The politics of a datascape transformed: ethnoracial statistics in Brazil in regional comparative perspective

Abstract

The first decades of the 21^st century witnessed a dramatic transformation of the datascape for the production of ethnoracial statistics in Latin America. This transformation is most strikingly evident in national censuses in the region. A few decades ago almost no Latin American countries included questions about race or ethnicity in their national census. By the 2010 census round, almost every country in the region collected information about ethnic or racial identification. This article argues that the production of ethnoracial population data in Latin America is both product of politics and productive of politics. Ethnoracial statistics contribute to shaping the terrain they seem to merely describe, but not always in ways that are predicted or intended. Placing the Brazilian experience in regional comparative perspective illuminates how official ethnoracial statistics can produce outcomes that are simultaneously productive and counterproductive to the aims of those who struggle for their production in the first place.

Keywords
demographic data; ethnoracial population data; Afrodescendents; indigenous populations; Latin America

Resumo

As primeiras décadas do século XXI presenciaram uma dramática transformação do cenário de dados para a produção de estatísticas etnorraciais na América Latina. Essa transformação é mais evidente nos censos nacionais da região. Há algumas décadas, quase nenhum país latino-americano incluía questões sobre raça ou etnia em seu censo nacional. Já no censo de 2010, quase todos os países da região coletaram informações sobre identificação étnica ou racial. Este artigo argumenta que a produção de dados populacionais etnorraciais na América Latina é tanto um produto da política quanto produz a política. Tais dados contribuem para moldar o terreno que parecem meramente descrever, mas nem sempre da maneira prevista ou pretendida. Colocar a experiência brasileira em uma perspectiva comparativa regional revela como essas estatísticas oficiais podem produzir resultados que são simultaneamente producentes e contraproducentes para os objetivos daqueles que lutam, primeiramente, por sua produção.^◊

Palavras-chave
dados demográficos; dados populacionais etnorraciais; afrodescendentes; populações indígenas; América Latina

Introduction: an avalanche of ethnoracial population data

The first decades of the 21^st century witnessed an unprecedented explosion of officially produced ethnic and racial demographic data in Latin America. This avalanche of ethnoracial statistics stems largely from the dramatic shift in how Latin American states classify and count their populations on national censuses.¹ Since the 2000s, almost every country in the region modified its national census to collect new types of data about individuals’ ethnoracial identification or color. Across Latin America, states that had long refrained from collecting ethnoracial statistics in national censuses shifted course, adopting new questions that recognize ethnic or racial difference within their populations.

The rather sudden regional embrace of ethnic and racial data collection on censuses in the first decades of the 21^st century is summarized in Table 1. The shaded cells in Table 1 indicate that a country took a national census in that decade. A white circle indicates that the census included a question that made indigenous populations statistically visible. A black circle indicates that the census included a question that made black or Afro-descendent populations statistically visible.

Table 1
Visibility of Afrodescendant and Indigenous peoples in Latin American Censuses, 1980-2020s

In the first decades of the 21^st century, Latin America’s national censuses increasingly made visible individuals who identify as indigenous or Afro-descendent. In the 1980s, approximately half of Latin American countries identified indigenous populations in the context of national censuses. By 2010, almost all of them had done so or planned to do so in the next census round. With respect to Afro-descendent populations, in the 1980s, only two countries – Brazil and Cuba – included census questions that differentiated them from others in the population. By 2010, nearly every Latin American country included a census question to count black or Afro-descendent identifying individuals, or planned to include such a question in their next census.² Taken together, the changes to Latin American censuses in this period marked a dramatic transformation of the datascape for the collection and production of ethnoracial demographic statistics.

This article argues that the transformation of the datascape for the production and use of ethnoracial population data in Latin America is both product of politics and productive of politics. The transformation breaks with several decades of official color-blindness in census enumeration in much of the region, a break that is attributable first and foremost to political struggle. Once in place, meanwhile, the transformed datascape contributes to shaping the ethnoracial realities it claims to merely describe, affecting the contours of the political terrain in which official ethnoracial statistics are produced and deployed. Placing the Brazilian experience in regional comparative perspective illuminates how official ethnoracial statistics can produce outcomes that are simultaneously productive and counterproductive to the aims of those who struggle for their production in the first place.

Historical background: national censuses and the idea of progress as demographic whitening

To appreciate how politically momentous the late 20^th and early 21^st century transformation of ethnoracial data production in national censuses is for Latin America, it is helpful to recall the longer history of ethnodemographic statistical production in the region. The recent shift in Latin America’s national censuses to include questions about race and ethnicity marks a radical departure from the dominant practice in most of the region since mid-century. Prior to the 1960 census round, however, the collection of ethnic and racial statistics in Latin American national censuses has substantial historical precedent.

Throughout the 19^th century and the first half of the 20^th, most Latin American countries collected ethnoracial statistics on at least one national census if not more.³ The ethnoracial demographic statistics produced in these decades were generated and deployed in heterogeneous ways. However, there were some shared themes that are evident in official census reports from across the region.

Prominent among these shared themes, many official census reports from the 19^th and early 20^th centuries use demographic data to substantiate arguments that national populations were becoming gradually more homogenous, and gradually whiter. Many authors of census reports noted declines in Black and indigenous populations, whether through differential mortality and fertility, or through mestizaje. They pointed to these supposedly objective demographic trends as positive signs of national progress.⁴ In the context of published national census reports, statistics were used to support nationalist, nationalizing narratives that emphasized the gradual dissolution or disappearance of racially distinct indigenous or afro-descendent groups within the population. With variations on the theme across contexts, Latin American political elites and scientists pointed to official statistics published in official census results to craft and support stories of national progress conceived as ethnoracial demographic progress.

Official statistics were arranged in misleadingly clear comparative tables, and presented as irrefutable evidence of populations trending “naturally” in a whiter direction. Most of the time, such statistical tables were presented matter-of-factly and without much analysis, suggesting that the numbers were believed to speak for themselves. In some contexts, however, the accompanying narrative was decidedly triumphant. An infamous example of such racist triumphalism comes from the lengthy introduction to the Brazilian 1920 national census, penned by Oliveira Vianna. Discussing changes in Brazil’s ethnoracial composition from 1872 to 1890, and ignoring the myriad known problems with census data he pointed to as evidence, Vianna wrote: “The delicate and complex mechanism of ethnic selection has been explained in the previous paragraphs; however, the demonstration of the excellence of its effects is this statistical table.”

Figure 1
Brazilian statistical ethnoracial composition in the 1920 National Census

Vianna celebrated (his reading of) statistical trends, which he presented to readers as clear evidence of racial progress of the nation. At one point, he enthused: “See how fast the destruction of the black population is in the extreme south . . . In contrast to the descendent evolution of the two inferior types [we can see] the magnificent ascendant movement of the aryan type.”⁵

The introduction to Brazil’s 1920 census stands out from other national censuses in the region in this period in the celebratory tone of the reading of statistics that purported to show the demise of distinct black and indigenous groups. It also stands out in its eccentric and problematic reliance on already-at-the-time anachronist strands of “race science.” But the main argument in Vianna’s essay is one that is echoed, in somewhat more neutral language and with context-specific variation, across many national censuses from this period. A common theme running through national census reports was to emphasize the gradual blending and demographic disappearance of distinct ethnic and racial groups as the national population became “mixed” and, ideally, “whiter.”

Notably, up until the 1960s, these types of narratives were presented in official censuses results when data on indigenous and afro-descent populations was collected, and also, often, when it was not. Thus, the presence or absence of questions about race or ethnicity in the first century or so of national census-taking in Latin America is not in itself a direct or clear signal of when, whether, or how much the producers of official demographic statistics were focused on the racial demographic trajectories of their populations. Brazil is again exemplary. After including a race/color question in the national census of 1872 and 1890, the question was omitted in 1900 and 1920. Yet the official report on the 1920 census focused much more intently on demographic trends in the racial composition of the population than in any of the decades prior (^{Loveman, 2009}; ^{Camargo, 2010}; ^{Piza; Rosemberg, 1999}; ^{Nobles, 2000}).

As in Brazil, so in the rest of the region, national censuses were a strategic site, and official demographic statistics were an authoritative scientific language for constructing and disseminating narratives of national progress via “racial progress.” These narratives were anchored in and through the putative objectivity and authority of demographic science and stamped with the imprimatur of the state (^{Loveman, 2014}; ^{Desrosières, 1998}). Scientific-demographic national-racial fictions complemented stories of national progress through whitening being told through other mediums. Beyond the world of official demographic data production, in literature and art, for example, the idea of national progress as and through racial progress became a pervasive theme. Again, this broad-brush generalization overlays enormous variations in the specifics of these narratives across and within countries over time. Still, as we know from a large and rich historiography, narrative and artistic constructions of national mythologies that celebrated the dissolution or disappearance of ethnoracial or ethnocultural difference became a common genre across the region, beginning especially in the latter decades of the 19^th century and well into the 20th.⁶

The most well-known Brazilian version of this genre crystallized as a narrative of whitening through racial mixture, conceived as a pacific process through which the nation would ultimately develop into a more homogenous, and whiter, national type. The core demographic notion of whitening through mixture is captured in the famous and frequently cited painting (Figure 2), Modesto Brocos’ “The Redemption of Cam” (1895), which depicts racial mixture as an intergenerational process through which the Brazilian nation formed, and also as a process through which the population was and would be, continuously and miraculously, whitened.

Figure 2
Painting “A redenção de Cam” of Modesto Brocos, 1895

There were of course dissenting voices, scientific and artistic both.⁷ But the national narrative of Brazil as a racially-mixed and gradually whitening nation, curated from the mid-to-late 19^th century, became an anchor for subsequent variations on the theme. In the lead-up to the 1920 census, for example, a political cartoon that publicized the event depicted Brazil as a singular, white, (male) national type – as if it were the baby from Modesto y Brocco’s painting, all grown up:

Figure 3
Cartoon of 1920 in the Brazilian magazine A Careta

Well into the twentieth century, political, scientific, and artistic elites in Brazil and throughout much of Latin America advanced versions of national narratives that explicitly or implicitly equated national progress with the disappearance of distinct indigenous or afro-descendent populations within the nation. National ideologies that championed race mixture as an alchemical ingredient of “racial democracy” (Freyre) or “a nation for all” (Martí) or a raza cosmica (Vasconcelos) recognized and even valued some aspects of racial difference, but only in the context of assuming its eventual demographic disappearance and cultural dilution – processes conceived almost always with a directionality toward whiteness.

From the 1930s, and especially in the wake of World War II, the political, scientific, and moral legitimacy of the race concept became internationally suspect. This shifting international normative and scientific context unsettled and undercut the explicit equation of national progress with racial progress – and the definition of racial progress as demographic movement toward whiteness. In this altered context, Latin American census officials sought ways to reconcile their commitments to the continued pursuit of national progress with the global delegitimization of “race” as a scientific concept and of “racial improvement” as a political project (^{Loveman, 2014}, ch. 6). In many countries, this reconciliation involved the omission of race questions from national censuses. Countries that continued to enumerate ethnic “difference” in the context of the national census shifted increasingly toward questions about cultural traits and behaviors; these questions allow for the continued statistical visibility of indigenous populations in several countries, while contributing to the statistical invisibilization of afro-descendents throughout most of the region. Brazil and Cuba became outliers in the region for the continuity of questions about “race” or “color” on the national census – with the exception of Brazil in 1970s, discussed below. Throughout almost all of Latin America, the second half of the twentieth century became an era of official statistical color-blindness.

By 2000, Brazil and Cuba were outliers no more. Almost every Latin American country had introduced, or re-introduced, a census question to make afro-descendent populations visible statistically. And the same was true for the statistical visibility of indigenous populations throughout the region. Viewed against Latin America’s long history of using official demographic data to document the blurring of ethnoracial distinctions and to mark, and even celebrate, the disappearance of differentiated afrodescendent or indigenous populations within the nation, the recent regional embrace of ethnoracial data collection on national censuses appears as a tectonic cultural and political shift. Instead of insisting on the blending and disappearance of ethnoracial distinctions in their populations, Latin American state actors now officially recognize enduring ethnoracial distinctions in their populations and increasingly proclaim the pluri-ethnic composition of their nations.⁸

This regional shift in Latin America’s censuses is the product of political struggle. At the same time, the transformed datascape is itself politically productive: producing new subjectivities, new organizational sites, and new stakes of politics. In the two next sections, I elaborate on this argument. First, I briefly describe how the transformation of Latin America’s ethnoracial statistical datascape resulted from political struggle, focusing on the articulation between actors embedded in social movement, the social sciences, and government agencies at local, national, regional, and international levels. I then turn to an analysis of some of the things that the production of ethnoracial statistics has become productive of, underlining the ways new data are transforming the political terrain that led to their production, and not always in ways that those who fought for their production intend.

How politics transformed the datascape

Existing accounts of the political struggles that led to the regional shift in census-taking practices in Latin America point to the critical role of mobilization by afro-descendent and indigenous activists, strategic collaboration with allied academics, and support and pressure from interested international organizations.⁹ In Brazil, where the continuity of racial classification on national censuses was disrupted by the removal of the question by the military government in 1970, targeted political advocacy by a small number of black movement activists, social scientists, and technocrats succeeded in getting the question reintroduced in the census of 1980.[¹⁰] Outside Brazil, advocates of census reform had to persuade national statistics agencies to introduce race or ethnicity questions for the first time in decades, or in some countries for the first time in their history. With Brazil’s experience as a crucial example, the addition of questions to national censuses crystallized as one concrete objective of activists who mobilized for ethnoracial recognition, rights, and redress within the broader political struggles for democratization and human rights in Latin America from the 1980s and into the early 2000s.

From the perspective of activists working to build the constituencies for mass movements domestically, and to increase the visibility of their struggles internationally, it made good sense to focus on the national census as strategic political target. Activists correctly identified the national census as a high-profile political stake to advance both symbolic and material goals. New questions on national censuses would, in themselves, repudiate national narratives that denied enduring ethnoracial difference or discrimination – narratives that historical census-taking practices had themselves helped to naturalize. To be named and counted on the national census was to win official recognition of the social existence of a collective identity or community. Official recognition, in turn, fortified the social reality of named groups. The census became a prime target of activists because official classification was rightly seen as key to undermine the symbolic violence of the past and open up new avenues for effective claims-making in the future.

Activists also targeted the national census because of the practical and political value of data that could speak to the material conditions of minoritized populations. Census data are not just any data; they are politically authoritative data, precisely because their production is supposed to be above and outside politics. Their objectivity is legitimated through their production by institutionalized expertise that combines the (supposedly neutral) authority of science and the state. Vested with this dual legitimacy, activists correctly recognized that ethnodemographic statistics could be deployed as ideal “tools of the weak” (^{Porter, 1995}); the same tool used in the construction and naturalization of the unjust social order could be deployed to expose the existence of injustices and to “furnish arguments” to contest them (^{Desrosières, 2014}, p. 351). In most Latin American countries, national censuses remain the primary source of statistical information about the population. Even in countries that conduct regular household surveys, the national census provides authoritative population data that serves as a baseline comparison for other surveys. Activists recognized that the inclusion of new questions on racial or ethnic identity or group membership would generate official, scientifically authoritative data to document the existence of and inequalities between distinct ethnoracial groups within the nation.

Brazil provided a model that activists elsewhere in the region sought to emulate and adapt to their own national contexts. The Brazilian experience suggested a more or less linear political trajectory: from the inclusion of a question on the census, to the production of statistical studies documenting inequalities between groups, to the development and implementation of policies to target those inequalities.¹¹ In reality, of course, the Brazilian story was much more complicated and conjunctural, and riddled with particularities and tensions that made it far from a simple template for activists elsewhere to follow.

Brazilian activists’ starting point for this political trajectory was quite distinct from the position of activists elsewhere in the region. As noted above, Brazilian national censuses had long included a race question. Its removal in 1970 was a deviation, so advocates were arguing for a return to the status quo ante rather than for a radical innovation, as was the case for several other countries. Also, unlike the situation in many other countries, in Brazil activists drew upon and built from a long history of black activism (^{Mitchell, 1992}; ^{Andrews, 1991}; ^{Hanchard 1994}). While the black movement in the 1970s was relatively small and geographically concentrated, and while one of its main challenges was to “raise consciousness” of blackness among broader sectors of the afro-descendent Brazilian population, the social fact of blackness in Brazil was broadly recognized (no one would question that someone who identifies as Brazilian could also identify as, or be, Black). This state of things contrasted to the situation faced by black activists in countries such as Argentina, Chile, Mexico, or Peru, for example). Most consequently, by the 1970s there was already an established social scientific and state interest in investigating questions of racial inequality in Brazil, an intellectual and political environment very distinct from the terrain in most of the rest of the region in these years.

The existence of prominent social scientists invested in the reintroduction of a race question in the 1980 census made possible an articulation of interests that proved politically effective.¹² In the 1970s, sociologists ^{Carlos Hasenbalg (1979)} and ^{Nelson do Valle Silva (1978)} published landmark studies that showed racial inequality in social mobility in Brazil. Their analyses were based on nationally representative statistics collected in the 1976 PNAD and used advanced statistical methods to document that Brazilians who self-identified as “white” experienced more intergenerational social mobility than “não branco” Brazilians. The discovery of “statistically significant” racial inequality in Brazil was not new. In the 1940s and 1950s, researchers sponsored by UNESCO analyzed Brazilian racial dynamics as part of the organization’s ill-fated post-WWII search for examples of “harmonious race relations” to serve as model for the world (^{Maio, 1999}; Azevedo, 1955; ^{Fernandes, 1972}; ^{Ianni, 1987}). These studies documented severe racial inequality, but they interpreted it as a legacy of slavery that would be resolved with (and by) Brazil’s modernization. The stratification models used by Hasenbalg and Silva disproved this interpretation by “controlling for” social origins to isolate contemporary societal dynamics and “reveal” the pervasive, present-day existence of racial discrimination.

These studies’ core findings aligned with the diagnostic and political vision of black movement leaders (^{Campos, 2013}). In addition to providing scientific proof of contemporary racial prejudice in Brazil, they used a dichotomous categorical comparison to undertake and interpret the findings. Hasenbalg and Silva chose to “lump” pardos and pretos into one “não branco” category for purpose of their analysis. They did so in part because the number of “pretos” was small for their statistical methods to productively analyze as a separate group, and in part because their models worked best with dichotomous data.¹³ The vision of a Brazilian population divided into those who are white and those who are not corresponded to arguments from within the black movement at the time that all Brazilians of any afro-descendent ought to recognize themselves/each other as negro. Though the preferred label for the category differed, the demarcation of the racial boundary was perfectly aligned. As ^{Campos (2013)} argues, this fundamental epistemological alignment smoothed the way to political alignment. That alignment secured the reintroduction of a race question on the census and continued to prove important when it came time to put racial statistics to use in the realm of public policy. Needless to say, this pivotal (and temporally circumscribed) epistemological and political alignment between activists and social scientists was not easily replicated in other countries in the region (nor, as we will see later, did it remain stable over time within Brazil).

Notwithstanding their very different situations, activists in other countries drew example from the Brazilian experience as they formulated political projects to combat racial inequality in the 1990s and 2000s. Inspired in part by the success of Afro-Brazilian activists, beginning in the early 1990s, Afro-Colombian and indigenous organizations lobbied the national statistics agency (DANE) to add new queries to Colombia’s national census (^{Buvinic; Mazza, 2004}). Activists called for new census queries on ethical grounds. They argued that the statistical invisibility of Colombia’s indigenous and afro-descendent populations in the national census perpetuated cultural violence against these segments of the Colombian population. The addition of new census questions was required, activists explained, to ensure that indigenous and afro-descendent populations garnered official recognition of their existence, protection of their rights, and redress for their historical marginalization (^{Paschel, 2010}).

Activists’ demands for census reform were bolstered by the shifting stance of the Colombian state on the issue of recognition of domestic minority populations. Colombia became a signatory to ILO Convention 169 in 1991 and new data were required for compliance. Signatories to ILO Convention 169 are expected to track and report numerous indicators of indigenous peoples’ well-being in comparison to the general population. States require data on socio-economic conditions, education, health, land, working conditions and employment, impacts of large development projects, infrastructure, and human rights violations (^{Rodríguez-Piñero, 2005}). The Colombian state also needed information about the size of indigenous reserves (resguardos) to carry out planned decentralization. Constitutional reforms granting special rights to Afro-Colombian communities, in turn, made it desirable to collect new information about the size and situation of these communities. Thus, momentum towards democratization domestically, growing pressure for official recognition of diversity internationally, and the specific targeted demands of indigenous and afro-descendent organizations converged to usher in a historic change to Colombia’s national census.

Building from their success, Afro-Colombian activists and their allies within DANE set their sights on empowering their counterparts in other Latin American countries to introduce analogous reforms. Bottom-up activism targeting national statistics agencies, pioneered in Brazil and then Colombia, put the issue of statistical visibility of minority populations on political agendas in a growing number of Latin American countries. By the early 2000s, pressure to adopt new census queries had disseminated across Latin America through the deliberate and coordinated efforts of domestic activists, their census agency allies, and regional and international organizations. In most cases, the appeals of domestic activists did not suffice to convince census agencies to introduce new questions on national censuses. Beyond Brazil and Colombia, it was the convergence of such appeals with new initiatives and demands from regional and international organizations that ushered in census reforms across much of the region.¹⁴

Activists in Brazil and Colombia took the lead in making census questions and categories a pivotal stake in broader political struggles for recognition, rights, and redress. In countries where national political elites resisted activists’ calls to introduce racial or ethnic data collection, international activist networks and international organizations played critical roles in pressuring national statistics agencies to introduce reforms. The pressure on national statistics agencies to add new ethnic or racial questions to censuses took varied forms. In some countries, pressure took the form of indirect but institutionally supported encouragement to voluntarily adopt new questions. International conferences and workshops brought together leaders of community groups, experts in the development of comparative economic and social indicators from regional organizations such as CEPAL, and academics with international comparative expertise to advise national statistics agencies on how to design and implement the new questions. In other contexts, pressure was more direct and coercive, such as understood conditionalities attached to loans from multilateral lending institutions for funding ongoing census operations.¹⁵

In sum: national statistics agencies became increasingly invested in ethnoracial data collection in response to targeted political pressure from coalitions of grass -roots activists, social scientists, and international development organizations, with the combinatory weight of these actors, and the means and modes of their alliances, varying substantially across different national contexts in the region. The particular political coalitions that managed to add new race and ethnicity questions to Latin American censuses differed in different Latin American countries, reflecting distinct histories of black and indigenous activism, relationships of activists to academics, technocrats, and political parties in power at the national level, and the relative status of national governments in the regional and international system of states.

Yet by the 2010s, with very few exceptions, the outcome of these political fights across almost the entire region was in one key sense the same: blacks and indigenous peoples were enumerated as such in national censuses. For the majority of Latin American countries, the statistical visibility of race, color, and ethnic identity in the national census departs from decades of de facto and de jure insistence on the absence or inconsequence of ethnic or racial distinctions within national populations. In the first decades of the 21^st century, grass-roots political claims and their successful articulation to social science research agendas and to national and international political projects brought a prolonged era of official color-blindness in Latin America to an end.

How the Datascape Transformed Politics

The fact that almost every country in Latin America now produces ethnoracial population data from the national census represents a momentous political accomplishment. In several countries, the inclusion of new questions and categories on national censuses has made indigenous and afro-descendent individuals statistically “visible” for the first time in decades, or in some contexts, for the first time ever. This statistical visibility, in itself, marks a victory for communities that have long struggled to gain official recognition. It also produces the conditions for rewriting Latin American narratives of nationhood, resetting discursive parameters for subsequent political struggles. On this altered political terrain, the processes of producing and using official ethnoracial statistics create new fronts and facets of political contestation within ongoing efforts to address contemporary inequalities and injustice. Ethnoracial statistics are not merely tools for engaging in politics, they contribute to constituting new sites, subjects, and stakes of political struggle.

Producing political sites

One striking feature of the transformed political terrain for the production of official ethnoracial statistics in some parts of Latin America is that it has opened novel sites for political participation. In a break from historical precedent, in some countries in the region, the enumerated have won a seat at the table to discuss the questions and categories that the census uses for their enumeration. In some contexts, the enumerated have also been invited to acquire training needed to themselves analyze the data that is produced.

For example, as part of the regional “Todos Contamos” workshops, government statistics offices were encouraged to facilitate the participation of representatives from indigenous and afro-descendent organizations in the process of ethnoracial data production. The UN, the World Bank, CELADE, and other agencies supported the creation of training programs to provide indigenous and afro-descendent individuals with access to the methods and software used by statisticians, policy analysts and government officials to analyze census data. As one illustration: a technical cooperation agreement between the Ministry of Health of Chile (MINSAL) and CELADE to produce a socio-demographic report on the metropolitan indigenous population using census data included “training workshops for indigenous technical personnel on the REDATAM software and the use of censuses” (^{ECLAC, 2009}). Financial and logistical support for such workshops demonstrated that calls by development agencies for greater inclusivity in the process of data production were not merely rhetorical.

Some agencies called for projects to go beyond the provision of technical training for data analysis, to incorporate input from community groups on the design of questionnaires. The UN’s Permanent Forum on Indigenous Issues, for example, flagged the need for more information on whether survey questions adequately addressed the needs of indigenous communities: “[…] indigenous peoples’ understanding of poverty, or land rights, often differs considerably from that of dominant or mainstream populations. This is rarely taken into account in the collection of relevant data” (^{UN, 2004}, p. 9). In the lead-up to the 2010 census round, the UN sponsored conferences and seminars to ensure participation by representatives of indigenous and afro-descendent groups. This was a primary objective of the 2008 CELADE/ECLAC conference held in Santiago, Chile, with funding from UNICEF, UNFPA, UNIFEM, and the WHO. The workshop brought together “more than 100 experts from more than 20 countries, including from governmental and non-governmental organizations, representatives of indigenous and afro-descendent organizations, academics and technical experts from international agencies.”

Significantly, the meeting emphasized the importance of generating statistics that would be recognized as valid not only by governments and development agencies, but also by indigenous and afro-descendent communities. Towards that end, a conference report recommended, “[. . .] we should aim to obtain this information with the participation of the peoples (pueblos) and communities, which is what will make it appear legitimate in the eyes of the entire population.” The conference participants underscored that participation should be construed in the broadest possible terms. The final recommendation of the conference report was that “participation of [indigenous and afro-descendent] pueblos in the design of questions and the collection and analysis of data that refer to them should be institutionalized.” The CELADE conference report denotes a critical shift in how international organizations publicly construe the relationship between the producers of demographic data, the individuals from whom data are collected, and the uses to which the data are put.

In practice, efforts to ensure that voices of those who will be enumerated actually get taken into account in designing instruments for enumeration remain the exception rather than the rule. Over time, it is conceivable that the continued legitimacy of the national census in the eyes of citizens will come to depend on the existence of formal venues for public input into census operations. In theory, such forums could become arenas of substantive democratization in action – where the historical objects of statistical inquiry become subjects empowered with the authority, expertise, and resources to investigate themselves. A much more likely scenario, however, is that national statistics officials, in collaboration with experts working for international development agencies, maintain de facto control over design and execution of national surveys, while encouraging popular participation on carefully delimited issues with a range of acceptable outcomes more or less predetermined.

The inclusion of the enumerated in the design of tools of enumeration has opened up new sites of political struggle that used to be entirely internal to the administrative-technocratic census bureaucracies. These are discussions where demands for recognition – which lines of ethnoracial distinction will get official sanction and which will remain statistically invisible – are tied to fights over representation – who gets to speak on behalf of whom? While these facets of official statistical production have always been political, this reality has been obscured historically through control of the means of enumeration in the hands of an elite few. Indeed, historically, the symbolic authority of the census as a source of objective demographic knowledge hinged on the successful obfuscation of the politics of production of that knowledge. Recent developments in the region have brought politics that inform every stage of data production into open view.

To the extent that the inherently political nature of producing official ethnoracial statistics becomes widely recognized, the authority of the census as a source of objective information risks being eroded. This erosion of legitimacy, in turn, may undermine what remains of the states’ capacities to pass off official census categories as mere description of demographic realities. The democratization of the process of data production, however limited and circumscribed, is likely to invite growing critiques of national census operations per se.

Producing political subjectivities

The politics that produced the transformed datascape of ethnoracial population data have also been productive of shifts in peoples’ subjectivities. To be clear, this is not an argument that census categories create new identities from scratch or out of thin air. The constitutive power of census categories on identities is not nearly as direct or automatic as their most ardent critics pretend. To the contrary, census categories are derived from and interact with pre-existing subjectivities and understandings of ethnoracial distinction in society (^{Emigh et al., 2016}). Their specific effects on self-understandings is contingent on their interactions with many other factors at play. Their influence hinges, especially, on the particular ways they are used – in media, in research, in administrative governance, in public policy etc. Thus, the influence of official census categories on individuals’ identities is often indirect and diffuse, rather than direct and instrumental. And their productive effects on collective self-understandings is not always neatly aligned what their producers may intend.

One prominent way that the introduction of new race and ethnicity questions on Latin American censuses became productive of shifts in ethnoracial subjectivities was through organized media campaigns targeted to this goal. In several countries, activists collaborated with census agency staff and international agencies to wage media campaigns that aimed to persuade would-be Indians and blacks to identify as such on national censuses. Testifying to the earlier successes of states’ ideological and political assimilationist projects, many Latin Americans who, by the varied criteria used by activists, social scientists, and others, “qualify” as indigenous or afro-descendent, nonetheless do not choose to identify as such when given a choice. For example, while many observers suggest that approximately one quarter of Colombian’s population is of African-descent, in the 2005 census, “only 11% of the population self-identified as Afro-Colombian.” Thus, in preparation for the census enumeration, organized campaigns sought to persuade those who would be enumerated to embrace new classifications.

Activists in several countries worked to strengthen domestic demand for government action on behalf of ethnoracially defined populations by increasing the number of their fellow citizens who identify as indigenous or afro -descendent. In one of the first such campaigns, in the lead-up to the 1990 census in Brazil, black movement activists launched a publicity campaign with support from the Ford Foundation to persuade Brazilians with any African ancestry to mark preto (“black”) instead of pardo or branco on the census (^{Nobles, 2000}). The campaign slogan admonished: Não deixe sua cor passar em branco. Prior to the 2005 census in Colombia, to take another example, Afro-Colombian organizations launched a campaign to encourage Colombians to recognize the “beautiful faces of my black people” (Las caras lindas de mi gente negra) (^{Estupiñón, 2006}; ^{Paschel, 2013}). The organizations produced a television commercial featuring individuals who identified themselves as morena, negra, mulata, zamba, or raizal, and concluding with the slogan: “In this census, make yourself counted. Proudly afro-descendent.” The advertisement was aired through an agreement with the national statistics agency (DANE) and also disseminated via social media. In Panama prior to the 2010 census, a campaign with the same slogan did double duty as a commemoration of afro-descendent women on the International Day of Women. The 30-second spot featured afro-descendent men engaged in traditionally women’s work (ironing, cooking, hanging laundry) and appealing to viewers to commemorate International Women’s Day and to show gratitude to afro-descendent Panamian women in particular by proudly reporting their afro-descent on the census.

Campaigns to educate Latin American populations about the importance of responding to questions about ethnic or racial heritage, and how to respond, escalated significantly in preparation for the 2010 census round. Domestic activists, census officials, national media outlets, and regional and international organizations worked together to launch major publicity campaigns to inform Latin Americans of African descent about the new census queries and to encourage them to acknowledge their African heritage in selecting a response. For example, a regional Census 2010 working group comprised of afro-descendent leaders partnered with the UN Development Fund for Women as well as a Brazilian communications firm to produce a four-part television series called “The Americas have color: Afrodescendents in 21st-century censuses.” According to a press release, the series was “created to inform the population of the Americas about the 2010-12 census round” and covered “the conditions of life of black men and women, black resistance throughout history and a panorama of public policies to confront racism.” The series described the living conditions of afro-descendent populations in Brazil, Ecuador, Panama, and Uruguay and underscored the importance of self-identification as afro-descendent in the census.

Publicity for new ethnic and racial ancestry questions often made explicit mention of racial prejudice, discrimination, and inequality as social ills that the new statistics would help to combat. In Brazil, for example, the “Americas have color” documentary series opened with a special episode of a weekly news program, Cenas do Brazil, in which journalist Lúcia Abreu discussed “the importance of declaring one's color on the 2010 Census, the evolution of demographic data that refer to color or race, and their contribution to the design of public policies.” Public campaigns encouraging Latin Americans to acknowledge and embrace the African part of their heritage, coupled with increasing mainstream media coverage of racial discrimination and prejudice, represents a significant break from decades of public silence around this issue.

While the net effect of public information campaigns on census results is difficult to isolate from other sources of influence on collective and individual subjectivities, Brazil and Colombia saw growth in the relative size of self-identified afro-descendent populations over the past two decades. Demographers and other social scientists have estimated how change in racial self-identification between censuses is shaping these shifts.¹⁶ Some observers attribute these trends to census outreach campaigns in the context of broader cultural shifts in the valuation of blackness, especially among younger cohorts; others note the likely influence of the growing use of official census categories in ethnoracially-targeted public policies. As discussed further below, when ethnoracial census categories are used to design, implement, or monitor affirmative action programs, instrumental incentives may play a role in shifting self-identifications as well (^{Muniz, 2010}; ^{Canessa, 2007}; ^{Bailey; Fialho; Loveman, 2018}).

When states collect and use ethnic and racial data, this shapes subjectivities partly through triggering public discussions and renegotiations of what particular racial categories signify, who belongs within them, and the criteria for belonging (^{Mora, 2014}). To the present in Brazil, there is active debate among and between activists, scholars, technocrats, politicians, and increasingly many other public figures over the meaning of racial categories that itself contributes indirectly to the constitutive effects of those categories. University undergraduates interviewed in Rio de Janeiro in the early 2000s were not always sure of the “correct” way to answer a question about their raça or cor. They often referenced their appearance and their family background in discussing whether or not to identify themselves as pardo or preto (^{Schwartzman, 2009}). In the years since then, there is indication that responding based on understandings of origins or ancestry have gained ground (^{Guimarães, 2011}). Additionally, there is evidence that between 1995 and 2008 (before the implementation of quota policies, and then some years after implementation) Brazilians became increasingly likely to choose one of the census race/cor categories to self-identify in an open-format survey question (^{Bailey et al., 2018}). This suggests the recursivity of questions and categories used in the census and public policy in shaping the social phenomenon they aim to describe.

Importantly, national media campaigns and public debates about ethnoracial census categories and their uses influence the subjective self-understandings of those who are not the principle targets of the campaigns or public policies, as well as those who are. In recent years in several Latin American countries, groups have emerged to defend official recognition of mestizos, using many of the same arguments invoked by afro-descendent and indigenous groups to demand visibility on censuses. For these critics, official ethnoracial classification is not inherently problematic; it is the depreciation or negation of “mixed” categories in favor of “absolute” or “pure” categories that is cause for concern. In Brazil, for example, in the mid-2000s, an NGO based in Amazonia with the name “Nação Mestiça” sought to advance “the valorization of the process of miscegenation (mixture) between the diverse ethnic groups that created the Brazilian nationality, the promotion and defense of pardo-mestiça identity and the recognition of pardo-mestiços as cultural and territorial inheritors of the people from which they are descended.” Arguing against campaigns to “unmix” Brazilians and echoing nationalists from the past, the group's slogan announced: “A miscigenação une a nação.”

Such slogans foreshadowed Jair Bolsonaro’s skillful deployment of the same nationalist trope to foment resentment against affirmative action policies as part of his presidential campaign, as seen in Figure 4.

Figure 4
Jair Bolsonaro wearing a T-shirt with the slogan “My color is Brazil!”

As is well-known, race-targeted social policies in Brazil produced fierce scholarly, public, and political debate.¹⁷ Above and beyond the immediate stakes and specific arguments of these debates, the existence and tenor of the debates themselves have arguably had productive effects on Brazilians’ subjectivities. Of particular importance, the public controversies helped catalyze the salience and broad popular resonance of a resurgent nationalist identity that insists on the (re)subordination of distinct ethnoracial identities to identification as Brazilian.

Thus, the production and use of official ethnoracial statistics may be productive of subjectivities that shift toward conformity with official categories and simultaneously productive of subjectivities of reaction, that take shape as alternatives to official categories or in opposition to official categorization per se. The recursivity of census categories – their ability to produce a “looping effect” in Hacking’s phrasing -- is actively contested and thus politically contingent. The productive effects of new census questions on subjectivities may not resound within closed loops. They may instead reverberate, as through an open-ended spiral. In some ways, the growing use of official ethnoracial categories in Latin America is shaping subjectivities in patterned and (more or less) predictable ways. In other ways, the constitutive consequences of official ethnoracial classification are spilling out in unanticipated forms and directions.

Producing political stakes

The way individuals answer census questions about their racial or ethnic identification reflects, in part, their assessments of what the census is really asking based on prevailing societal understandings of the terms used, why it is being asked, and what they are expected to take into account in providing their response. Their answers reflect, as well, shifting understandings of what exactly is at stake.

By the latter part of the 2000s, political struggles to get Latin American states to collect ethnoracial statistics on their national censuses mostly gave way to increasing demands to do things with the new data. As the data were analyzed and “put to use” to motivate, develop, implement, and monitor public policies, new political stakes emerged -- stakes that were directly tied to official ethnoracial classification. The existence of new stakes tied to ethnoracial categories reverberated through the political terrain, shifting identities, alliances, and alignments (epistemological and political) that had fueled the success of political struggles to transform the datascape in the first place.

Before the new data collected in Latin American censuses could become “tools of the weak” in ^{Porter’s (1986)} sense, the avalanche of raw numbers needed to be transformed into useful/usable knowledge. As the authors of a report sponsored by the Interamerican Development Bank pointed out, “The inclusion of the ethnic variable in censuses and surveys is pointless if it isn’t used in analyses” (^{IDB, 2000}). Thus ensued the production of a wave of social scientific studies and government reports. These assembled descriptive statistics and statistical analyses to document inequalities between ethnoracial categories on various indicators of material well-being, including studies of disparities in life expectancy, educational attainment, access to electricity and sanitation, income, among other statistical “outcomes.”¹⁸ International and regional development agencies also began to generate comprehensive reports on the condition of afro-descendent or indigenous communities in particular Latin American countries and in comparative perspective. The Economic Commission for Latin America and the Caribbean, for example, committed resources to the production of statistical reports on the status of indigenous and afro -descendent youth in Latin America based on results from the 2000 censuses. The World Bank and agencies of the United Nations also sponsored the production of reports on various aspects of health, education, and living conditions of minority populations in several parts of Latin America.

Armed with a growing body of statistical evidence of ethnoracial inequality, activists and their allies within state governments began to push for targeted social policies. Statistical documentation of ethnoracial inequality supported calls for states to supplement legal promises of protection from discrimination with proactive, corrective, and reparative measures to reduce existing disparities of condition and opportunity. In contexts where social movement actors or their allies occupied positions within the state, policies that directly address ethnoracial inequality became a visible priority.

Brazil pioneered such policies in the region, and to date, Brazil has gone farthest with their use, with a broad range of affirmative action initiatives in government agencies, university admissions, and industries such as fashion and television.¹⁹ In Brazil, affirmative action policies came to rely upon official ethnoracial statistics to justify their introduction, guide their implementation, and to monitor the results. The implementation of affirmative action programs created new stakes of racial classification, and thus drew sharp attention to the categories used in the census, in social science, and in social policy.

The production of new political stakes introduced fissures into the political terrain that stressed, and in some contexts fractured, alliances that had been strong in the political struggle to get affirmative action legislation passed in the first place. In different ways, these fissures ran through the social sciences, social movements, universities as institutions, and the political domain (^{Bailey; Peria, 2010}; ^{Carvalho, 2005}). As just one example of such a fissure, it is instructive to consider how the politics and practice of drawing boundaries around and within the não branco population shifted from the period of mobilization for affirmative action policies, to the period of their implementation. As discussed above, social science research in the 1970s and 1980s favored the use of a dichotomous analytic comparison between brancos and não brancos, with the latter grouping together census pardos and pretos. This scheme was easily translated into comparisons between whites and negros, using black movement activists’ preferred framing at that time, which advocated for an expansive definition of negro that included census pretos, pardos, and anyone who identified as afrodescendent (^{Campos, 2013}). In scholarship on racial inequality in Brazil written in English this translation was even more straightforward, since the same word – black – is used to translate both preto and negro.

In the period of political mobilization leading up to the passage of racial quota policies, methods that downplayed distinctions within the non-white population were favored by social scientists, just as they were important to activists’ vision building the broadest possible constituency to press for social policies to combat racial inequality. Fast forward some years, and there is a much more active debate among social scientists over whether it is appropriate or misleading to group pretos and pardos together in statistical analyses.²⁰ Citing evidence of socioeconomic heterogeneity within the pardo category, variation in perceptions of discrimination by class status, and regional variation in its meaning, some researchers advocate avoiding binary models because they obscure class and regional inequality within the pardo category, and the severity of disadvantage suffered by pretos.²¹ Alongside these emerging strands of research, there are political realignments within the coalitions of activists and academics supportive of affirmative action policies. Programs that tied opportunities for university admission or government employment to categorical ethnoracial identification raised thorny questions about who qualifies for such programs and who does not. This question opened the door to debates over who decides who qualifies, and on what basis? The distributional stakes of affirmative action programs produced reassessment and revision to earlier views of how racial boundaries in Brazil should be drawn.

Whereas in a mobilizational moment of politics, the favored approach of activists and allied scholars alike was to group pardos and pretos together, in a distributional moment of politics, the favored approach has shifted to one that differentiates pardos from pretos, “brightening” a boundary they had previously blurred. This is especially evident in some of the discourse and practices of “anti-racial fraud” campaigns.

Figure 5
Poster “Who are the new beneficiaries of UFES’ black quotas?”

Figure 5 is a poster taken from a website of a black student group at a Brazilian university. The poster points out that 51% of the Brazilian population is black, and that the other half has double the opportunities. To arrive at 51%, they must include census pretos and census pardos, since self-identified census pretos alone would be less 7.6% (according to the 2010 census). Nonetheless the picture is clearly one that represents brazil as a population made up of black and White. At the bottom it notes that false self-identification is a crime. Most media coverage of this campaign and legislation focused on people accused of falsely claiming to be non-white when they are “actually” (socially) white. However, racial fraud charges were also levied against self-identified pardos seen as not being black enough. An article that appeared in English language press in 2017, in the magazine Foreign Policy, told the story of a woman who stood accused of racial fraud for identifying herself as parda in the application process for a public prosecutor job. Some activists criticize people falsely self-identified as pardas for using programs that were in fact set up for pardos and pretos, arguing that really these spaces were not meant for “people with black grandmothers”; rather they were meant for those with “very dark skin” (^{Oliveira, 2017}). Criteria used to draw distinctions within the negro category – distinctions that were downplayed or even denied in a mobilizational moment of politics were being resurrected and redrawn to more narrowly delimit beneficiaries in the context of determining distribution of scare resources. An unintended consequence of Brazil’s racial quota policies and the varied reactions to them has been to (re)introduce political and sociological fissures through the não branco population, sharpening the boundary between pretos and pardos that was previously deliberately blurred.

Ethnoracially-targeted affirmative action programs of various sorts are in place or on political agendas in several other Latin American countries. These initiatives range from policies focused on health services, nutrition, housing, poverty alleviation, and land titling, to educational benefits and guaranteed political representation. As ethnoracially-targeted social programs spread in Latin America, controversies over why and how states classify their populations by race or ethnicity escalate. As occurred in Brazil, critics of these programs point to opportunistic self-identification; they note how official categories will create or grow constituencies along those categorical lines; and they warn of unintended consequences of well-meaning policies, including the consolidation of more rigid group boundaries and attendant polarization or fragmentation of national societies.

Those who support the aims of targeted social policies are likely to raise concerns about official ethnoracial classification as well. Policymakers, social scientists, and activists will debate whether the categories used for implementation effectively funnel resources or opportunities to the most deserving and intended beneficiaries. The ethnic or racial categories used to implement targeted social policies will never correspond perfectly to all of the categorical distinctions that are operative in the lives of individuals and communities. In some contexts, official ethnoracial categories may inadvertently exclude individuals who merit inclusion based on policymakers’ intent. In others, official categories will be deemed too broad, diffusing the impact of initiatives to address explicitly ethnoracial facets of poverty and inequality. There is an inherent tension between census question formats or social science analytic methods that aim to maximize the total number of those who count as indigenous or black, and the design of social policies that use such categories to target those who are most disadvantaged.

The transformed datascape for the collection, production, and use of official ethnoracial statistics in Latin America can be used both as information and as ammunition in political struggles over targeted social policies and over competing visions of social progress. In previous decades, the disaggregation of poverty and health statistics by sex revealed the gendered nature of inequality in Latin America. Such statistics helped support a shift toward development strategies focused on women’s roles in the distribution of scarce resources within families and communities. The emergent avalanche of racial and ethnic statistics will likewise continue to expose severe ethnic and racial inequalities in some regions, and provide support for government interventions that aim to reduce them. In some contexts, these projects may respond to real and pressing needs of historically neglected populations. In others, they may amount to what anthropologist Charles Hale terms “multicultural neoliberalism,” serving primarily to appear to address inequalities while effectively dissipating bottom-up pressure for development strategies that would entail more substantive structural change. Either way, the production and use of official ethnoracial statistics produces new political stakes which are themselves productive: as they are put to use in government administration and public policy, they produce tremors, fissures, and fractures that may coalesce in movements that fundamentally alter the political terrain.

Conclusion

Looping effects or spirals? Navigating a datascape transformed

The inclusion of new race, ethnicity, and color questions on national censuses and nationally representative household surveys in Latin America in recent decades is a momentous political accomplishment. This transformed datascape for the collection, production, and use of ethnic and racial statistics, in turn, has become productive of new sites, subjectivities, and stakes of politics.

The new politics of ethnoracial data production have seeded new struggles within the political field of grass-roots activism, social scientists, state actors, and international organizations. This is especially evident in contexts where the inclusion of new questions or the revision of question wording or response options remain open for discussion.

Struggles over the terms of official recognition – which ethnoracial categories and boundaries will be officially sanctioned and which will remain officially invisible – cede easily into struggles over representation – who gets to speak on behalf of whom? These latter struggles have shaped the field of social movement organization and ties between NGOs, activists, political parties, and national and international actors (^{Paschel, 2016}).

The unprecedented availability of ethnic and racial population data has predictably unleashed an avalanche of new statistical descriptions and analyses of ethnoracial inequalities.

In most countries in the region, it remains an open question whether or how the accumulation of quantitative studies documenting pervasive ethnoracial disparities in indicators of income, health, and education will translate into successful political struggles for either incremental or more transformative change. The new availability of ethnoracial population data has bolstered activists’ demands for expanded benefits of social citizenship, including demands for ethnoracially targeted social benefits as a means of redress for historical marginalization and/or contemporary discrimination. As noted above, affirmative action policies for ethnoracially – defined groups are already in place in several Latin American countries and there is pressure on states from both domestic activists and international organizations to introduce more initiatives of this kind in the future. Such policies focus on targeted delivery of benefits ranging from health services to housing, poverty alleviation to political representation. Among the most visible and contentious initiatives have been those focused on affirmative action in the field of higher education.

In Brazil, the proliferation of statistical social scientific studies of racial disparities in the late 1980s and 1990s supported claims made by the black movement and allies for targeted policy interventions; Brazil became a leader in the introduction of affirmative action in government employment and university admissions. More recently, however, Brazil has also become a leading example of organized backlash. Brazil had been seen as a regional model for how political mobilization could produce changes to the census, which could then produce ethnoracial statistics, which could then be used to produce social science research on racialized inequalities, which could in turn inform public policies that aim to address those inequalities. Even as this example remains, the more recent Brazilian experience of reactionary backlash to this political project -- and its links between activists, social scientists, and state actors – is now weighed by its neighbors as a precautionary example as well.

As the Brazilian experience demonstrates all too well, social policies that explicitly aim to address ethnoracial inequalities through interventions that target ethnoracially-defined beneficiaries often draw attention to the political processes that inform the production of ethnoracial data in the first place. Political battles fueled in part by the statistical documentation of ethnoracial inequalities tend to circle back around to political battles over the production of ethnoracial statistics per se. Thus, as ethnoracially-targeted social programs spread, controversies over why and how states classify citizens by race or ethnicity will likely escalate. As in the United States, so in Brazil, Bolivia, and a growing list of countries in the Americas are likely to witness escalating opposition to the idea that states may legitimately produce ethnoracial statistics at all.

The future of ethnoracial demographic data production in Latin American remains uncertain. Indeed, it is possible that challenges to the legitimacy of ethnoracial data production may unsettle the legitimacy of government collection and analysis of population data per se. ^{Desrosières (2014, p. 352)} argues that “for a statistic to play its social role as a neutral reference, above the conflicts of social groups, it must be instituted, guaranteed by democratic procedures, themselves legitimate. It then contributes to making reality and not simply reflecting reality.” While the analysis in this article could lend support for the first part of this claim, it questions the second: it is evident ethnoracial demographic statistics may contribute to constituting the realities they aim to measure even when their legitimacy is contested. Indeed, this is so even when their legitimacy is contested at every stage of their production, institutionalization, analysis, and use. The productive power of a statistic and the democratic acceptance of its means of production and institutionalization are not so tightly coupled.

Whether ethnoracial statistics contribute to (re)making reality in the ways their advocates and producers intend is another matter altogether. The official ethnoracial categories, institutionalized in statistics and operationalized through public policy, constitutive of subjectivities but not in as controlled a way as we might imagine by invoking ^{Ian Hacking’s (1995)} “looping effect” metaphor. Rather than a closed loop, the productive reverberations of official ethnoracial statistical categories reverberate in an open-ended spiral, wound more or less tightly and with the end pointing inward or outward in relation to movements of political actors and shifts in the political terrain. As the metaphor of a “spiral effect” suggests, the analysis in this article, and growing evidence from the region, affirms the productive effects of ethnoracial statistics include predictable and upredictable; thus, the creation and use of these data are apt to be politically productive in ways at once productive and counterproductive to the aims of those who produce them.

The transformed datascape is not only fueling the creation of new knowledge about ethnoracial inequalities in Latin America; it is also helping to define new sites, subjectivities, and stakes of political struggle over recognition, rights, and redress for historically marginalized individuals and communities – while also stoking reactionary nationalist opposition. Thus, while the early 21^st -century boom in the production of ethnoracial statistics about Latin America populations is a historic political and social scientific accomplishment, it is simultaneously a politically contentious and tenuous accomplishment that could well be short-lived.

New data have generated new knowledge about ethnic, racial, and color inequalities in Latin American societies; this new knowledge, in turn, together with the process and politics of its production, has stoked new sites, subjectivities, and stakes of politics. These have contributed to the political constitution of ethnoracial identities and boundaries, to the advancement of claims for greater equality and recognition for ethnoracially-defined individuals and communities, and to stoking the rise of organized reactionary opposition. Thus, in the context of unfolding political struggles to challenge existing relations between states, scientists, and (non)citizens, as social scientists grapple with the avalanche of ethnoracial population data in Latin America with the aim to advance understanding ethnoracial inequalities, they must also keep the politics of the production of such data, and their productive effects on politics, within the frame of analysis.

• 1
  The use of the term “avalanche” in this context is in reference to ^{Ian Hacking’s (1982)} essay “Biopower and the avalanche of printed numbers”. While Hacking analyzes a very different context and moment in the history of demographic statistics, echoes of that earlier avalanche still reverberate in this one.
• 2
  Table 1 provides a summary overview of the change in statistical visibility of indigenous and afro-descendent populations, but it also obscures the considerable variation in the particularities of census question formats and response options across countries, which make visible distinctions within these categories in some countries. Table 1 also omits consideration of other lines of ethnoracial distinction enumerated in some national censuses in the region (for example, some censuses allow responses for “whites” or “Asian”, while some deliberately omit such response options). For more on these variations and their justifications and consequences, see ^{Loveman (2014, chapter 7)} and ^{De Popolo (2008)}.
• 3
  For an overview of ethnoracial classification on censuses in Latin America in this period, see Tables in ^{Loveman (2014, p. 233 and p. 241)}.
• 4
  These generalizations are developed at more length and with many examples in ^{Loveman (2014, chapters 4-6)}. There are also many excellent studies of these practices for specific countries and census years – too many to cite them all here, but see, for example, ^{Otero (2006)} and ^{Camargo (2010)}.
• 5
  Brazil. Directoria Geral de Estatistica, Recenseamento do Brasil, 344.
• 6
  The historiography on this theme is much too long to cite comprehensively here. An important recent contribution to this large body of work is found in ^{Vejo and Yankelevich (2017)}; also, ^{Sommer (1991)} and ^{Martinez-Echazábel (1998}; ¹⁹⁹⁶⁾.
• 7
  On dissenting voices, see for example the views of a military doctor who used data collected from soldiers to argue against Vianna’s thesis of evolution toward a singular, whitened, national Brazilian “type” (^{Loveman, 2009}). See also ^{Skidmore (1993)}, ^{Schwarcz (1993)} and ^{Borges (1993)}.
• 8
  An extended analysis of the recent shift in state practices of ethnic and racial classification in Latin America can be found in ^{Loveman (2014)}.
• 9
  See ^{Nobles (2000)}, ^{Htun (2004)}, ^{Hooker (2005}; ²⁰⁰⁹⁾, ^{Paschel (2010}; ²⁰¹⁶⁾, ^{De Popolo (2008)}, and ^{Loveman (2014)}.
• 10
  Cf. ^{Nobles (2000, p. 98-110; 116-119)}. Brazil’s census almost omitted the race question in 1940, but it was ultimately retained with the justification and expectation that it would allow Brazil to show the world its continued progress toward whitening (^{Camargo, 2010}, p. 255-266).
• 11
  Illustratively, Argentina’s statistics agency explained on its website that the question on afro-descent on the 2010 census would aid “…the elaboration of future studies and investigations … with an eye to the realization of targeted public policies.”
• 12
  ^{Nobles (2000, p. 98-110; 116-119)} argues that absent organized pressure from academics and activists, Brazil’s 1980 census would have been fielded without a color question.
• 13
  They also argued later that the grouping of pardos and pretos made substantive sense given their proximity in social status in comparison to “whites”. To this day, scholars debate whether it is best to use dichotomous, trichotomous, an analytic combination of both, or none of the above to analyze racial inequalities in Brazil.
• 14
  For more on international organizations’ role, see ^{Loveman (2014, chapter 7)}.
• 15
  One example of the former type is the international conference in Lima, Peru 2016, which brought together social movement spokespersons, representatives of the national statistics agency, government officials, social scientists from independent research institute and universities, UN representative, and “international academic experts” to discuss the adoption of new questions on the 2017 census. For an example of the latter, see account of Nicaragua’s 2005 census in ^{Loveman (2014, p. 290-93)}.
• 16
  For example, ^{Carvalho, Wood and Andrade (2004)}.
• 17
  On these debates see: ^{Guimarães (1999)}, ^{Feres Jr, Campos and Daflon (2011)}, ^{Fry et al. (2007)}, ^{Bailey and Peria (2010)}, ^{Heringer and Johnson (2015)} and ^{Teixeira (2003)}.
• 18
  For example, see studies produced by CEPAL and the World Bank.
• 19
  For detailed description and analysis of these programs and their politics, see for example: ^{Feres Jr. (2007}; ²⁰⁰⁸⁾; ^{Fry et al. (2007)} and ^{Petruccelli (2015)}.
• 20
  There were dissenting voices on the use of dichotomous analytic scheme earlier, and some scholars continued to produce analyses that kept pardos and pretos separate throughout these years. But the approach that was seen and presented as most in line with politics of racial justice was to adopt a dichotomous lens.
• 21
  See ^{Bailey, Loveman and Muniz (2013)}, ^{Loveman, Muniz and Bailey (2012)}; ^{Daflon, Carvalhães and Feres Jr. (2017)}, ^{Muniz (2010)} and ^{Bailey and Telles (2006)}.

Referências

• ¹
  ANDREWS, George. Blacks and whites in São Paulo, Brazil, 1888-1988 Madison: University of Wisconsin Press, 1991.
• ²
  BAILEY, Stanley R.; TELLES, Edward. Multiracial versus collective black categories: examining census classification debates in Brazil. Ethnicities, v. 6, n. 1, p. 74-101, 2006.
• ³
  BAILEY, Stanley R.; PERIA, Michelle. Racial quotas and the culture war in Brazilian academia. Sociology Compass, v. 4, n. 8, p. 592-604, 2010.
• ⁴
  BAILEY, Stanley R.; LOVEMAN, Mara; MUNIZ, Jeronimo. Measures of ‘race’ and the analysis of racial inequality in Brazil. Social Science Research, v. 42, n. 1, p. 106-119, 2013.
• ⁵
  BORGES, Dain. Puffy, ugly, slothful and inert’: degeneration in Brazilian social thought, 1880-1940. Journal of Latin American Studies, n. 25, p. 235-256, 1993.
• ⁶
  BRAZIL. Ministério da Agricultura Indústria e Commercio. Directoria Geral de Estatística. Recenseamento do Brazil realizado em 1 de setembro de 1920 Rio de Janeiro: Typografia da Estatistica, 1920.
• ⁷
  BUVINIC, Mayra; MAZZA, Jacqueline. Social inclusion and economic development in Latin America Nova York: Inter American Development Bank, 2004.
• ⁸
  CAMARGO, Alexandre de P. R. Classificações raciais e formação do campo estatístico no Brasil (1872-1940). In: SENRA, Nelson de C.; CAMARGO, Alexandre de P. R. (orgs). Estatísticas nas Américas: por uma agenda de estudos históricos comparados. (Estudos & Análises). Rio de Janeiro: IBGE, 2010. p. 229-264.
• ⁰⁹
  CAMPOS, Luiz A. O pardo como dilema político. Insight Inteligência, n. 63 [online], 2013.
• ¹⁰
  CANESSA, Andrew. Who is indigenous? Self-identification, indigeneity, and claims to justice in contemporary Bolivia. Urban Anthropology, v. 36, n. 3, p. 195-237, 2007.
• ¹¹
  CARVALHO, José A. M. de; WOOD, Charles; ANDRADE, Flávia. Estimating the stability of census-based racial/ethnic classifications: the case of Brazil. Population Studies, v. 58, n. 3, p. 331-343, 2004.
• ¹²
  CARVALHO, José Jorge de. Usos e abusos da antropologia em um contexto de tensão racial: o caso das cotas para negros na UnB. Horizontes Antropológicos, v. 11, n. 23, p. 237-246, 2005.
• ¹³
  CELADE - Centro Latinoamericano y Caribeño de Demografía. Pueblos Indígenas y afrodescendientes de América Latina y el Caribe: relevancia y pertinencia de la información sociodemográfica para políticas y programas. CEPAL, 27 de abril al 29 de abril de 2005.
• ¹⁴
  DAFLON, Verônica T.; CARVALHAES, Flávio; FERES JÚNIOR, João. Sentindo na pele: percepções de discriminação cotidiana de pretos e pardos no Brasil. Dados, v. 60, n. 2, p. 293-330, 2017.
• ¹⁵
  DEL POPOLO, Fabiana. Los pueblos indígenas y afrodescendientes en las fuentes de datos: experiencias en América Latina. Santiago de Chile: Comisión Económica para América Latina y el Caribe, CEPAL, 2008.
• ¹⁶
  DESROSIÈRES, Alain. Statistics and social critique. Partecipazione e conflitto, v. 7, n. 2, p. 348-359, 2014.
• ¹⁷
  DESROSIÈRES, Alain. The politics of large numbers: a history of statistical reasoning. Cambridge: Harvard University Press, 1998.
• ¹⁸
  ECLAC - Economic Commission for Latin America and the Caribbean. Progress report on the biennial programme of regional and international cooperation activities of the Statistical Conference of the Americas of the Economic Commission for Latin America and the Caribbean (ECLAC) 2007-2009 Santiago: ECLAC, 2009.
• ¹⁹
  EMIGH, Rebecca J.; RILEY, Dylan; AHMED, Patricia. Changes in censuses from imperialist to welfare states How societies and states count. Nova York: Palgrave Macmillan, 2016.
• ²⁰
  ESTUPIÑÓN, Juan P. Afrocolombianos y el Censo 2005. IB - Revista de la Información Básica, v. 1, n. 1, art. 7, 2006. http://www.dane.gov.co/revista_ib/html_r1/articulo7_r1.htm
  » http://www.dane.gov.co/revista_ib/html_r1/articulo7_r1.htm
• ²¹
  FERES JÚNIOR, João. Ação afirmativa: política pública e opinião. Sinais Sociais, v. 3, n. 8, p. 38-77, 2008.
• ²²
  FERES JÚNIOR, João. Comparando justificações das políticas de ação afirmativa: Estados Unidos e Brasil. Estudos Afro-Asiáticos, v. 29, n. 1-3, p. 63-84, 2007.
• ²³
  FERES JÚNIOR, João; CAMPOS, Luiz A.; DAFLON, Veronica T. Fora de quadro: a ação afirmativa nas páginas d’O Globo. Contemporânea - Revista de Sociologia da UFSCar, v. 1, n. 2, p. 61-83, 2011.
• ²⁴
  FERNANDES, Florestan. O negro no mundo dos brancos São Paulo: Difusão Europeia do Livro, 1972.
• ²⁵
  FRY, Peter; MAGGIE, Yvonne; MAIO, Marcos C.; MONTEIRO, Simone; SANTOS, Ricardo V. (Orgs.). Divisões perigosas: políticas raciais no Brasil contemporâneo. Rio de Janeiro: Editora Civilização Brasileira, 2007.
• ²⁶
  GUIMARÃES, Antônio S. Raça, cor, cor da pele e etnia. Revista Cadernos de Campo, n. 20, p. 1265-1271, 2011.
• ²⁷
  GUIMARÃES, Antônio Sergio. Racismo e anti-racismo no Brasil São Paulo: Editora 34, 1999.
• ²⁸
  HACKING, Ian. The looping effect of human kinds. In: SPERBER, Dan; PREMACK, David; PREMACK, Ann J. (eds.). Causal cognition: a multidisciplinary debate. Nova York: Clarendon, 1995.
• ²⁹
  HACKING, Ian. Biopower and the avalanche of printed numbers. Humanities in Society, n. 5, p. 279-295, 1982.
• ³⁰
  HALE, Charles. Does multiculturalism menace? Governance, cultural rights, and the politics of identity in Guatemala. Journal of Latin American Studies, v. 34, n. 3, p. 485-524, 2002.
• ³¹
  HANCHARD, Michael. Orpheus and power: the Movimento Negro of Rio de Janeiro and São Paulo, Brazil, 1945-1988. Princeton: Princeton University Press, 1994.
• ³²
  HASENBALG, Carlos. Discriminação e desigualdades raciais no Brasil contemporâneo Rio de Janeiro: Graal, 1979.
• ³³
  HASENBALG, Carlos; SILVA, Nelson do V. Notes on racial and political inequality in Brazil. In: HANCHARD, Michael (ed.). Racial politics in contemporary Brazil Durham: Duke University Press, 1999. p. 154-178.
• ³⁴
  HASENBALG, Carlos; SILVA, Nelson do V. Estrutura social, mobilidade e raça Rio de Janeiro: IUPERJ, 1988.
• ³⁵
  HERINGER, Rosana; JOHNSON, Ollie; JOHNSON III, Ollie A. (orgs.). Race, politics and education in Brazil: affirmative action in higher education. Londres: Palgrave Macmillan, 2015.
• ³⁶
  HOOKER, Juliet. Race and the politics of solidarity Oxford University Press, 2009.
• ³⁷
  HOOKER, Juliet. Indigenous Inclusion/black exclusion: race, ethnicity and multicultural citizenship in Latin America. Journal of Latin American Studies, v. 37, n. 2, p. 285-310, 2005.
• ³⁸
  HTUN, Mala. From ‘racial democracy’ to affirmative action: changing state policy on race in Brazil. Latin American Research Review, v. 39, n. 1, p. 60-89, 2004.
• ³⁹
  IANNI, Octavio. Raças e classes sociais no Brasil São Paulo: Brasiliense, 1987.
• ⁴⁰
  IDB - Interamerican Development Bank. BID, Banco Mundial y Colombia auspician seminario internacional sobre factores raciales y étnicos en censos de América Latina Press release, 7 nov., 2000.http://www.iadb.org/es/noticias/comunicados-de-prensa/2000-11-07/bid-banco-mundial-y-colombia-auspician-seminario-internacional-sobre-factores-raciales-y-etnicos-en-censos-de-america-latina,799.html
  » http://www.iadb.org/es/noticias/comunicados-de-prensa/2000-11-07/bid-banco-mundial-y-colombia-auspician-seminario-internacional-sobre-factores-raciales-y-etnicos-en-censos-de-america-latina,799.html
• ⁴¹
  LOVEMAN, Mara. National colors: racial classification and the state in Latin America. Oxford: Oxford University Press, 2014.
• ⁴²
  LOVEMAN, Mara. The race to progress: census-taking and nation-making in Brazil, 1870-1920. Hispanic American Historical Review, v. 89, n. 3, p. 435-470, 2009.
• ⁴³
  LOVEMAN, Mara; MUNIZ, Jeronimo; BAILEY, Stanley. Brazil in black and white? Race categories, the census, and the study of inequality. Ethnic and Racial Studies, v. 1, n. 18, p. 1466-1483, 2011.
• ⁴⁴
  MAIO, Marcos C. O Projeto Unesco e a agenda das ciências sociais no Brasil dos anos 40 e 50. Revista Brasileira de Ciências Sociais, v. 14, n. 41, p. 141-158, 1999.
• ⁴⁵
  MARTINEZ-ECHAZÁBAL, Lourdes. Mestizaje and the discourse of national/cultural identity in Latin American 1845-1959. Latin American Perspectives, v. 25, n. 100, p. 21-42, 1998.
• ⁴⁶
  MARTINEZ-ECHAZÁBAL, Lourdes. O culturalismo dos anos 30 no Brasil e na América Latina: deslocamento retórico ou mudança conceitual? In: MAIO, Marcos C.; SANTOS, Ricardo V. (orgs.). Raça, Ciência e Sociedade Rio de Janeiro: Ed. Fiocruz, 1996. p. 107-124.
• ⁴⁷
  MITCHELL, Michael. Racial identity and political vision in the black press of São Paulo, Brazil, 1930-1947. Contributions in Black Studies, v. 9, art. 3, 1992.
• ⁴⁸
  MORA, G. Cristina. Cross-field effects and ethnic classification: the institutionalization of Hispanic panethnicity, 1965 to 1990. American Sociological Review, v. 79, n. 2, p. 183-210, 2014.
• ⁴⁹
  MUNIZ, Jerônimo. Sobre o uso da variável raça-cor em estudos quantitativos. Rev. Sociol. Política v. 18, n. 36, p. 277-291, 2010.
• ⁵⁰
  NOBLES, Melissa. Shades of citizenship: race and the census in modern politics. Stanford: Stanford University Press, 2000.
• ⁵¹
  OLIVEIRA, Cleuci de. One woman’s fight to claim her ‘blackness’ in Brazil. Foreign Policy, 24 jul. 2017.
• ⁵²
  OTERO, Hernan. Estadística y nación: una historia conceptual del pensamiento censal de la Argentina moderna 1896-1914. Buenos Aires: Prometeo Libros, 2006.
• ⁵³
  PASCHEL, Tianna. Becoming black political subjects Princeton: Princeton University Press, 2016.
• ⁵⁴
  PASCHEL, Tianna. The beautiful faces of my place people: race, ethnicity and the politics of Colombia’s 2005 census. Ethnic and Racial Studies, n. 36, p. 1544-1563, 2013.
• ⁵⁵
  PASCHEL, Tianna. The right to difference: explaining Colombia’s shift from color-blindness to the law of black communities. American Journal of Sociology, v. 116, n. 3, p. 729-769, 2010.
• ⁵⁶
  PETRUCCELLI, Jose L. Brazilian ethnoracial classificaiton and affirmative action policies: where are we and where do we go? In: SIMON, Patrick; PICHE, Victor; GAGNON, Amelie A. (orgs.). Social statistics and ethnic diversity: cross-national perspectives in classifications and identity politics. Londres: Springer, IMISCOE Research Series, 2015.
• ⁵⁷
  PIZA, Edith; ROSEMBERG, Fulvia. Color in the Brazilian census. In: REICHMANN, Rebecca (ed.). Race in contemporary Brazil: from indifference to inequality. University Park: Pennsylvania State University Press, 1999. p. 37-52.
• ⁵⁸
  PORTER, Theodore M. The rise of statistical thinking, 1820-1900. Princeton: Princeton University Press, 1986.
• ⁵⁹
  RODRÍGUEZ-PIÑERO, Luis. Indigenous peoples, postcolonialism and international law The ILO regime (1919-1989). Oxford: Oxford University Press, 2005.
• ⁶⁰
  SCHWARCZ, Lilia M. O espetáculo das raças: cientistas, instituições, e questão racial no Brasil 1870-1930. São Paulo: Companhia das Letras, 1993.
• ⁶¹
  SCHWARTZMAN, Luisa F. Seeing like citizens: unofficial understandings of official racial categories in a Brazilian university. Journal of Latin American Studies, v. 41, n. 2, p. 221-250, 2009.
• ⁶²
  SILVA, Nelson do V. Black-white income differentials in Brazil Tese de Doutorado em Sociologia. Universidade de Michigan, Michigan, 1978.
• ⁶³
  SKIDMORE, Thomas. Black into white: race and nationality in brazilian thought. Oxford: Oxford University Press, 1993.
• ⁶⁴
  SOMMER, Doris. Foundational fictions: the national romances of Latin America. Berkeley: University of California Press, 1991.
• ⁶⁵
  TEIXEIRA, Moema de P. Negros na universidade: identidade e trajetoórias de ascensaão social no Rio de Janeiro. Rio de Janeiro: Pallas, 2003.
• ⁶⁶
  UN - United Nations. Secretariat of the Permanent Forum on Indigenous Issues. Workshop on Data Collection and Disaggregation for Indigenous Peoples (Relatório). Nova York, 19-21 de janeiro de 2004. http://www.un.org/esa/socdev/unpfii/documents/workshop_data_ilo.doc
  » http://www.un.org/esa/socdev/unpfii/documents/workshop_data_ilo.doc
• ⁶⁷
  VEJO, Tomás P.; YANKELEVICH, Pablo (orgs.). Raza y política en Hispanoamérica Diásporas. Madrid: Iberoamericana Vervuert; Mexico City: Bonilla Artigas Editores; Mexico City: El Colegio de México, 2017.

Publication Dates

• Publication in this collection
  24 May 2021
• Date of issue
  Jan-Apr 2021

History

• Received
  11 Dec 2020
• Accepted
  22 Mar 2021