx Reference Marx1998). The coincidence of race and class cleavages simply cannot account for the timing of the reclassification reversal in Brazil. What’s more, the period of time that coincides with the reclassification reversal was one of declining inequality and a narrowing of the gap between white and nonwhite Brazilians (IPEA 2016; Klein et al. Reference Klein, Mitchell and Junge2018; López-Calva and Lustig Reference López-Calva and Lustig2010; Neri Reference Neri2011). If anything, the structural coincidence of race and class that are said to underpin identity politicization was becoming increasingly crosscutting just as racial identities have become more politicized in Brazil. As we’ll see, Brazilians exhibiting a politicized racial consciousness often point to structural conditions to legitimate the view that race deserves greater attention in Brazilian politics, but the structure of racial and class cleavages in particular cannot account for the timing of the reclassification reversal. To boot, like symbolic national institutions, structure also cannot explain microlevel variation in who chooses to reclassify or assume a politicized black identity.
Research Design and Methods
This book employs multiple methods and analyzes multiple sources of data to provide support for the educational expansion argument and assess alternative explanations. In particular, I build and test this argument by relying on qualitative and quantitative data collected during more than 18 months of in-depth fieldwork in São Paulo and Recife – two major urban centers located in culturally distinct and the most densely populated geographic regions of Brazil. The data I present in this book is the product of sequential stages of data collection, beginning with inductive field research that included participant observation and in-depth interviews with reclassifiers. This initial field research generated the hypothesis that better-educated Brazilians were most likely to adopt nonwhite identities over time, and provided additional avenues of inquiry to situate these patterns in Brazil’s historical and macro-level context. Having refined and developed the hypothesis, I then set out to test these insights rigorously and systematically with a wealth of additional data. In the end, the data analyzed in this book include:
Ethnographic interview data with thirty-four Brazilians of various educational attainments, including both reclassifiers and non-reclassifiers;
A synthetic panel dataset of birth cohorts, comprised of more than 137,000 observations, constructed from annual demographic surveys between 1992 and 2015, as well as additional surveys from 1976 to 1990;
A multilevel panel dataset of 5,500 municipalities in Brazil between 1991 and 2010, as well as state-level affirmative action laws, decrees, and policies;
A panel dataset of federal university students’ racial identifications in high school and when registering for the university entrance exam, constructed post hoc from the Ministry of Education’s embargoed surveys of public-school students;
Survey experiments conducted face-to-face with randomly sampled Brazilians in 2002 and 2018; and
Public opinion surveys, including sociological studies of racial identification collected between 1986 and 2008, surveys of protesters from 2013 to 2016, and national surveys for five presidential elections between 2002 and 2018, among others.
This book joins other recent books that leverage longitudinal and subnational variation in a single context to distill generalizable theoretical insights (see Pepinsky Reference Pepinsky2019). For these purposes, the Brazilian case is particularly well positioned to shed light on the processes of identity formation and politicization. Indeed, to the extent that Brazil has appeared in the literature on ethnic and racial politics, scholars have noted the weak politicization of racial differences (Bueno and Dunning Reference Bueno and Dunning2017; Hanchard Reference Hanchard1994; also see Yashar Reference Yashar2005). In comparative studies, Brazil is typically analyzed as a “negative” case lacking racial politics, one that demands explanation or offers crucial variation on variables of theoretical interest (e.g., Lieberman Reference Lieberman2003, Reference Lieberman2009; Marx Reference Marx1998; Nobles Reference Nobles2000).Footnote 14 By and large, scholars have agreed that race has been politically relevant in this context only insofar as elites have disarticulated racial differences by constructing a racially inclusive nation that whitewashes Brazil’s history as the single largest and longest-running participant in the slave trade, as well as the legacies of slavery in shaping present-day inequalities (Andrews Reference Andrews1991; Loveman Reference Loveman2014; Skidmore Reference Skidmore1993; Telles Reference Telles2004). Previous scholarship suggests, therefore, that this ought to be an unlikely case for the formation of political identities rooted in racial categories of social membership. Yet at the same time, the fluidity of racial boundaries renders the Brazilian case “exceptional” (Pepinsky Reference Pepinsky2017), one where such identity change is not only possible on such a wide scale, but possible to detect and analyze empirically and systematically.
The downside to mining a single domestic context for data and insights is that doing so potentially limits whether these findings generalize to other contexts. But there are two reasons to not so quickly dismiss this phenomenon or these findings as singularly Brazilian. The first pertains to the applicability of these findings elsewhere in Latin America, a region that has undergone a sea change in its ethnoracial politics. Over the past three decades, the region has witnessed the rise of indigenous social movements and political parties, the constitutional codification of ethnoracial rights, and the implementation of race-targeted affirmative action policies. Brazil has charted some of this new territory. But these significant changes have arrived to a wide array of countries, including Bolivia, Ecuador, Peru, Guatemala, Colombia, and Mexico. What’s more, national census bureaus across the region have begun to more uniformly and consistently collect ethnoracial population data, rendering visible for the first time, in some cases, ethnoracialized populations previously obscured by national myths of racial unity. Mexico’s census bureau has even gone so far as to include measures of skin tone in its national household surveys – an indication of the state’s greater willingness to recognize and potentially redress racialized inequities.
More data ought to facilitate extensions of the longitudinal analysis I conduct of Brazil to other cases, as well as mapping of variation across cases. And while data collection efforts do not yet permit longitudinal comparisons in the same depth across the region as a whole, there are nonetheless indications of significant growth in black (Afro-descendant) and indigenous identification in a number of cases. This is true even outside of Brazil and the Caribbean, where Afro-descendant identifiers already comprise large shares of national populations. Indeed, Table 1.1 shows that between countries’ two most recent censuses, the relative size of self-identified indigenous populations grew at significantly high rates in Argentina, Colombia, Costa Rica, Guatemala, Paraguay, and Venezuela.Footnote 15 These proportions more than doubled in Brazil and Nicaragua and nearly tripled in Chile. Similarly, the number of self-identified black or Afro-descendant identifiers grew significantly in Guatemala, Puerto Rico,Footnote 16 and Mexico, more than doubled in Ecuador, and tripled in Uruguay – a country that once celebrated itself as “the white nation” (Andrews Reference Andrews2010). To be sure, variation in enumeration practices in these cases complicates inter-census and cross-case comparisons, so these statistics should be interpreted with caution. But nonetheless, these statistics seem to suggest that the patterns and trends that the Brazilian case has thrown into sharp relief are potentially representative of broader, region-wide change.
Table 1.1 Relative proportion of black (or Afro-national) and indigenous populations in Latin America
Indigenous ID | Black or Afro-national ID | |||||
---|---|---|---|---|---|---|
2000s Census | 2010s Census | % Change | 2000s Census | 2010s Census | % Change | |
Argentina | 1.55% | 2.47% | 59 | −−− | 0.39% | −−− |
Bolivia | 61.97 | 40.28 | –35 | −−− | 0.23 | −−− |
Brazil | 0.20 | 0.43 | 115 | 6.21 | 7.52 | 21 |
Chile | 4.58 | 12.44 | 172 | −−− | 0.06 | −−− |
Colombia | 3.35 | 4.31 | 28 | 10.40 | 6.75 | –35 |
Costa Rica | 1.68 | 2.42 | 44 | 1.91 | 1.05 | –45 |
Ecuador | 6.83 | 7.03 | 3 | 2.23 | 5.25 | 135 |
El Salvador | −−− | 0.23 | −−− | −−− | 0.13 | −−− |
Guatemala | 39.41 | 43.43 | 10 | 0.04 | 0.32 | 606 |
Honduras | −−− | 7.25 | −−− | −−− | 1.39 | −−− |
Mexico | 9.23 | 21.50 | 133 | 1.16 | 2.15 | 86 |
Nicaragua | 1.84 | 4.43 | 141 | −−− | −−− | −−− |
Panama | −−− | 11.91 | −−− | −−− | 8.94 | −−− |
Paraguay | 1.19 | 1.73 | 45 | −−− | 0.06 | −−− |
Peru | −−− | 25.80 | −−− | −−− | 3.57 | −−− |
Uruguay | −−− | 4.90 | −−− | 2.00 | 7.84 | 292 |
Venezuela | 2.22 | 2.66 | 20 | −−− | 3.60 | −−− |
Puerto Rico | 0.17 | 0.17 | 1 | 7.57 | 11.30 | 49 |
Black categories exclude mestizo/a, mulato/a, and other mixed categories except in Colombia, where black and mulato/a categories are lumped in the census. Statistics come from national census bureaus in each country, and the decennial census in nearly all cases. Exceptions include Puerto Rico, for which statistics come from the 2010 and 2020 American Community Survey (identification as black or African American alone); Mexico, for which the 2000s figure comes from the census bureau’s annual household survey; and Uruguay, the 2000s figure for which is also drawn from the annual household survey, as presented in Andrews (Reference Andrews2010, table I.1). Guatemala’s 2018 black population figure combines black and Garifuna identification. The Dominican Republic did not collect ethnoracial data in 2002 or 2010. See Supplementary Tables A5 and A6 for detailed information on sources.
Latin America is also far from the only region where one encounters fluid ethnic or racial boundaries. Indeed, analyses of the United States – where boundaries are believed to be rigid – have documented racial reclassification following changes to racial measurement on the census (Davenport Reference Davenport2018, Reference Davenport2020; Masuoka Reference Masuoka2017), as well as prior to these changes in enumeration practices (Waters Reference Waters, Denton and Tolnay2002). Beyond the Western Hemisphere, the type of identity change manifest in reclassification also finds a parallel in the adoption of the national identities of titular nations in Laitin’s (Reference Laitin1998) post-Soviet republics of Eastern Europe. And similar forms of ethnic reclassification have also occurred with non-Han ethnic minorities in China, aboriginal populations of Australia, bumiputeras (sons-of-the-soil) in Malaysia, and Assamese linguistic identities in India (Hoddie Reference Hoddie2006). Even Horowitz (Reference Horowitz1985), in his classic tome on intergroup conflict that spans much of the global south, acknowledges that “[i]t is not merely what is asked and how the [census] results are to be interpreted that counts. Individual answers are also manipulable, since there is an element of self-definition in ethnic affiliation” (195). In short, the instability in social boundaries and demographic structures as constituted by censuses is more widespread than is commonly acknowledged by political scientists. This study represents an effort to center and make sense of this variation in so-called “political demography” (Hoddie Reference Hoddie2006; McNamee and Zhang Reference McNamee and Zhang2019).
A second potential limitation to generalizability stems from the institutionalization of racial boundaries, which shapes not only how easily boundaries can be crossed in a given context, but the extent to which racial categories are top-of-mind among citizens, whether they “see” race at all (Brubaker et al. Reference Brubaker, Loveman and Stamatov2004). Where boundaries are less porous and racial membership rules are more strictly enforced, one is unlikely to find such wide-scale and uncontroversial reclassification as one does in Brazil. But even if such boundary crossing does not occur everywhere, there is good reason to expect the mechanisms and processes that underlie reclassification – the processes of identity formation and politicization – to generalize to other contexts or other categories of social membership, especially stigmatized categories. The transformations and processes that I detail in this book are fundamentally about how certain types of social differences become the basis of individuals’ self-understandings (identities), how those self-understandings come to shape their beliefs about power (become politicized), and how these then shape political action and behavior (articulation). In this case, I argue that reclassification is an expression, or indicator, for these identity processes. But I do not argue that reclassification is a necessary ingredient or output of all identity processes pertaining to other kinds of social categories. What is more easily detected in the Brazilian case may be harder to detect in other contexts and with other social categories, and their empirical manifestations may well vary. Yet the mechanisms and processes of exposure could easily generalize to subordinate gender or sexual identities, for example. In the book’s conclusion, I return to this discussion and discuss possibilities for generalizing the mechanisms and processes I identify to other stigmatized or subordinate social groups.
Positionality
As with any study, but particularly when the analyst is directly involved in data collection, we must carefully consider our positionality, that is, how one’s presentation in racial, gender, class, and other terms can impact the methods and approaches one chooses, the data and information one can access, and one’s interpretation of that information. There is no doubt that my presentation as a light-skinned, foreign researcher from the United States impacted avenues for data collection and the willingness of individual interlocutors to share their perspectives with me. This is especially pertinent to the one-on-one interviews that I personally conducted for this research. Though at times I relied on the help of Brazilian research assistants who aided this project in various ways (including conducting focus group interviews), the qualitative interview data I present throughout this book come from the in-depth interviews I conducted. Even under the best of circumstances, ensuring interviews come to fruition requires incredible persistence and logistical flexibility. But despite my best efforts, there were times where potential interviewees I pursued declined to be interviewed, often offering a polite explanation for why they could not participate. I cannot know for certain why these individuals did so. In one instance, I was made aware that a group of student activists with whom I had made contact discussed amongst themselves whether or not they would agree to speak with me about my research. In the end, many did, and these were among my most fruitful interviews. But in any case, the relevant point is simply that I cannot know what information I failed to get, or how my presence may have impacted the narratives told by my interviewees.
By the same token, however, it would be disingenuous to say that my positionality served solely as an obstacle to conducting this research. Indeed, the combination of my being read as white and introducing myself as American also served as an asset. Fulfilling the stereotypical image of the white American in the minds of Brazilians undoubtedly opened doors, greased wheels, and frankly generated some excitement on the part of my interlocutors. While some interviewees may have been turned off by my positionality, others expressed intrigue about the foreigner from the north who took an interest in their racial transformations. All this is not even to mention the countless other ways that my status as an American researcher aided my research, especially when contacting government agencies, politicians, and bureaucrats whose assistance and insights I also relied on.
Regardless of whether my positionality was an asset or an obstacle, the methodological concern is simply that this introduced bias into data collection and my analysis of that and other data. There is no denying or skirting this issue. Such bias is hard to measure or quantify, and as a counterfactual is impossible to observe. But the concern of bias is partly what motivates the triangulated and multi-method research design of this project. By relying on different methods employed sequentially, I am able to partially mitigate some of the effects of bias by verifying and cross-checking the information collected qualitatively with independent sources of data and analysis, including macro-level and microlevel census and survey data. As a result, the argument and findings I develop in this book do not hinge on any one particular method or piece of data, and instead piece together conclusions based on multiple independent data sources. While my analyses will no doubt raise questions that future analysts will further probe and interrogate, this research design offers some measure of insurance against the influence of my positionality on the findings.
Plan of the Book
The chapters that follow define the puzzle at the heart of this book, develop the conceptual and theoretical argument, and provide empirical support for the causes, mechanisms, and consequences of the reclassification reversal. Chapters 2 and 3 elaborate the book’s empirical, conceptual, and theoretical framework. Chapter 2 lays out the empirical and theoretical puzzle and lays to rest simple explanations for these patterns based on census enumeration practices and intergroup differences in demographic trends. I then situate these patterns historically, and emphasize that conventional wisdom in sociology and anthropology would not have predicted the reclassification reversal. With the puzzle established, Chapter 3 elaborates the educational expansion argument introduced here. I detail the macro-level institutional reforms responsible for setting off the reclassification reversal, and situate recent improvements in educational access against the backdrop of the twentieth century. I then transition to the microlevel, identifying and specifying the specific causal pathways through which educational expansion impacts individuals’ racial subjectivities: by increasing their exposure to new information, social networks, and experiences in the labor market.
Chapters 4 through 7 present empirical analyses of the causes, mechanisms, and consequences of the reclassification reversal. Chapter 4 focuses on specifying and illustrating the causal pathways that link education to reclassification and racial consciousness at the individual level. I present qualitative evidence from in-depth interviews with reclassifiers. These data, which helped to generate the central hypothesis of this book, bring the theoretical mechanisms to life by illustrating what the processes of reclassification look like on the ground. In this chapter, I also present systematic tests of the mechanisms by analyzing public opinion surveys from 1986 to 2008. I show that greater education is indeed correlated with racial consciousness and black identification, but that this relationship only emerged as access to education became more inclusive in the 2000s. Chapter 5 focuses on testing the implications of the educational expansion hypothesis more rigorously. I conduct longitudinal analysis of a pseudo-panel of birth cohorts to show that better-educated Brazilians are those most likely to adopt nonwhite, and especially black, identities over time. I supplement this analysis with panel analysis of Brazilian municipalities, which lends further support to the argument.
Chapter 6 focuses on alternative arguments based on affirmative action. Drawing on original survey experiments and two panel datasets, I assess instrumental and symbolic arguments pertaining to these policies. My analyses include priming and list experiments, as well as difference-in-differences analyses of state-level affirmative action policies and the federal affirmative action law on university students. These analyses provide mixed support, at best, for these alternative arguments, but nonetheless show some added effect of these policies on reclassification. Overall, they suggest that affirmative action cannot be ruled out, but also is not central, to the reclassification reversal.
Chapter 7, the final empirical chapter of the book, draws out the implications of the reclassification reversal for political behavior in Brazil and situates these implications in the context of five presidential elections between 2002 and 2018. Contrary to the common view that race is irrelevant in Brazilian politics, my analyses reveal that highly educated black identifiers – those most likely to exhibit racial consciousness – have come to constitute a loyal leftist constituency in the Brazilian electorate. Finally, the concluding chapter, Chapter 8, situates my findings within and against prevailing theories in the comparative ethnoracial and identity politics literatures, and offers possibilities for extrapolation and generalization from this case. I conclude the book with reflections and speculation on what these findings may hold for the future of Brazilian politics.
Notes on Racial Terminology and Official Racial Categories
Before proceeding, it is worth clarifying my analytical focus on official census categories and the racial terminology I employ throughout this book. First, the analytical and empirical focus of this book is on changes in how individuals choose to self-classify in the official racial categories determined by the state. This is motivated by the empirical puzzle centered in this study, and should not be interpreted to mean that identification with official census categories is the most appropriate, or singularly comprehensive, way to measure racial identity in this or any context. Indeed, a rich social science literature has analyzed the resonance of official categories with those employed colloquially (Bailey et al. Reference Bailey, Fialho and Peria2018; Harris et al. Reference Harris, Consorte, Lang and Byrne1993; Sheriff Reference Sheriff2001; Telles Reference Telles2004), and the group consciousness literature reminds us that identification is but one aspect of social identities, which vary in their strength of attachment as well as political content and attitudes. Where possible, my analyses go beyond census identification and include analysis of multidimensional measures of racial identity. But by and large, my analyses focus on census categories for the simple and practical reason that they provide the only consistent and longitudinal empirical data source for analyzing and testing claims about the reclassification reversal. A full accounting of the reversal’s implications for colloquial racial terminology and discourses would surely be welcome, but this lies beyond the scope of the study at hand.
Second, those familiar with Brazil and other contexts in Latin America know that racial labels must be employed with care and precision. Because my focus is on patterns of census reclassification, I refer to official census categories when I use the words white, black, and brown (branco, preto, and pardo, respectively). Simply to avoid confusion, I steer clear of other racial labels despite their widespread usage in the Brazilian context. These include the notoriously ambiguous term moreno (roughly meaning dark) and negro, a label promoted by the black movement that is generally understood to encompass all Afro-descendants, but that can also stand in for black (preto) colloquially. When English translation is unclear or obscures some other meaning, I refer to the original Portuguese parenthetically. Additionally, I use the term “mixed-race” interchangeably with brown. And when I use the term “nonwhites,” I refer to black and brown identifiers together. I do not include yellow or indigenous census categories among nonwhites, simply because these samples are too small to analyze with precision in my analyses. As a general rule, I refrain from referring to “whites” or “blacks” as self-evident categories or groups (Brubaker Reference Brubaker2004), referencing instead white or black identifiers. I do this in recognition of the fact that racial categories are simply labels, not inherent or immutable characteristics. For the sake of readability, I at times refer to “black voters,” for example. In these instances, such references indicate voters who identify with particular racial categories.