Skip to navigation – Site map

HomeNuméros7Dossier : Intégration éducative d...Who are the Blacks?

Dossier : Intégration éducative des minorités en Amérique latine

Who are the Blacks?

The Question of Racial Classification in Brazilian Affirmative Action Policies in Higher Education
Qui sont les noirs ? Construction de la classification raciale dans les politiques d’action positive concernant l’enseignement supérieur au Brésil
Luisa Farah Schwartzman
p. 27-47


Debates about racial classification and its agreement with the uses of “race” and “color” in everyday life have been central to the discussion about affirmative action in Brazil. Using quantitative and qualitative data regarding the relationship between socio-economic status and racial identification in Brazilian universities, this paper investigates how particular kinds of policies may have different impact in terms of which particular “kinds” of individuals are benefited: both the labels that are used and the socio-economic limits that are imposed may have significant and not always intuitive consequences for which individuals are admitted, and for how contestable their eligibility will become. The label negro, when used as the sole criterion for admissions, may be too restrictive and exclude “deserving” non-whites from these policies. On the other hand, because potential non-whites from higher socio-economic classes are more likely to come from “multi-racial” families, the absence of a socio-economic criterion may lead to a substantial number of candidates who may feel that they can lay claims to a wide range of racial labels, not all of which may be acceptable to policy designers and scrutinizers concerned with restricting eligibility for quotas to “deserving” candidates.

Top of page

Full text


1Since 2001, several universities in Brazil have implemented race-based quotas for undergraduate student admissions, sometimes combined with socioeconomic criteria. The question of racial classification has played a central role in public debates about affirmative action in Brazilian universities, and therefore will be the focus of this paper. Understanding racial classification in affirmative action also has broader implications, since racial classification for social policy is becoming a much larger trend in Brazil, being applied in areas outside of higher education (Nobles, 2000; Htun, 2004; Telles, 2004; Fry, 2005).

2Two important questions emerged in the public and social scientific debates regarding this trend. One question is whether this explicit “racialization” of formal state institutions and discourse will increase racial divisions in Brazilian society (Fry, 2005; Maggie, 2005; Maio & Santos, 2005) or if it will, instead, only allow government institutions to deal with the problems that arise from a society that is already divided among racial lines, but that stubbornly denies this division (Medeiros, 2004; Telles, 2004; Carvalho, 2005). This larger and more ethical question is often discussed in conjunction with a smaller, more “technical” question of whether it is possible to “objectively” (or “fairly”) classify Brazilians by race for affirmative action. Both questions are related to the empirical question of how Brazilians use and understand racial categories in everyday life and in their relationships with state institutions.

3My goal in this paper is not so much to advocate for or contest specific policies, but to provide a different perspective on the question of racial classification in Brazil that will hopefully allow for a more informed debate on this issue. In the heated debate about whether people should be classified racially for affirmative action purposes, social scientists have given less attention to how different ways of classifying can produce different effects in terms of which population the policy applies to.

4Debates on the feasibility of racial classification for affirmative action and other purposes in Brazil have often focused on the consistency between native classification systems (i.e., those generally used in everyday life) and official classification systems (see Schwartzman L.-F., s.d.). Rather than simply examining the consistency between these two types of racial classification –which presumes that native racial categories are rigid and immutable, rather than the result of on-going construction and negotiation of categories as experienced in everyday life and in previous interactions with state institutions– I suggest that we look at how students to whom affirmative action policies are applied understand official categories and their relation to everyday categorization.

5Understanding the students’ point-of-view is important because affirmative action in Brazilian universities rely (at least partly) on self-classification. This does not mean that classification processes are not being scrutinized: the most extreme example is the university of Brasilia, which, as I will discuss below, required until recently that self-selected students have their pictures analyzed by a commission and who could reject a students’ claims to eligibility. In most cases, however, the legitimacy of classification is often being observed and contested more indirectly (and with less immediate consequence), by colleagues, courts, and the general public (Peria, 2004; Tavolaro, 2006). This scrutiny –or the threat thereof– often puts constraints on the classification choices that students perceive they have.

6Classification processes assign labels to individuals. After a brief background on the current Brazilian context, I will show that which particular labels are used matters for which individuals get selected and which get excluded from the policies. In particular, I will call attention for the need for policymakers who want to design affirmative action policies based on self-classification to pay closer attention to how the potential targets of those policies understand those labels. After that, I will shift my focus to individuals and the extent to which a wider or narrower range of racial categories can legitimately be applied to them. This is important because the wider the range of racial categories that can be applied to an individual, the more likely there will be disagreement between people who are making claims on categories that will make them eligible for racial quotas and people who are scrutinizing and policing the eligibility criteria. The breadth of this range, which Teixeira (2003) calls realm of possibility, will vary across individuals, depending on their skin color, the “racial” composition of the person’s family, socioeconomic status, and the way that the latter affects skin color and family composition.


7Income inequality in Brazil is one of the highest in the world, and education mediates much of this inequality (Lam, 1992). Brazilian society is also highly stratified by skin color (Telles, 2004). For lower social strata, differences in educational attainment and family background account for most racial disparities in income and other socioeconomic indicators, but there is evidence of a “glass ceiling” where, at the top of the social hierarchy, it is harder for non-whites (pretos and pardos) to transform their socioeconomic background into educational attainment, their educational attainment into occupations, and their occupations into income (Telles, 2004; Ribeiro, 2005; Schwartzman S., 2006).

8Higher education is still the privilege of a minority of Brazilians: in 2002, only 9,8% of Brazilians ages 18-24 attended higher education. The selection of this minority is highly skewed both by color and by socioeconomic background: 76% of students in 2002 were white (branco) and a significant proportion (48% for private schools and 34% for public schools) was in the top 10% income bracket of the population. Secondary education is still a major bottleneck for access to higher education: in 2002, only 40% of people ages 15-17 were attending high school. Although the proportions of whites and nonwhites enrolled in high school is now similar to the proportions in the population, whites are still disproportionately likely to earn a high school degree. Public universities in Brazil are free and are generally regarded as providing a better education than private universities, but most higher education (70,5% of students in 2002) is done in private universities (Schwartzman S., 2004 & 2006).

9Aspiring undergraduate students apply to a specific department in a university, and take a university entrance exam, called vestibular. One’s performance in the vestibular –which will affect one’s chances of getting into the more competitive universities and departments– depends heavily on the quality of one’s previous schooling. Primary and secondary education tends to be better in private schools and in a few elite public schools. Private preparatory courses for the exam (called « cursinho pré-vestibular »), may partly compensate for some deficiencies in previous education. Public universities have less spots than applicants, so one’s score in the vestibular is important for guaranteeing access, especially in the most prestigious departments. In private universities, apart from medicine and dentistry, there is no shortage of slots relative to the number of applicants, so having a high school diploma and being able to pay tuition are usually enough to guarantee a spot (Schwartzman S., 2004).

10In sum, there is no question that Brazilian access to higher education –especially in some of the most prestigious departments and universities– is extremely exclusive in terms of both socioeconomic background and color. There is also no question that much of the selection happens long before students attempt to enter the university and that, therefore, in the long term, improvements in basic public education are needed to make universities more democratic.

11Since the 1990s, there have been attempts to affect admissions to universities In a more short-term and direct way. In the early 1990s, some non-profit “cursinhos” have emerged to help prepare lower-income students to go to universities, some of which are led by black movement activists and try to promote black consciousness » (consciência negra). In the early 2000s, many public universities started implementing quotas for admissions, either race-based or targeted at students from public schools, or both (see Telles, 2004; Htun, 2004 & Schwartzman L.-F., s.d. for an history). Different institutions reached different compromises between different views of what affirmative action policies should be about, and the merits and disadvantages of the policies and their criteria are still hotly debated.

12Two public universities have dominated the public debate and provided two different models to be followed (or avoided): the State University of Rio de Janeiro (UERJ) and the University of Brasilia (UnB). In 2001 the Rio de Janeiro State legislature passed a law required that UERJ and the other major state university, the Universidade Estadual do Norte Fluminense (UENF) to implement quotas for students from public schools and from disadvantaged racial groups. In the case of the University of Brasilia, racial quotas were first proposed by two of its anthropology faculty members in 1999 and, after much internal debate (and influenced by the UERJ experience) approved in 2002 by an internal committee (Tavolaro, 2006).

13UERJ and UnB both developed ways to deal with the possibility of “fraud” (that is, the possibility that people who were socially considered “white” would make themselves eligible for the policy). They also reached different kinds of compromises between different views regarding which “kinds” of people quota policies should target. In the case of UERJ, state government officials and representatives, university administrators and black movement activists negotiated two important modifications to the quota law. First, in contrast to the original law, which conditioned eligibility to classification as either negro (“black”) or pardo (“brown”), the new law eliminated the category pardo, because negro was already presumed to include pardo and because of the fear that too many brancos would feel comfortable with the label pardo. Second, the new law restricted quota eligibility to students with low income, giving prominence to a class-based interpretation of the legitimate beneficiaries of racial quotas (Peria, 2004; Machado, 2004; Fry, 2005; Ramos, 2005; Moraes Dia Da Silva, 2006; Tavolaro, 2006).

14From the beginning, the University of Brasilia’s policy only used the category negro. UnB decided to privilege “race” over “class” –there were no income limits nor quotas for public school students– and interpreted the label negro as designating people who have suffered discrimination. Consistent with this view, UnB’s solution to “fraud” was to require candidates who wanted to qualify quotas to have their pictures analyzed by a committee of “experts” (Maio & Santos, 2005; Tavolaro, 2006). Due to heightened public opposition to this method, UnB has recently abandoned the use of pictures, substituting it for an interview.

15Like UERJ, most universities that subsequently adopted racial quotas relied exclusively on students’ self-classification. Many included additional socioeconomic criteria, or restricted eligibility to public school students (Silva, 2006). Some, like the Federal University of Bahia, kept the categories used by the Brazilian census office: pretos (dark blacks) and pardos. Private universities were also affected by this “wave” of affirmative action through a federal government program called PROUNI (Programa Universidade Para Todos), where the federal government pays universities to open extra slots –and to offer full of partial scholarships– for low-income students a percentage of which should be self-declared negros or pardos.

Who are the “negros”?

16Before few could predict that racial quotas would be implemented in Brazil, there were debates about which labels the Brazilian census office (IBGE) should use (Nobles, 2000). This debate was not restricted to Brazilian borders: an interesting version occurred in the American journal Social Forces between Edward Telles and Marvin Harris, two U.S. sociologists who have become prominent in the field of race relations in Brazil. Harris and his colleagues (1993) argued that IBGE should use an “emic” approach to racial categorization, using terms that people actually employ in everyday life (such as “morena”), instead of an “etic” approach, where « racial categories can be defined by clusters of phenotypical traits (skin color, hair form, nose width) by trained observers. » In the 1980 census, they argued, IBGE used a mixture of an “emic” and an “etic” approach by employing terms foreign to everyday vocabulary.

17Telles (1995) challenged Harris’s “emic” approach, arguing that the difference between pardo and branco reflects a difference in social status in Brazilian society, while the difference between branco and moreno does not. While being branco and being morena/o are not mutually exclusive (the category moreno has a range of different meanings, such as brunette, sun-tanned, an euphemism for black or a racially mixed person), pardo is understood as a mixed-race (thus non-branco) term. For him, it is more important for social scientists to reliably diagnose racial inequality than to use the labels that respondents prefer.

  • 1 Evidence from my interviews with students at UERJ is in line with Telles’s claim that Brazilians un (...)

18However, Telles is not advocating a completely “etic” approach either. Since he is not challenging self-classification, his argument rests on his assessment that Brazilians may not express but nonetheless understand what pardo means. Therefore, the method he advocates still requires an understanding of how respondents interpret census categories.1

19Those in charge of designing the criteria for affirmative action did not adequately take into account how candidates would understand and sort themselves out according to the labels that the policies would use. Instead, they relied exclusively on social scientists’ and black movement activists’ understandings of these labels (Schwartzman L.-F., s.d.). In particular, policies have often been designed with the assumption that students would understand the label negro as a broad racial category which would include pardos (thus including a wide range of people with visible African descent), while in fact many students understood the label in much more restrictive ways, as applying only to people at the darker end of the color spectrum.

20The broader interpretation of the label negro was developed by an alliance between black movement activists and some (mostly quantitative) social scientists. IBGE uses the labels branco, pardo, preto (black), amarelo (yellow) and indígena (indigenous), which corresponded, respectively, to 54%, 39%, 6%, 0,5% and 0,4% of the population, in the 2000 census. Quantitative social scientists found that socioeconomic differences between brancos and pardos is higher than differences between pretos and pardos (Telles, 2004), which gave them reasons to understand pardos and pretos as a single group, originally called “nonwhite” (Hasenbalg & Silva, 1988). Some social scientists have allied with black movement activists and decided to call this nonwhite group negro (Henriques, 2001). This resounded with black movement activists who wanted to understand blackness as including mixed-race people, in order to unite Brazilians of visible African descent in a common struggle against discrimination, and to give visibility to the claim that negros constitute a significant proportion of Brazilians (40%) and thus should figure prominently in our understanding of the nation (see Nobles, 2000). Proportions of negros in each state, calculated according to this definition, were also used as rationales for advocating for large sizes of quotas to be implemented.

21However, many students I interviewed at UERJ did not understand the label negro as including pardos. The native meaning of negro is, for many, a more restricted term that applies to Afro-Brazilians of darker skin. Many interviewees also do not believe that pardos are discriminated against or deserve quotas. Though results of the interviews could be biased by my non-representative sample, a nationally representative survey confirms these results. The survey showed pictures of people with different skin colors to respondents, asking whether each person in the picture deserved racial quotas. Only a third of respondents considered that Afro-Brazilians with lighter skin tones (whom 90% of respondents labeled pardo) deserved quotas, while 60% of respondents thought that the two individuals with the darkest skin tones in the photographs deserved quotas (Bailey, forthcoming).

22One of the motivations for eliminating the pardo category was the desire by some defenders of affirmative action policies to foster a broader negro identity (Carvalho, 2005), which implied that a forced choice between branco and negro would lead many students to declare themselves as negros. However, evidence suggests that giving people a forced choice between negro and branco, even while giving them incentives related to affirmative action, leads to the exclusion of pardos as well: Rosemberg (2004) found that only about 50% of candidates for a Ford Foundation fellowship who marked pardo in a questionnaire using IBGE categories declared themselves negro in the form that influenced their eligibility and that did not have pardo as an option. Moreover, the fear that people who were socially seen as brancos but who have distant negro ancestors would take advantage of the policy led activists, policymakers and the general public to discouraged a broad interpretation of the negro label, which (probably unintentionally) reinforced students’ perception that the category negro did not include pardos for the purposes of the policy (Tavolaro, 2006; Schwartzman L.-F., s.d.).

23In order to see how categories in affirmative action policies may be affecting admissions in practice, it is useful to look at data that compares the characteristics of students who entered the university before and after quotas were implemented. For this purpose, I use responses to the socioeconomic questionnaire from ENADE (National Exam for Student Performance). ENADE is a national examination (plus a socioeconomic questionnaire) given to students in the first and last years of selected departments in all Brazilian universities, collected by the National Institute for Educational Research (INEP), an agency tied to the Brazilian Ministry of Education. ENADE’s sample is not representative of all college students, since only some departments were selected, and taking the exam was optional, but its coverage is reasonably good for the departments selected and thus general patterns observed are informative.

Figure 1: Percentage of students taking ENADE exam, by color, year, type of institution, and state

Figure 1: Percentage of students taking ENADE exam, by color, year, type of institution, and state

24Figure 1 compares students attending the first and in the last year of college, by type of institution and by state. I selected three states that illustrate a good range of variation in policy for my purposes that implemented quotas relatively early: Brasilia, Rio de Janeiro and Bahia. ENADE does not identify universities, but once we narrow down to the state and type of university, we can have a good idea about which type of policies are affecting which students. None of the last-year students entered the university through quotas (the first UERJ quota students entered in 2003), while many of the first-years did. Comparisons across Cohort, state and type of institution will thus allow us to see how different policies (and the racial labels they use) may be affecting the racial composition of the student body.

25The graph confirms the pattern, noted in the background section, that universities admit branco students disproportionately to their share in the population (54% in the 2000 census) Although the state of Bahia admitted more nonwhite than white students, it is still disproportionate to the total share of the populations, which is 25% for the state (it was 55% for Rio de Janeiro and 49% for Brasilia). Comparing first-year and last-year students across types of universities suggests that the introduction of affirmative action may have made federal universities less “white,” making their racial composition more similar to that of state and private universities.

26PROUNI alone does not affect enough students to change racial composition of private universities (Schwartzman S., 2006; Carvalho, 2006). Therefore, the rise of negros and pardos in those universities is probably due to other factors. Therefore, we will focus our attention on what is happening among public universities, leaving private universities as a point of comparison.

27How are different quota policies affecting 1) the proportion of non-whites in universities and 2) the distribution between negros and pardos, as measured in ENADE? Regarding the first question, comparisons between first- and last-year cohorts suggest that places that have implemented quotas did increase the proportion of non-whites. In Rio de Janeiro, where both state universities (UERJ and UENF) have quotas and the federal universities do not, the figure suggests a trend for federal universities to become whiter and for state universities to become “darker” (private universities also became “darker” but the change was much less radical). In Brasilia, where federal university (UnB) implemented quotas for negros (not using the category pardo), the total proportion of brancos diminished between the last year and the first year cohorts, while the racial composition of the private universities remained very similar across cohorts. In Bahia, where quotas were introduced in both federal and state universities, all types of universities decreased their proportion of brancos to some extent, though the state universities made less progress in that regard.

28Answering the second question, the distribution between negros and pardos is highly dependent on whether the label pardo was included as a criterion for admissions. If students who saw themselves as pardo understood that they were included in the negro category for the quotas, they might switch between those two labels depending on the availability of choices. It does not seem that this switching is occurring. In state universities in Rio de Janeiro, where the category pardo was not available for the 2005 cohort, there is a growth between cohorts in the proportion of negros and only a marginal increase of pardos. Similarly, the number of negros at UnB has increased more than the number of pardos. In Bahia, the federal university (UFBA) has quota that admit both students who classified themselves as pretos and pardos, while the largest state university in Bahia (UNEB) has quotas for negros. While UFBA increased admissions mostly of pardos, state universities in Bahia increased admission of negros but not pardos.

29I see two possible interpretations of the above pattern. One is that people who think of themselves as pardo are selecting themselves out of quotas that are targeted only at negros. If this is the case, policy advocates are not being consistent with statistics on racial inequality that they often cite (which suggests that pardos suffer as much discrimination as pretos and should therefore be counted as negros). On the other hand, using both labels (at least in UFBA) means privileging pardos over negros, who may be (contrary to what contemporary social scientific findings suggest) the relatively more advantaged among “non-whites”. The alternative interpretation is that people who would otherwise classify themselves as pardos decide to mark negro for entering the university through quotas. Interestingly, in this case, they maintain this classification even when responding to a survey that is not meant for admissions, and that gives them the option of marking pardo. This would mean that racial quotas are provoking a longer-term, non-opportunistic shift in racial self-identification. Those who advocate for quotas in order to foster a negro identity may be having their wishes come true (which are also the nightmares for those who fear bi-polarization of identities). In any case, being admitted to places like UERJ and UnB seems to require a longer-term commitment to the label negro.

Who are the quotistas? “Multi-raciality”, racial identity and socioeconomic status

30Popular debates on the feasibility of affirmative action in Brazil often focus on two images of the country: on one extreme, a country sharply divided between negros and brancos and, on the other, a “racially mixed” country where any racial classification is arbitrary.

31Some social scientists have decided to focus less on how Brazil is and focus Instead on how it should be. On one side, people like Maggie (2005) and Fry (2005) argue that policymakers should be cautious to not destroy the “myth of racial democracy”, i.e., the idea that all Brazilians are racially mixed and therefore racial distinctions do not or should not matter. While not a reality, they argue, this vision of Brazil is a useful utopia that can help dismantle racism. On the other side, there are those who advocate strengthening black identity and naming racial “groups” in order to fight racism (Carvalho, 2005). Regardless of whether a “racial democracy” or a “multiculturalist” utopia is better, it is important for social scientists to examine empirically how previously existing Brazilian classification systems are interacting with newly-implemented, “official” classification systems. In the long-run, this may allow us to examine how official classification may be changing Brazilian society. In the short-run, this can allow a more fine-grained understanding of which specific Brazilians are most affected by those policies.

32The fluidity of racial boundaries, and therefore the potential for disagreement over the labels to be applied to individuals, should not be discussed abstractly for the country as a whole, but should be examined 1) in relation to specific contexts of classification and 2) in relation to the specific individuals that are being classified. Contemporary scholars of race in Brazil have made significant contributions to our understanding of the first point, showing how multiple racial categorization systems coexist and how the use of one system over another varies according to the context of social interaction (Maggie, 1991; Sherriff, 2001; Sansone, 2003; Telles, 2004). In previous work, I have also been concerned with the first point, showing how classification in the current Brazilian affirmative action context – where people self-classify, where classification is often attached to perceived “deservingness” to qualify for the policy and where the process of classification is being intensely scrutinized by the public – differs from classification in everyday life for discriminatory purposes – where the question of public legitimacy is not at stake (Schwartzman L.-F., s.d.). In this section I am concerned with the second question, that is, how the degree of determinacy of racial boundaries may vary depending on which specific individuals are being targeted.

  • 2 The main difference between Brazil and the United States is that in Brazil the realm of possibility(...)

33In doing so, I use an approach that Teixeira (2003) developed while studying the racial identity of university students before affirmative action was introduced. She argues that Brazilians classify themselves within a « realm of possibilities ». She explains (Teixeira, 2003: 64): « as much as the Brazilian system of racial classification may be broad, ambiguous and encompassing, each individual keeps for herself, based on her physical features, of which the main one is skin color, a certain “limit” on the possibilities for classification both by third parties and by themselves ». This “limit”, she says, is not the same for everyone, but is broader for « the so-called pardos, or mestiços […] or for those that have characteristics that are common to brancos or negros, especially the skin color of the former and other characteristics such as hair, nose, lips of the latter »2.

  • 3 I use pseudonyms in this paper in order to protect the confidentiality of my interviewees.

34Interviews with UERJ students are consistent with Teixeira’s approach, defying both the idea that “race” is always fluid and optional, and the idea that it is a simple matter to everyone. Although many see Brazil as a “mixed country,” deciding which “race” or “color” one belongs to is not always problematic. In the middle-class world of UERJ, being branco is especially unproblematic. When asked why she thought she was branca, “Helena” laughs, then says: « Because, well, it’s visible, right? And also because of my ancestry and such »3. As many brancos in Rio de Janeiro, Helena descends from Europeans who came to Brazil in the beginning of the 20th century. But even for “Sergio,” who is not completely sure of his ancestry, marking branco in a questionnaire is not a difficult decision. I ask him what color marks he in questionnaires such as IBGE’s, to which he simply responds: « I put “branco” » Upon further probing, Sergio says that, since Brazil is a racially mixed country and since he does not know about his distant ancestors, he cannot be sure that he is branco, but since his known ancestors are brancos « since going by my skin color I’m branco, in objective terms like this, skin color, I put branco ».

35Skin color also strongly determines racial status at the other end of the color spectrum. For very dark-skinned students, the question is more about whether to use a “hard” term like negro or a “softer” term like moreno, that is, whether to downplay or display their blackness, and also whether they are “conscious” of their color (that is, whether they perceive color as a significant dimension in their life). Many times, the “consciousness” comes from everyday experiences with discrimination. “Angela,” who replied positively when I asked if she considered herself negra, describes the first time she thought about her color the following way: « I remember, it was like this, when I got into school. Because, well, I didn’t have a consciousness of my color. I was negra, but for me, I didn’t even know that. For me everyone was the same. So when I got into school, there was a little bit of joking […] So they called me monkey, and this is when I became aware of it ».

36When asked about Camila Pitanga, a TV actress who has relatively light skin color but who publicly labels herself negra, Angela responds: « Look, from her features I would say, well, that no, I would call her morena. But, I mean, she says she is negra, right? So... she is negra. She goes by her father. […] I think that the thing is like this, it depends a lot on the person, because it’s like this […] my parents are negros too. So if I were light-skinned and my parents were negros, what would I say? Am I negra or not? Her father is negro, her mother says she is [negra]. She was already born like this, with something, features not so much of a negra, I don’t know, but she says that she is negra, so I think it is very difficult to say that negro is a racial thing. I go more by the color ».

37Angela differentiates between Camila Pitanga and herself. Given Angela’s dark skin color, she has no doubts about her own status as negra. If she had lighter skin but still had parents who were negros (like the actress), it would be more difficult for her to say « what she was ».

38The question becomes more complicated for people who are from “multi-racial” families, by which I mean people whose immediate family members are seen as being of different “races” or “colors”. The word immediate here is key, for it matters less what the color of distant ancestors are. I ask Joana whether she had thought about her color sometime in her life outside of filling out questionnaires:

Joana: « On the Internet, when you are meeting someone, they ask: what do you look like? […] You say like this, I am tall, have short hair, or long hair, and such, but he will ask: and your color? […]

Author: And what do you say?

Joana: I say like this, joking, that I am cor-de-burro-quando-foge [color-of-a-donkey-when-it-runs away], color of nothing, dirty color... Because in Brazil you can’t say very well. My mother is completely white, straight hair, my father is very moreno, my sister is preta, and I have a very white little sister. And we are three colors, three totally different tones. We are three and each of us is one way, nobody would say that any of us three are sisters ».

39Joana could choose to classify herself as negra, but the question has a less straightforward answer than for other people. This is not only because her immediate family is of different “colors” but also because, within her own family, she had an intermediary skin color.

  • 4 See Schwartzman L.-F. 2007 for details on methodology.

40It seems clear from the quotes above that physical appearance, and one’s immediate family of origin (which in turn affects one’s physical appearance) are important factors that constrain one’s range of “options” for racial labels. Household survey data confirms this hypothesis. Table 1 shows what the “rules” of racial inheritance were in Brazil in 2005 for different combinations of parents and their children, where one child under 15 was randomly selected per household: children of endogamous marriages are classified like their parents most of the time, but children of mixed marriages where one parent is branco have a very high likelihood of being classified as branco. While children’s racial categorization is almost pre-determined in intra-racial marriages, inter-racial marriages lead to a less predictable outcome in terms of how children are classified. Intermarriage therefore expands the realm of possibility for the classification of children.4

Table 1: Percentages of Children Under 15 Classified in Each Racial Category by Race of Parents, 2005

Mother’s Race

Father’s Race

Child’s Race


Absolute Numbers









































































Source: Schwartzman L.-F. 2007 elaborated from PNAD 2005.
Note: one child was randomly selected per household.

41As I have shown elsewhere (Schwartzman L.-F., 2007), socioeconomic status increases both the probability that non-whites (pretos or pardos) will marry whites (brancos) and the probability that white-nonwhite couples will have children whom they classify as brancos. As we have discussed above, having people of different racial categories in one’s immediate family increases one’s realm of possibility, which means that people with both white and nonwhite relatives see a greater range of choice regarding the labels that they could apply to themselves, which partly reflects the different labels that others apply to them. The combination of the two statements in this paragraph means that potential negros or pardos of higher socioeconomic status will also be more likely to be potential brancos in certain circumstances. They are more likely to have a branco parent and to have been labeled branco by their parents, but on the other hand may still feel that the label pardo or negro legitimately applies to them.

42In the current context of affirmative action, policymakers and the general public are actively scrutinizing who qualifies for quotas and the prospect of “fraud” is a widespread concern. At the same time, claiming that one is negro (and in some cases pardo) can offer significant rewards in the form of access to free higher education. This means that, while candidates have an incentive to expand the definition of negro or pardo (which they can do by giving more weight to ancestry over skin color, or by accepting a wider range of skin tones as fitting within the classification), scrutinizers have an incentive to restrict them (which they can do by restricting the range of criteria that are considered acceptable). Therefore, the broader the range of possible labels that can be applied to a candidate (i.e., the wider a candidate’s realm of possibility), the more likely it is that there will be disagreement between the candidate and potential scrutinizers. This potential for disagreement is likely to increase with the candidates’ socioeconomic status.

43Below I will show realm of possibility is likely to be larger among potential candidates for university admissions than among the population in general and, moreover, that it may become especially large in the most selective departments (universities that have implemented quotas usually have separate quotas for each department). I will do so in two steps. First (using Table 2), I will show some of variation in selectivity by socioeconomic status across academic programs. Then (using Table 3) I will show the racial composition of the families in the socioeconomic strata that these programs serve, which is likely to influence the realm of possibilities that potential targets for these programs (and their potential scrutinizers) see for their racial classification.

44Table 2 (next page) shows the socioeconomic background of students from different types of universities and programs. Of the departments for which ENADE data is available, Education (Pedagogia) is the one that admits the most economically disadvantaged students, while Production Engineering (Engenharia de Produção) has the most economically well-off student body.

45This table shows that, in all kinds of universities, about half the students in Education have family incomes lower than three minimum wages, and more than half have fathers with less than five years of education. In comparison, Production Engineering departments are much more elitist: less than 10% of students admitted have family incomes lower than 3 minimum wages, and most have fathers who studied beyond primary school (9 years of education). Production Engineering at Federal Universities is even more selective: about 60% of admitted students have a family income of more than 10 minimum wages, almost 60% have fathers who attended college, and more than 80% have fathers who attended high school.

Table 2: Students’ family income, by course and type of university

Table 2: Students’ family income, by course and type of university

Source: ENADE, 2005.
Note: Total percentage includes missing values not included in the table. One minimum wage in 2005 was equal to R$ 300, or about US$ 130. Courses surveyed by ENADE are: Architecture and Urban Studies, Biology, Social Sciences, Computer Science, Engineering, Philosophy, Physics, Geography, History, Literature, Mathematics, Education and Chemistry.

  • 5 About 60 % of students in ENADE reported living with their parents. The age distribution for first- (...)

46ENADE does not have information on the parents’ race, but we can look at another dataset and deduce what potential candidates for affirmative action in these departments might look like. Table 3 shows the percentage of people who had at least one branco parent, among those who were 17-34 years old in 2005, lived with their parents, and had at least one preto or pardo parent (which means that they could legitimately claim eligibility for quotas).5 In general, those with sufficient education to be eligible to apply for university (10 years or more) are already more likely to have a white parent than average. In a quota system without socioeconomic restrictions, most students likely admitted to Production Engineering at a federal university would have a father that went to college, and almost all would have fathers with high school education. 45% of potential (and eligible) negros or pardos whose fathers went to high school have a white parent, while 60% of those whose fathers went to college have a white parent. A similar pattern can be observed by looking at people’s family income. These differences are especially striking compared to the average Brazilian youngster with a nonwhite parent and whose family income is the lowest and whose parents are least educated, where only a little more than 30% have white parents.

Table 3: Percentage of people ages 17-34 with one branco parent among students with at least one pardo or preto parent, by socioeconomic status, among those who live with their parents


People with 10 or more years of education

Father’s education

0-4 yrs



5-8 yrs



9-11 yrs









Family income (in minimum wages)

Up to 3


















Source: PNAD 2005.

47Therefore, the more selective the admissions process in terms of socio-economic status, the more likely they will come from “multi-racial” backgrounds and, consequently, the more likely that the range of possibilities for classification that are socially acceptable for them will cross white-nonwhite boundaries. This may also imply that there will be disagreement between the way that they decide to classify themselves and the way that potential scrutinizers of the process will find legitimate.


48Affirmative action policies in Brazil were the product of a negotiation between different groups, with different goals in mind regarding which “kinds” of people should be targeted (Moraes Dia Da Silva, 2006; Tavolaro, 2006; Moraes Dia Da Silva & Schwartzman L.-F., s.d.). Different institutions have designed criteria for admissions that were intended to reflect different compromises between these goals. However, there has been little investigation of which “kinds” of students are actually benefiting (or not) from these policies. From the discussion in this paper, we can conclude that 1) affirmative action policies that rely exclusively on the negro label may be excluding many lighter-skinned Afro-Brazilians who may still suffer disadvantages based on skin color, but who tend to classify themselves as pardo and 2) affirmative action policies, especially those in more selective departments and universities, may be admitting students from a relatively well-off segment of the population, who are disproportionately likely to come from “multi-racial” families, and therefore might have a wider realm of possibility in their choice of labels. Because this wider realm of possibility can be perceived by both candidates themselves and scrutinizers of the policy, and because they have opposite incentives how broadly to interpret labels like negro and pardo, the potential for accusations of “fraud” becomes heightened in the most elite departments. Those elitist departments admit only a tiny minority of Brazilians, but are also the most visible by the general public, and the ones where quotas are most likely to make a difference.

49The uneasiness with the difficulties of classification are, therefore, not unfounded. However, the fear that students deemed “too white” would try to quality for quotas reflects more the reality of a relatively well-off, “multi-racial” population that only now is getting access to Brazil’s most prestigious – and recently white-dominated – undergraduate programs, and less the reality of the population as a whole. For most Brazilians (and even for most of those applying to universities), the challenge is to allow people to use broader definition of the eligibility criteria, and to make it clear that policies are also aimed at Brazilians who consider themselves pardo or “mixed,” and not only those at the extreme of the color spectrum. The exclusive use of the label negro in some universities, combined with a heightened public concern over whether people who are “really negro” are benefiting from the policy, may be serving to exclude people whom the policies were designed to include, which is the majority of the Brazilian population, who does not understand themselves as negros and do not necessarily perceive discrimination but at the same time do not enjoy the privileges of whiteness. Putting socioeconomic restrictions on eligibility for racial quotas may be therefore a more sensible way of dealing with “fraud” while not excluding underprivileged people who understand racial categories differently. Policies based on socioeconomic status have also the advantage of being more widely accepted in the Brazilian population than race-based ones, as a recent survey suggests (Queiroz & Santos, 2006).

50For those concerned with designing a policy for the “black middle-class”, it is relevant for them to ask which “black middle-class” they want, that is, how “black” (negro) do they have to be either in skin color or in spirit and, once those are admitted, who should fill the remaining spots opened up by quota policies.

51This paper has not delved on the issue of how affirmative action policies are operating to change people’s racial identities (and cleavages) in the long run, although data from section 3 could be interpreted as suggesting that the use of the label negro for affirmative action has a lasting effect on classification choices for other purposes (the alternative interpretation is that it is selecting students who have previously classified as negro and leaving out those who saw themselves as pardos). However, more thorough research on this subject, particularly studies involving intensive participant observation of social interactions among students in universities that have implemented affirmative action, are needed to assess the possibility of a more long-term impact of racial categorization in Brazil.

52Finally, it is clear that the problem of access of non-white and poor Brazilians to higher education cannot be fully resolved with quotas alone, since most Brazilians are not even eligible to apply to universities. In this context, improving the quality of primary education and expanding access to secondary education is still a major challenge that needs to be resolved in order to achieve true equity and diversity in higher education.

Top of page


Bailey (S.), Forthcoming, « Unmixing the Mixed-Race Population in Brazil », American Journal of Sociology.

Carvalho (J.-J.), 2005, « Usos e abusos da antropologia em um contexto de tensão racial: o caso das cotas para negros na UnB », Horizontes Antropológicos, vol. 11, n° 23, pp. 237-246.

Fry (P.), 2005, A Persistência da Raça – Ensaios Etnográficos sobre o Brasil e a África Austral, Rio de Janeiro, Civilização Brasileira.

Harris (M.), Gomes Consorte (J.), Lang (J.) & Byrne (B.), 1993, « Who are the Whites? Imposed Census Categories and the Racial Demography of Brazil, Social Forces, vol. 72, n° 2, pp. 451-462.

Hasenbalg (C.-A.) & Silva (N.), 1988, Estrutura social, mobilidade e raça, Rio de Janeiro, IUPERJ, Vértice.

Henriques (R.), 2001, « Desigualdade Racial no Brasil: Evolução das Condições na Década de 90 », Texto para Discussão 807, Instituto de Pesquisa Econômica Aplicada, Rio de Janeiro, Brazil.

Hilu Da Rocha Pinto (P.-G.), 2006, « Ação Afirmativa, Fronteiras Raciais e Identidades Acadêmicas: Uma Etnografia das Cotas para Negros na UERJ », in J.-J. Feres & J. Zoninsein (eds.), Ação Afirmativa e Universidade: Experiências Nacionais Comparadas.

Htun (M.), 2004, « From “Racial Democracy” to Affirmative Action: Changing State Policy on Race in Brazil », Latin American Research Review, n° 39, pp. 60-89.

Lam (D.), 1992, « Age, Experience and Schooling: Decomposing Earnings Inequality in the United States and Brazil », Sociological Inquiry, n° 62, pp. 218-45.

Machado (E.-A), 2004, « Desigualdades ‘Raciais’ e Ensino Superior: Um Estudo sobre a Introdução das “Leis de Reserva de Vagas para Egressos de Escola Públicas e Cotas para Negros, Pardos e Carentes” na Universidade do Estado do Rio de Janeiro (2000–2004) », PhD Dissertation, Programa de Pós-Graduação em Sociologia e Antropologia, Instituto de Filosofia e Ciências Sociais, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil.

Maggie (Y.), 1991, « A Ilusão do Concreto: Análise do Sistema de Classificação Racial no Brasil », Tese de Professor Titular, IFCS/UFRJ, Rio de Janeiro, Brazil.

Maggie (Y.), 2005, « Mário de Andrade Ainda Vive? O Ideário Modernista em Questão », Revista Brasileira de Ciências Sociais vol. 20, n° 58, pp. 5-25.

Maio (M.-C.) & Santos (R.-V.), 2005, « Políticas de cotas raciais, os “olhos da sociedade” e os usos da antropologia: o caso do vestibular da Universidade de Brasília (UnB) », Horizontes Antropológicos, vol. 11, n° 23, pp. 181-214.

Medeiros (C.-A.), 2004, Na Lei e na Raça: Legislação e Relações Raciais, Brasil-Estados Unidos, Rio de Janeiro, DP&A.

Moraes Dias Da Silva (G.), 2006, « Ações afirmativas no Brasil e na África do Sul », Tempo Social, Revista de Sociologia da USP, vol. 18, n° 2, pp. 131-165

Moraes Dias Da Silva (G.) & Schwartzman (L.-F.), s.d., « Glocalizing Affirmative Action », Unpublished Manuscript.

Nobles (M.), 2000, Shades of Citizenship: Race and the Census in Modern Politics, Stanford (CA), Stanford University Press.

Peria (M.), 2004, « Ação Afirmativa: Um Estudo Sobre Reserva de Vagas Para os Negros nas Universidades Públicas Brasileiras: o Caso do Estado do Rio de Janeiro », Masters Thesis, Museu Nacional, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil.

Queiroz (D.-M.) & Santos (J.-T.), 2006, « Sistema de Cotas: Um Debate dos Dados à Manutenção de Privilégios e Poder », Educação e Sociedade, vol. 27, n° 96, pp. 717-737.

Ramos (C.), 2005, « “Nem tão pobres, nem tão negros”, Um estudo de caso sobre os alunos indeferidos no vestibular/2004 da UERJ », Masters Thesis, Programa de Pós-Graduação em Sociologia e Antropologia, Instituto de Filosofia e Ciências Sociais, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil.

Ribeiro (C.-A. Costa), 2005, « A Raça da Classe e Suas Desvantagens », Insight Inteligência, n° 29, pp. 80–86.

Rosemberg (F.), 2004, « O branco no IBGE continua branco na ação afirmativa? », Estudos Avançados, vol. 18, n° 50, pp. 61-66.

Sansone (L.), 2003, Blackness Without Ethnicity: Constructing Race in Brazil, New York (NY), Palgrave MacMillan.

Schwartzman (L.-F.), 2007, « Does Money Whiten? Intergenerational Changes in Racial Classification in Brazil », American Sociological Review, n° 72, pp. 940-963.

Schwartzman (L.-F.), s.d., « Seeing Like Citizens: Unofficial Understandings of Official Racial Categories in a Brazilian University » Unpublished Manuscript.

Schwartzman (S.), 2004, « Equity, quality and relevance in higher education in Brazil », Anais da Academia Brasileira de Ciências, vol. 76, n° 1, pp. 173-188.

Schwartzman (S.), 2006, « A questão da inclusão social na Universidade Brasileira », Paper prepared for the symposium Universidade e Inclusão Social Experiência e Imaginação, Universidade Federal de Minas Gerais (

Sheriff (R.-E.), 2001, Dreaming Equality: Color, Race and Racism in Urban Brazil, New Brunswick (NJ), Rutgers University Press.

Tavolaro (L. Gonçalves Magalhães), 2006, « Race and Quotas, “Race” In Quotes: The Struggle over Racial Meanings in Two Brazilian Public Universities », Ph.D. Dissertation, New School for Social Research.

Teixeira (M. De Poli), 2003, Negros na Universidade: Identidade e Trajetórias de Ascensão Social no Rio de Janeiro, Rio de Janeiro, Pallas.

Telles (E.-E.), 1995, « Who are the Morenas? », Social Forces, vol. 73, n° 4, pp. 1609-1611.

Telles (E.-E.), 2004, Race in Another America: The Significance of Skin Color in Brazil, Princeton and Oxford, Princeton University Press.

Top of page


1 Evidence from my interviews with students at UERJ is in line with Telles’s claim that Brazilians understand the meaning of the label “pardo” in ways that are consistent with IBGE’s interpretation, that is, as something that is “neither black nor white”, or as a “type of mulato”. This is not entirely accidental, since they have often “learned” the meaning of this label from previous interactions with the state, such as in the military, during census interviews or from registration in birth certificates (Schwartzman L.-F., s.d.).

2 The main difference between Brazil and the United States is that in Brazil the realm of possibility can often cross the boundary between “mixed-race” and “white” identities and, though less frequently, between “black” and “white” as well. Internal differentiation among “afro-descendants” by skin color is present in both countries, and therefore the question of whether different labels are used to mark different shades of skin color does not per se present obstacles for affirmative action.

3 I use pseudonyms in this paper in order to protect the confidentiality of my interviewees.

4 See Schwartzman L.-F. 2007 for details on methodology.

5 About 60 % of students in ENADE reported living with their parents. The age distribution for first-year ENADE students was : 27 % were younger then 20, 39 % were between 20 and 24, 15 % were between 25 and 30, 8 % was between 30 and 34 and 12 % were over 34. Pretos are the oldest group and brancos are younger than both pardos and pretos. College graduates were not excluded because, especially in more competitive courses and universities, it is not uncommon for students who already have a (usually less prestigious) degree elsewhere to apply.

Top of page

List of illustrations

Title Figure 1: Percentage of students taking ENADE exam, by color, year, type of institution, and state
File image/jpeg, 596k
Title Table 2: Students’ family income, by course and type of university
Caption Source: ENADE, 2005.Note: Total percentage includes missing values not included in the table. One minimum wage in 2005 was equal to R$ 300, or about US$ 130. Courses surveyed by ENADE are: Architecture and Urban Studies, Biology, Social Sciences, Computer Science, Engineering, Philosophy, Physics, Geography, History, Literature, Mathematics, Education and Chemistry.
File image/jpeg, 996k
Top of page


Bibliographical reference

Luisa Farah Schwartzman, Who are the Blacks? Cahiers de la recherche sur l’éducation et les savoirs, 7 | 2008, 27-47.

Electronic reference

Luisa Farah Schwartzman, Who are the Blacks? Cahiers de la recherche sur l’éducation et les savoirs [Online], 7 | 2008, Online since 01 May 2012, connection on 28 September 2023. URL:

Top of page

About the author

Luisa Farah Schwartzman

Ph.D. candidate in Sociology at University of Wisconsin, Madison.

Top of page



The text only may be used under licence CC-BY-NC-ND-4.0. All other elements (illustrations, imported files) are “All rights reserved”, unless otherwise stated.

Top of page
Search OpenEdition Search

You will be redirected to OpenEdition Search