SciELO - Scientific Electronic Library Online

vol.26 issue100The New Adult Education and Training Policy in Portugal: critic to the neoliberal governance of this sector in the context of europeanizationTransdisciplinary education: training in skills for the good living author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand




Related links


Ensaio: Avaliação e Políticas Públicas em Educação

Print version ISSN 0104-4036On-line version ISSN 1809-4465

Ensaio: aval.pol.públ.Educ. vol.26 no.100 Rio de Janeiro July/Sept. 2018 

Artigo Original

Democratizing the access to college education: Brazilian race/color classification in affirmative action’s debate1

Democratizando o acesso ao ensino superior: classificação brasileira de cor/raça no debate sobre a ação afirmativa

Democratización del acceso a la educación universitaria: clasificación brasileña de la raza/color en el debate sobre la acción afirmativa

Moema De Poli Teixeiraa 

a Universidade Federal do Rio de Janeiro. Rio de Janeiro, Rio de Janeiro, Brasil.


One of the principal aspects to develop affirmative action in Brazil comprises how to define target population, which includes uses and perceptions of ethnic/racial/color categories. The present paper has the main objective of bringing the analysis about IBGE’s2 race/color classifications contribute to the design of affirmative action in Brazil using categories historically constructed with the endorsement of official statistics. The color issues in the Brazilian Census and the experiences, including the racial designation as an open response, has been studied since 1872, and it is noted that there are two dimensions to be observed in the affirmative action debate: a structural and other cultural involving race/color classifications in Brazil. The statistics are fundamental to build the best evidence in shaping public policy. On the other hand, we must recognize ethnic and racial identities as cultural phenomena that are susceptible to change, which drives us to continue the discussion, trying to capture the meaning of these transformations. The affirmative action debate may not disqualify any of these approaches to knowledge about race relations in Brazil.

Key words: Race/color Classification; Brazilian Census; College Education; Affirmative Action


Um dos principais aspectos para o desenvolvimento de ações afirmativas no Brasil compreende como definir a população-alvo, que inclui usos e percepções de categorias étnico-raciais. O presente trabalho tem como objetivo principal trazer a análise sobre as classificações de cor ou raça do IBGE para contribuir com o desenho de ações afirmativas no Brasil, utilizando categorias historicamente construídas com o aval da estatística oficial. As questões de cor nos Censos Brasileiros e as experiências incluindo a designação racial como resposta aberta, são estudadas desde 1872, e constata-se que há duas dimensões envolvendo as classificações raciais no Brasil a serem observadas no debate sobre ação afirmativa: uma estrutural e outra cultural. As estatísticas são fundamentais para construir as melhores evidências na formulação de políticas públicas. Por outro lado, reconhece-se as identidades étnicas e raciais como fenômenos culturais suscetíveis de mudança, o que nos leva a continuar a discussão, tentando captar o significado dessas transformações. O debate sobre a ação afirmativa não pode desqualificar nenhuma dessas abordagens para o conhecimento das relações raciais no Brasil.

Palavras-Chave: Classificação de cor/raça; Censo Demográfico; Educação Superior; Ação Afirmativa


Uno de los principales aspectos para el desarrollo de acciones afirmativas en Brasil comprende cómo definir la población objetivo, que incluye usos y percepciones de categorías étnico-raciales., El presente trabajo tiene el objetivo principal de hacer que el análisis sobre las clasificaciones de raza/color del IBGE contribuya al diseño de la acción afirmativa en Brasil usando categorías históricamente construidas con el respaldo de estadísticas oficiales. Los problemas de color en el Censo Brasileño y las experiencias, incluida la designación racial como respuesta abierta se estudiaron desde 1872, y se constata que hay dos dimensiones que se deben observar en el debate de acción afirmativa: una estructural y otra cultural que involucre clasificaciones de raza/color en Brasil. Las estadísticas son fundamentales para construir la mejor evidencia en la configuración de las políticas públicas. Por otro lado, debemos reconocer las identidades étnicas y raciales como fenómenos culturales susceptibles de cambio, lo que nos impulsa a continuar la discusión, tratando de captar el significado de estas transformaciones. El debate de acción afirmativa no puede descalificar ninguno de estos enfoques del conocimiento sobre las relaciones raciales en Brasil.

Palabras-clave: Clasificación de color/raza; Censo Brasileño; Educación universitaria; Acción Afirmativa

One of the principal aspects to develop affirmative action in Brazil comprises how to define target population. Many authors recognize this (ROSEMBERG, 2004; SANTANA, 2010; RATTS; CIRQUEIRA, 2010, GONÇALVES, 2014), which includes uses and perceptions of ethnic/racial/color categories. The present paper has the main objective of bringing the analysis about IBGE’s3 race/color classifications contribute to the design of affirmative action in Brazil using categories historically constructed with the endorsement of official statistics.

Since the 1940s, when the first census conducted by IBGE occurred, population data on race/color have been used by social scientists in the analysis of race relations in Brazilian society (AZEVEDO 1955; PINTO 1952; FERNANDES, 1978).

These data and the corresponding information analysis were useful to subsidize the discussion of some concepts such as “racial democracy” and patterns of miscegenation (interracial marriage), which contributed to develop studies on racial inequalities in Brazilian society. These studies included inequalities in schooling, income distribution, access to the labor market and social security. More recently, these same data have been the substrate for the debate on affirmative action for native Brazilians and Black people in Brazil. As expected, when data leave the abstract realm of research to steer public policies, there is a resurgence of the debate about these surveys, the kind of information/data they produce and which categories they contribute to reinforce. These discussions are not entirely new and have been present at all times since prior surveys (OLIVEIRA 2003; BELTRÃO; TEIXEIRA, 2009). For this reason, it is important to recover past debate in order to understand the present and to analyze prospects for improvements that allow us to better capture the phenomenon in the best possible way.

1 Different dimensions of a racial question as an object to statistical research

Studying the race/color questions in Brazilian Censuses conducted between 1872 and 1960, Costa (1974, p. 100), came to the following conclusions:

  1. there has not been a common criterion by the censuses, which has undermined the comparability of data;

  2. both self-classification and the classification made by an interviewer were based on more than one criterion;

  3. there was a consensus about the importance of collecting this kind of data to understand the contribution that different ethnic groups give for the formation of the current population.

Concerned about the quality of data during the preparation of the 1970 Census, IBGE consulted experts on the race/color issue and their opinions pointed to the need of taking into account three levels:

  1. classification criteria;

  2. race/color terms used in everyday discourse by the population; and

  3. multiple relationships between these two levels.

Therefore, the research question would be how people classify themselves and, in doing so, which terms they normally use (COSTA, 1974).

As it is known, the consultation led to the withdrawal of the topic in the 1970 Census with the proposal that the Office would implement further studies in order to answer the question – done for the first time in 1976 with a household sample survey.

The working hypothesis were:

  1. the classification by race/color is a proxy for an ethnic/racial identification;

  2. research conducted so far indicated that the terms, primarily those of color, also included several other physical features besides skin pigmentation;

  3. the perception and the resulting classification by color is influenced by physical criteria as well as social prestige and other situations of interaction between different groups and individuals;

  4. the perception of color is expressed in a vocabulary with a rich variety of terms (HARRIS, 1970; TEIXEIRA, 1987, SHERIFF, 2002);

  5. this vocabulary is a sort of cultural manifestation – the results of the 1976PNAD4 and July 1998 PME5 clearly stated that (Table 1);

    Table 1 Distributions of responses to race/color self-identification (open answer)*. 

    Distributions of responses to race/color self-identification

    Categories PME/98 Accumulated PNAD/76 Accumulated
    Branca (White) 54.24 49.45
    Morena (Tanned, Dark haired, light Brown) 20.89 75.13 24.8 74.25
    Parda (Mixed-race) 10.4 85.53 8.47 82.72
    Preta (Black) 4.26 89.79 5.61 88.33
    Negra (Negro) 3.14 92.93 0.1 88.43
    Morena Clara 2.92 95.85 2.75 91.18
    Amarela (Asian) 1.11 96.96 1.53 92.71
    Mulata (Mulatto) 0.81 97.77 1.24 93.95
    Clara (Fair) 0.78 98.55 1.5 95.45
    Morena Escura 0.45 99.00 0.54 95.45
    Escura (Dark) 0.38 1.08 97.07
    Indígena (Native Brazilian) 0.13
    Brasileira (Brazilian) 0.12
    Mestiça (Mixed-race) 0.08 0.05 97.12
    Loira (Blonde) 0.05 0.39 97.51
    Branca + ** 0.04 0.01 97.52
    Other 0.19 2.48 100.00
    Total 100.00 100.00

    Source: IBGE, 1976; 1998.

    * Not considered in the analysis, no response, which amount ed to 0.26% in 1998 and 0.94% in 76.

    ** This category included answers with the term whit e plus some complement.

  6. then the objective would be to study the classification expressed in the vocabulary to obtain a classification which reflects the variety of criteria Brazilian population effectively use to identify themselves racially or ethnically, turning this classification comprehensible for society as a whole.

In this first experience (IBGE, 1976) IBGE asked interviewee’s designation of color in a question with an open answer. It was a way to test the use of categories and if people would use to identify themselves IBGE’s traditional categories – brancos (whites), pretos (blacks) and pardos (mixed-races). Results showed that even if interviewees declared more than 130 different terms to identify themselves by color, the most frequent categories were those used historically by the institute.

Twenty years later, IBGE conducted a similar survey as a supplement questionnaire in PME (IBGE, 1978), again as part of the discussions during the preparation of the 2000 Demographic Census. The Table 1 shows the results of these two researches conducted by IBGE trying to capture in an open question the vocabulary used by the population to identify their own race/color6.

We can perceive that branca is the modal category in both surveys, but an even higher percentage in 1998. The second most cited category is morena, which was more representative in 1976 while the reverse happened with parda category. The category preta has increased its incidence between 1976 and 1998, and most recently in a 2008 survey7 (further, this research will be better explained). Looking specifically to PME’s data and to the differences between Brazilian regions (Table 2), we can see that the perfect coincidence between the terms of closed categories and the use of their namesakes in answering the open question, defined as those with a higher percentage of coincidence – all above 90% – are the categories branca in São Paulo, Rio de Janeiro and Porto Alegre and amarela in São Paulo. The other categories have a lower degree of acceptance by the population. It is in Rio de Janeiro and Porto Alegre metropolitan areas that the categories preta and parda appear with a higher percentage of coincidence, between 52% and 60%. In Porto Alegre metropolitan area, the modal categories for the open question are always namesakes of the corresponding closed category and in the Rio de Janeiro metropolitan area, the discrepancy occurs only for the indígena population who spontaneously prefer the omnibus category of morena.

Table 2 Proportion of coincidences between closed categories of color/race and Terms in answering the open question by metropolitan areas. 

PME – JULY 1998

Metropolitan Areas Branca Preta Parda Amarela Indígena
Recife 74.5 16.9 4.2 28.3 3.9
Salvador 87.3 44.9 47.2 67.2 29.8
Belo Horizonte 75.6 27.0 9.8 40.0 7.7
São Paulo 91.7 41.6 33.0 93.8 15.6
Rio de Janeiro 94.8 59.3 52.4 44.4 14.0
Porto Alegre 95.4 62.1 52.5 48.0 41.8

Source: IBGE (1998).

Analyzing the categories of the closed question which spontaneously preferred options in the open question that were not homonymous (Table 3), we observed that in Recife metropolitan area there seems to be a dichotomy in which the population is classified in the open question as branca or morena. The amarela category is mainly in the first option and all the other closed categories in the second option. It is important to observe the preferred option of indígena by the open category morena in all metropolitan areas, with the exception of Porto Alegre (though both open and close questions mention color or race). At the other extreme we have the open homonymous option always modal for individuals classified as branca in the closed question. Between 14 cells not empty in the Table 3, 11 refer to the open option morena reinforcing the view already expressed by Silva (1996) when analyzing data from 1976.

Table 3 Proportion and categories of first option of an open category given a closed category in the case of homonymous not open is not the modal category according to metropolitan areas. 

PME – July 1998.

Metropolitan areas Branca Preta Parda Amarela Indígena
Recife - Morena 82.0% Morena 60.5% Branca 29.1% Morena 78.9%
Salvador - - - - Morena 44.7%
Belo Horizonte - Negra 33.9% Morena 70.4% - Morena 69.8%
São Paulo - Negra 42.7% Morena 59.2% - Morena 61.5%
Rio de Janeiro - - - - Morena 68.6%
Porto Alegre - - - - -
Brazil - Morena 56.7% - - Morena 64.3%

Source: IBGE (1998).

Note: In cells with no information, the open namesake category is modal.

There are other implications for the study of vocabulary and their meanings. Studies about racial and ethnic identities in addition to the multiplicity of terms and categories that can be used according to local culture (TEIXEIRA, 1987; SHERIFF, 2002) inform us that we could, in principle, recognize and distinguish at least five levels of classification:

  1. individual point of view about himself;

  2. individual point of view about someone close to him (a relative, for example);

  3. individual point of view about a person unknown, so only based on the appearance of this person;

  4. individual point of view about how he is perceived by society in general and

  5. a description of how an individual wants to be perceived in a given context (Cf. (BELTRÃO; TEIXEIRA, 2008).

This diversity may generate different perceptions, even conflict in some cases, from the way individuals understand the question to which terms and criteria are used to classify people, especially in the case of people classified as parda (census category) and morena (preference of respondents in surveys with open response). There are many examples given by field researches.

In 2008, a new research entirely dedicated to investigate the dimensions involved in ethnic and racial classification of population was conducted by IBGE. With a sample of 15.000 households in five Brazilian states and the Federal District: Amazonas, Paraíba, Mato Grosso, São Paulo, Rio Grande do Sul and Distrito Federal, the research provides a methodological innovation: the exclusive selection of one single respondent per household to guarantee the color or race self-declaration. The research sought to reproduce in an objective questionnaire, the approach of field research on racial identity. So the PCERP started the survey with a question that sought to introduce the subject to the respondent, asking him about the importance of such information: 63.9% said they believe that color or race have influence on people’s lives in Brazil.

In sequence, an initial inquiry different from the traditional question was made. In place of “what is your color or race”, it was asked whether the respondent “could tell what color or race”: 96% of respondents said yes, that they could say what their color or race were. When responding, openly, what would be the majority of, respondents ranked among six most frequent categories, representing about 93% of total respondents (Table 4). Among the alternatives, we can see that there was an increase in the identification as negro (category relating to family origin) over as preto (category relating to color), compared to the previous experience conducted by PME in July 1998. It is noticed that the more frequent categories remain the same: branco (white), moreno (dark), negro, preto (black), pardo (brown) (Tables 4 and 5).

Table 4 Frequency of race/color categories by state (open answer) – PCERP 2008. 

Color or Race AM PB SP RS MT DF
Branca 15.9 30.7 51.1 59.5 29.8 28.6
Morena 48.9 45.9 18.6 11.8 34.5 19.8
Parda 22.9 10.5 14.1 3.3 17.9 27.4
Negra 3.4 4.5 9.2 5.1 10.3 10.8
Preta 1.7 0.8 1.3 1.6 1.8 0.7
Aamarela 1.6 1.0 1.6 0.4 0.9 1.0
Sub Total 94.4 93.4 95.9 81.7 95.2 88.3
Total 100.0 100.0 100.0 100.0 100.0 100.0

Source: IBGE (2008a).

Table 5 Frequency of race/color categories by state (close answer) – PNAD 2008. 

Color or Race AM PB SP RS MT DF
Amarela 0.7 0.1 1.4 0.1 0.4 0.4
Branca 22.4 37.2 64.4 80.8 39.2 41.4
Indígena 0.4 0.2 0.2 0.4 1.4 0.3
Parda 73.9 57.2 28.4 13.8 50.9 49.6
Preta 2.6 5.2 5.4 4.9 7.9 8.1
Sem declaração 0.1 0.0 0.1 0.1 0.2 0.1
Total 100.0 100.0 100.0 100.0 100.0 100.0

Source: IBGE (2008b).

Comparing these data with the PNAD’s data in the same year and in the same Brazilian states investigated by PCERP, we realize that, in relative terms, the PNAD category “parda” could represent the sum of “brown” and “dark” PCERP categories in all states. It is interesting to note that the “white” declaration is much higher in the PNAD, showing that when people who declare themselves as “white” before the classic question with closed response alternatives can choose they prefer other terms, possibly under the category compound adjectives: branco brasileiro (white Brazilian); branco moreno (dark white), etc. In other words, when left to choose the categories in which they better identify, fewer people identify themselves as “white” and more people identify themselves as “negros” (except from Paraíba state).

The survey also included an interviewee’s classification by color or race made by the interviewer, to investigate the level iii mentioned above, i.e., the dimension of race/color classification related to how a person (the respondent) can be perceived by society (the interviewer). The results are presented below (Table 6).

Table 6 the respondent color/race self-declaration according to the classification given by the interviewer – PCERP 2008. 

self-declaration Interviewer

Branca Morena Parda Negra Preta Amarela Indígena Outras Total
Branca AM 74.7 8.8 14.2 0 0 1.1 1.1 0.1 100
PB 76.9 0.2 21.3 1.6 0 0 0 0 100
SP 91.5 2.1 5,.3 0.1 0 0.3 0 0.7 100
RS 92.7 2.1 4.1 0.3 0 0 0.1 0.7 100
MT 85.9 2.6 10.6 0.7 0 0 0 0.2 100
DF 70.1 7.7 16.4 0 0 0 0 5.8 100

Morena AM 7.5 40.2 43.6 5.1 2.0 0.4 1.0 0.2 100
PB 17.8 7.6 55.1 15.5 3.2 0. 0 0.8 100
SP 29.6 21.8 39.0 7.0 0.9 0,.4 0 1.3 100
RS 43.8 21.6 22.7 7.1 3.1 0 0 1.7 100
MT 12.4 21.1 54.3 10.6 0.8 0 0.8 0 100
DF 19.3 28.1 42.2 9.5 0 0 0 0,9 100

Parda AM 16.1 22.9 58.8 0.8 0.2 0.5 0.2 0.5 100
PB 26.5 5.5 65.9 1.5 0.6 0 0 0 100
SP 24.8 13.0 54.8 5.9 1.1 0.1 0 0.3 100
RS 43.9 10.3 39.3 3.4 1.7 0 0.5 0.9 100
MT 4.2 13.9 24.3 56.2 1.0 0 0.4 0 100
DF 18.6 34.6 39.0 7.8 0 0 0 0 100

Negra AM 2.7 24.8 28.3 44.1 0 0 0 0.1 100
PB 7.3 1.2 257 65.0 0.8 0 0 0 100
SP 2.2 11.2 16.2 67.3 2.7 0 0 0.4 100
RS 7.7 9.0 17.6 51.8 13.0 0 0 0.9 100
MT 18.6 25.7 37.8 15.8 0.8 0 1.2 0.1 100
DF 2.4 29.9 5.8 61.9 0 0 0 0 100

Preta AM 0 35.3 25.1 10.3 18.2 0 11.0 0.1 100
PB 0 0 84.1 15.9 0 0 0 0 100
SP 1.6 5.5 21.6 45.1 25.8 0 0 0.4 100
RS 2.5 12.1 6.6 58.8 18.0 0 0 2.0 100

Amarela AM 32.0 48.6 5.8 0 0 13.6 0 0 100
PB 38.1 0 61.9 0 0 0 0 0 100
SP 14.7 2.1 4.2 0 0 71.7 0 7.3 100
MT 7.7 28.9 18.5 36.8 6.3 0 1.7 0.1 100
DF 16.1 43.6 40.3 0 0 0 0 0 100

Indígena AM 4.6 2.9 20.7 0 0 0 68.0 3.8 100

Outras AM 18.5 32.6 39.5 3.2 0 4.8 0 1.4 100
PB 34.2 2.9 52.8 10.1 0 0 0 0 100
SP 49.9 7.3 26.5 4.6 0 5.3 0 6.4 100
RS 67.7 8.1 18.0 0.9 1.5 0 0.8 3.0 100
MT 43.1 16.8 23.7 13.0 0 0 0 3.4 100
DF 34,0 22,5 31,8 11,7 0,0 0,0 0,0 0,0 100,0

Source: IBGE (2008a).

Among people who see themselves as “white”, it is not so much considered by the interviewer, especially in Amazonas, Paraíba, Mato Grosso and the Federal District, where 14.2%, 21.3%, 10.6% and 16.4% were respectively classified as “mulatto”. In São Paulo and Rio Grande do Sul the classification as “white” was more consistent between interviewee and interviewer. Anyway, this is for all categories that offer the highest degree of consistency between the two types of classification − by the self and by the “other”.

Among those who classified themselves as “browns”, the greater consistency with regard to the classification by the interviewer happened in Paraíba 65.9%, followed by Amazon, 58.8 and São Paulo, 54.8%. In Rio Grande do Sul, more people − 43.9% − were considered “white” by the interviewer while in Mato Grosso, 56.2% were considered “black”. In the Federal District, 39% were also considered by the interviewer as “browns” and 34.6% as “dark”.

Among those who said they were “black” there was reasonable consistency; over 50% to 67% in Rio Grande do Sul, Distrito Federal, Paraíba and São Paulo.

Regarding the classification of “black”, it is perceived that this was a category used more by the interviewee than by the interviewer, since the majority of which were classified as qualified by other words, in Paraíba, 84.1% as “brown”; Amazonas, 35.3% as “brown” and 25.1% as “mulato”, 45% in São Paulo and Rio Grande do Sul 58.8% as “black”.

Self-declaration of “yellow” appears more consistent with the declaration of the interviewer in São Paulo – 71.7%. In Amazonas, 48.6% were declared as “dark”; in Paraíba 61.9% as “mulatto”; in Mato Grosso 36.8% as “black” and 28.9% as “dark” while in the Federal District 43.6% were considered “dark” and 40.3% “brown”.

The self-declared “indigenous” only appeared in the Amazon and 68% were also considered as such by the interviewer.

Regarding the criteria, we know that in Brazil the color or race of people has always been associated with the phenotypic features or the phenotypic characteristics (NOGUEIRA, 1985). The PCERP also investigated this dimension of ethno-racial classification, detailing the main criteria (up to 3, in order of importance) that the society uses (in interviewee’s opinion) and the criteria that each respondent used to characterize himself. The criteria were identified based on literature by skin color, physical traits in general, descent, social and economic origin (social class), political or ideological motivation or cultural traits.

The chart below presents the results of a combination of criteria to set the color or race, both for the general population (according to the interviewee’s opinion) as to the actual interview. We added alternatives to “phenotypic features criteria (physical features and skin color) as Code1, “origin criteria” (culture/tradition; family origin; social and economic origin) as Code 2 and “political and ideological criteria” as Code 3. We can observe that most interviewees state three criteria (a question asked for information of up to 3) and very few (less than 1%) do not report any criteria. It is also interesting to note that in describing the behavior of the “other” respondents reporting a greater number of criteria in which to inform its own color or race (Graphic 1).

Source: IBGE (2008a).

Graphic 1 Criteria’s combination for color/race definition according to number (up to 3) of criteria used. 

In the Table 7 we can see the distribution of the criteria by type in the order of importance in which they appeared, in the opinion of the interviewee, as the criteria used by society in general and to himself. We realize, for example, that the “phenotypic features criteria” appear in first place in both situations, but more representative of the people in general (70%) than for itself (60%). Moreover, rules of origin are more representative to set the color or race of that person (38%) than that used for people in general (28%). Otherwise, political and ideological criteria are more representative of people in general and appear in descending order of importance, more like third criterion, then as a second and less as the first criterion. It is also interesting to note that people use fewer criteria to define their own color or race than what they think people use in general to classify others. This can be verified in the amount of non-responses in the second and third positions declared by the interviewee talks about people in general and when they speak of themselves. We can think of two possible reasons for this: that the respondents consider that more criteria is needed to set the color or race of people in general or that they are trying to represent the thinking of many people about it. Anyway, the fact that most people do use more than one criterion to identify themselves or to identify the other and as a second more frequent criterion is the family origin (even more frequent to identify themselves) may explain why people use this criterion in the first place to identify themselves to be a candidate to affirmative action. In other words, thinking in only one criterion, policymakers think in skin color or “phenotypic features criteria” while candidates manipulate their identities putting family origin in the first place to define themselves. Although in most cases both criteria may be coincidental, people with particular “phenotypic features” have the corresponding descent. But we know that in many cases the criteria for family background have a more flexible scope for manipulating the identity. This also explains why many institutions have resorted to photos, as UnB8 or for personal interviews, so they could ensure the application of the phenotypic features criterion on the process of identifying candidates to affirmative action9.

In this sense we can say that the advent of affirmative action policies have posted a new dimension to the agenda: people come to racially characterize based on criteria of “descent” to have access to seats reserved for negros, black or brown. Evidence of these changes in grading standards, in our view related to change of criteria, can be seen in official statistics by increasing the population self-reported black and brown (see data from PNAD in Table 8).

Table 7 Distribution by type of criteria in order of importance that appeared to people in general and to the interviewed himself – PCERP 2008. 

Type of criteria To people in general To the interviewed

First Second Third First Second Third
Did not answer (0) 0.69 17.30 35.72 0.41 25.84 42.85
Phenotypic features (1) 69.97 46.26 23.53 60.25 42.14 25.04
Origin (2) 28.68 34.91 38.77 38.21 30.92 30.63
Ideological-Political (3) 0.57 1.49 1.82 0.39 1.01 1.41
Other (7) 0,08 0,04 0,16 0,74 0,09 0,07

Source: IBGE (2008a).

Table 8 Distribution of race/color categories – PNAD 1999, 2004, 2008 e 2015. 

Race or Color 1999 2004 2008 2015
Amarela (Asian) 0.5 0.4 0.6 0.5
Branca (White) 54.0 51.3 48.4 45.2
Indígena (Indian) 0.2 0.2 0.3 0.4
Parda (Brown) 40.0 42.2 43.8 45.1
Preta (Black) 5.4 5.9 6.8 8.9
Sem declaração (None) 0 0 0.1 0
Total Global 100,0 100,0 100,0 100,0

Source: IBGE. (1999; 2004; 2008b; 2015).

This change in the way of qualifying appears to be occurring in young age groups, as can be seen in the Graphics 2, 3 and 4.

Source: IBGE (2008b; 2015).

Graphic 2 White population distribution by age groups. 

Source: IBGE (2008b; 2015).

Graphic 3 Black population distribution by age groups. 

Source: IBGE (2008b; 2015).

Graphic 4 Mixed-race population distribution by age groups. 

The question seems to be how to bring this experience of color or race surveys of official statistics for the universities to implement their affirmative action programs since people seem to respond equally to surveys that have different characteristics and purposes.

Based on this research everything indicates that the different views that comprise the universe of Brazilian racial/ethnic classification cannot be understood from a single question in a survey, whatever its nature, may it be a census or a registration form in the entrance exam to university.

We cannot fail to highlight the issue of the nature or objectives of the survey, which also may influence how individuals can be identified. Rallu, Piché and Simon (MORNING, 2005, p. 243) describe four types of governmental approach to ethnic enumeration:

  1. Enumeration for political control (compter pour dominer),

  2. Non-enumeration in the name of national integration (ne pas compter au nom de l’intégration nationale),

  3. Discourse of national hybridity (compter ou ne pas compter au nom de la mixité), and

  4. Enumeration for antidiscrimination (compter pour justifier l’action positive).

In this sense, different directions can be perceived in the processes of survey conducted by the IBGE census of population and the most recent surveys conducted by universities, which keep a tighter relationship with the fourth type of approach mentioned above. And by all indications, this latest move of the surveys for the purpose of affirmative policy is to reflect on how the population has traditionally responded to the question in surveys of the IBGE.

The PCERP also investigated categories of family origin. The distribution of combinations of declaration of origin in the color / race declared shows that the declaration of origin is the single most common source for those who said, among all groups of color and race, being higher among those who declared yellow (81.5%), followed by those who declared white (78.6%) (Table 9).

Table 9 Distribution of combinations of origin declaration by complexity of response according to colour stated – PCERP 2008. 

Branca Morena Parda Negra Preta Amarela Total da Amostra
No declaration 25.9% 52.6% 40.0% 40.8% 45.4% 7.3% 35.0%
Single origin once declared 78.6% 67.5% 60.8% 58.7% 59.1% 81.5% 72.9%
Multiple origins once declared 21.4% 32.5% 39.2% 41.3% 40.9% 18.5% 27.1%

Source: IBGE. (2008a).

The Table 10 presents the same information from the previous table, but with the specification of combinations of origins most cited. The category with the highest incidence among all groups of color/race was declared a single origin in the “Far East” to the yellow (72.6%), followed by European origin among whites. Indigenous origin exclusively was the category of those who classified themselves in moreno (14.5%), pardos (Browns) (15.2%) and Blacks (15.3%), and incidences very close together. Among Blacks, the exclusive African origin (17%) was the combination of origins most cited. However, among those who declared themselves black, the exclusive Indian origin was the most cited (15.3%), followed by Africa alone (14.9%) and by the combination of African and European (14.5%). Among the yellows, the most frequent origin was unique Far East (72.6%).

Table 10 Distribution of origin declared combination by colour declared – Answers more significantes – PCERP 2008. 

Origin Branca Morena Parda Negra Preta Amarela
None 25.9% 52.6% 40.0% 40.8% 45.4% 7.3%
sul americana 0.8% 1.0% 1.0% 0.2% 0.5% 0.0%
extremo oriente 1.4% 2.5% 1.6% 0.5% 0.9% 72.6%
oriente médio 0.6% 0.1% 0.4% 0.1% 0.0% 0.0%
norte-americana 0.1% 0.0% 0.0% 0.0% 0.0% 0.0%
judaica 0.0% 0.1% 0.1% 0.0% 0.0% 0.0%
indígena 4.5% 14.3% 15.2% 11.8% 15.3% 0.3%
européia 50.1% 10.5% 11.4% 5.1% 0.8% 2.7%
centro-americana 0.1% 0.0% 0.1% 0.2% 0.0% 0.0%
africana 0.8% 3.5% 6.6% 17.0% 14.9% 0.0%
européia+indígena 6.1% 5.0% 6.9% 1.2% 0.5% 2.3%
africana+européia 2.3% 1.6% 4.0% 5.8% 14.5% 0.2%
africana+indígena 0.6% 2.7% 3.4% 9.6% 4.0% 0.6%
européia+indígena+africana 1.5% 2.0% 3.8% 4.6% 0.0% 0.4%
européia+extremo oriente 1.8% 0.2% 0.8% 0.0% 0.0% 4.3%
indígena+extremo oriente 0.4% 0.8% 0.8% 0.5% 0.0% 5.4%
européia+sul americana 0.6% 0.3% 0.1% 0.0% 1.5% 0.0%
indígena+sul americana 0.4% 0.3% 0.7% 0.1% 0.0% 0.0%
européia+extremo oriente+indígena 0.3% 0.3% 0.4% 0.1% 0.0% 1.3%

Source: IBGE (2008a).

2 Final considerations: dimensions for analysis

Observing all these attempts to understand the uses of a terminology to ethnically or racially characterize Brazilian population, we conclude that it is necessary to see this field of study involving racial/ethnic classifications in Brazil as comprising two different approaches to two major strands of research. One that works with cultural and social representations, where we discuss what our national identity would be and which ethnic categories could define it, if they are representations of color, race or origin - who is white, black or mixed race in the country, ultimately, who we are, what we want to be as we elaborate our ethnic and cultural identity, how we conceptualize people and nation, how we represent ourselves and others according to our own notions and concepts, and also considering personal views of how each one develops its own identity based on its references, experiences and history.

Another strand of research, I think, refers to ways of approaching the same levels of social reality through broader, more comprehensive representation and relate it to what we have become from those principles that guide our actions. It is within this perspective that we treat (or invent) analytical concepts such as social, racial, or gender inequalities. We are talking here from the perspective of social structures and how they operate specifically in real society.

It is evident that these two tests intercommunicate and interact in terms of social reality, one helping the understanding of the other. We are talking about the effect of specific looks that different social scientists bring to the subject, which are complementary in my view. Many of the arguments and misunderstandings in respect to affirmative action policies can be translated by the confrontation between these two points of view that represent different areas of knowledge that have always worked together in building the field of studies on race relations in Brazilian society while these studies were not brought to the public policy arena.

These two forms of analysis involve only different approaches to the same social reality. In the first approach, we emphasize the ethnic and racial diversity with its richness and variety of cultural identities of almost endless regionalism. Portrayed in such analysis, Brazil is dynamic and changes easier than portrayed in the second approach, the social structure that allows us to speak about the same country in terms of its great polarities. From this perspective it is possible to represent this country, often in the form of two Brazils, of the poor and of the rich ones (people that have access to goods and services), the Brazil of whites and the Brazil of nonwhites (or blacks), the Brazil of those who have employment and of those who are unemployed, the Brazil from those who have access to education and of those who do not. That Brazil, in the structural point of view, is more difficult to change. This Brazil is a more lasting. It is organized and structured upon inequalities that seem to be crystallized in society, being difficult to overcome in the short term. And precisely on that point lays the strength of his analysis, the emphasis on more permanent aspects in our social reality.

In Brazil the investigation of color and race of the population exists since the first Census of 1872, with little variation (1872: whites, blacks, browns and shifting cultivators; 1890: whites, blacks, mestizos, caboclos, 1940, 1950, 1960 and 1980: whites, blacks, browns, yellows, 1991 and 2000: the same, with the inclusion of indigenous people). Very differently from other Latin American countries, we have a long history of surveys about this issue, with all problems and criticisms we can make. This information has enabled Brazil to discuss the level of structures.

We can say that data since census of 40, have enabled researchers to point to the underprivileged situation that blacks and browns have enjoyed in Brazilian society – studies of Unesco in the 50s: Costa Pinto analyzed data for Rio de Janeiro (PINTO, 1952); Thales de Azevedo for Bahia (AZEVEDO, 1955); Florestan Fernandes studied data from 1950 for São Paulo (FERNANDES, 1978). Since 1976 when it was made a particular PNAD to prepare the return of the item to the census of 80 − it was not part of the Census of 70 − many researchers have done tests that seek to denude the Brazilian racism by official statistics: in IBGE, Tereza Cristina Araújo, Rosa Maria Porcaro, and Lucia Elena de Oliveira (OLIVEIRA; PORCARO; ARAÚJO, 1985); in IUPERJ, Carlos Hasenbalg (HASENBALG, 1979) and Nelson do Valle Silva, 1996, who formed a group of researchers currently working on the topic, have focused on these same data as well as many others. More recently, specifically after the Durban Conference in 2001, even the IPEA10 has incorporated new researchers working on such data in the same direction and tackling racial inequalities.

Since the success on demonstrating it through the “impartiality” of numbers, the questioning about the existence of racial prejudice in Brazil has stopped. Some social investments were made, but it seems that they were insufficient to alter significantly the structural framework of racial inequalities. Data analysis of the 90s – from the PNAD – may demonstrate this. This shows that we are dealing with an issue that lies at the level of the social structure and thus makes these initiatives take time to reflect on a structural change.

But with changes in the political sphere depicted by the movement succeeded in favor of affirmative action, some began to doubt these sixty years of studies of racial inequality based on official statistics. When we bring this knowledge to support the projects of intervention in society through affirmative action policies, especially in higher education (where the bottleneck was always bigger and just a change only in the base would take many years to bear fruit on it) takes a step back the discussion of what is believed to be overtaken by social scientists who study the subject.

In my view, this is a battle that reflects these two dimensions of analysis of the racial issue. This is the point that is shown in this article. The statistics are fundamental to build the best evidence in shaping public policy. If they are not good or appropriate at the time as in the past, also because we move forward in their study, and we know how we can improve them. Foolhardy to think they are good for nothing, is false or any other adjective that serves to disqualify all reviews of several researchers that in these 50 years of studies have sought to advance the knowledge of the racial issue in Brazil from official statistics.

On the other hand, we must recognize ethnic and racial identities as cultural phenomena that are susceptible to change, which drives us to continue the discussion, trying to capture the meaning of these transformations. Until the late 40’s we had in formation the ideology of the three races in Brazil – whites, blacks and indians, the foundation of the idea that we are primarily a country of “mestizos”. The analysis sought to uncover the persistent racial prejudice and discrimination in Brazilian society, used the device of joining the census categories “blacks” and “browns” in the same category of origin (negro) − which might also be regarded as a synonym of African descent − once they also came close in terms of socioeconomic background. Those who used this trick always knew that this analytical strategy approximate sizing of the problem and not all “browns”, of course, have “black” origin. By the 1991 census − when the Indians and their descendants have earned a separate category in order to identify themselves − they, even with a relatively small (less than 1% of the population) were always aggregated under the term “brown”. Prof. José Jorge de Carvalho (2005) makes a very interesting reflection on the browns of Midwest Brazil, in his view, historically identified more with whites than with blacks. At this moment of reflection on ways of ethno-racial classification and its implications for the construction of public policies based on it, we need to resume the studies done in the past, recognizing their merits and achievements, without forgetting their limitations. In addition, in that sense, we must remember that one thing is to build a category of analysis as an artifice, and quite another is to bring it to the real world of individual or collective identities.

We must think about the fact that making this implementation – the structural analysis for the reality of diversity and complexity – has contributed to the design of a framework that may be headed towards strengthening also ideologically, or in the cultural field, that way bipolarizing the representation of the country. In fact, the discussions around the projects of Affirmative Action have walked in this direction, to polarize the debate around the issue of race in the country, to the point of bringing these two directions of research mentioned before, that of interconnecting factors of the different perspectives of the same issue, has gone to join the same vision about it. That is precisely the discussion that, in my view, must be intensely debated. The debate may not be by a way of disqualification of any of these approaches to knowledge about race relations in Brazil. If so, we will be giving “a shot in our own foot”. What happens is that those who formulate public policy seem to have in mind the structural model as dichotomous and those who have applied to politics may be driving racial-ethnic origins on the basis of other criteria, such as family origin. In this sense, we need to ask ourselves if whether this is exactly what we want – for example, get like the Americans of the past, for whom a drop of black blood makes one person as black, or if, unlike this, we reaffirm our vision mixed and pacifist, or if we want something different, to overcome our inequalities, stating our ethnic and cultural diversity in an even wider way. We need to reflect if we are not substituting one bias for another, closing the doors of affirmative action for those who wish to assert a different identity – e.g. of the mestizo or mixed race, or whatever that represents the racial mix, but without that connotation of “false consciousness of black” of the 70s and the 80s. That is why the affirmative action draft should be alert to what is being proposed not to leave out exactly one population segment – also deleted – that to be seen included in the project has to be stated as “black” perhaps denying their own proposal of ethno-cultural identification.


AZEVEDO, T. As elites de cor: um estudo de ascensão social. São Paulo: Companhia Editora Nacional, 1955. (Brasiliana, v. 282). [ Links ]

BELTRÃO, K. I.; TEIXEIRA, M. D. P. “Moreno” escolha dos brasileiros desde 1940: o quesito “Aberto” de cor no recenseamento geral. Rio de Janeiro: IPEA, 2009. Textos para Discussão na Escola Nacional de Ciências Estatísticas (ENCE). [ Links ]

CARVALHO, J. J. Inclusão étnica e racial no Brasil: a questão das cotas no ensino superior. São Paulo: Attar Editorial, 2005. [ Links ]

COSTA, T. C. N. A. O princípio classificatório ‘cor’, sua complexidade e implicações para um estudo censitário. Revista Brasileira de Geografia, Rio de Janeiro, v. 36, n. 3, p. 91-106, jul./set. 1974. [ Links ]

FERNANDES, F. A integração do negro na sociedade de classes. São Paulo: Ática, 1978. [ Links ]

GONÇALVES, M. A. R. Políticas educacionais, ações afirmativas e diversidade. Crítica e Sociedade: Revista de Cultura Política, Uberlândia, v. 4, n. 1, p. 142-162, 2014. [ Links ]

HARRIS, M. Referential ambiguity in the calculus of Brazilian racial identity. Southwestern Journal of Anthropology, Albuquerque, v. 14, no. 4, p. 1-14, 1970. [ Links ]

HARRIS, M. The Brazilian pattern. In: HARRIS, M. Patterns of race in The Americas. New York: Walker and Company, 1964. p. 112-24. [ Links ]

HARRIS, M. et al. Who are the whites? Imposed Census categories and the racial demography of Brazil. Social Forces, Oxford, v. 72, no. 2, p. 451-462, 1993. https://doi:10.2307/2579856Links ]

HARRIS, M.; KOTTAK, K. C. The structural significance of Brazilian racial categories. Sociologia, São Paulo, v. 25, n. 3, p. 203-208, 1963. [ Links ]

HASENBALG, C. A. Discriminação e desigualdades raciais no Brasil. São Paulo: Graal, 1979. [ Links ]

HUTCHINSON, H. W. Race relations in a rural community of the Bahian Recôncavo. In: WAGLEY, C. (Ed.). Race and class in rural Brazil. Paris: Unesco, 1952. p. 16-46. [ Links ]

IBGE. Pesquisa das Características Étnico-Raciais da População: microdados. Rio de Janeiro: IBGE, 2008a. [ Links ]

IBGE. Pesquisa Mensal de Emprego - Julho. Rio de Janeiro: IBGE, 1978. [ Links ]

IBGE. Pesquisa Mensal de Emprego - Julho. Rio de Janeiro: IBGE, 1998. [ Links ]

IBGE. Pesquisa Nacional por Amostra de Domicílios. Rio de Janeiro: IBGE, 1976. [ Links ]

IBGE. Pesquisa Nacional por Amostra de Domicílios: microdados. Rio de Janeiro: IBGE, 1999. [ Links ]

IBGE. Pesquisa Nacional por Amostra de Domicílios: microdados. Rio de Janeiro: IBGE, 2004. [ Links ]

IBGE. Pesquisa Nacional por Amostra de Domicílios: microdados. Rio de Janeiro: IBGE, 2008b. [ Links ]

IBGE. Pesquisa Nacional por Amostra de Domicílios: microdados. Rio de Janeiro: IBGE, 2015. [ Links ]

KOTTAK, C. Emics and Etics of Racial Classification in Brazil: Based on a Recent National Survey. In: AMERICAN ANTHROPOLOGICAL ASSOCIATION, 91., 1992, San Francisco. Annual Meeting... San Francisco, 1992. Paper Presented. [ Links ]

KOTTAK, C. Race relations in a bahian fishing village. Luso-Brazilian Review, Madison, v. 4, no. 2, p. 35-52, 1967. [ Links ]

MAGGIE, I. Aqueles a quem foi negada a cor do dia: as categorias de cor e raça na cultura brasileira. In: SEMINÁRIO INTERNACIONAL SOBRE RACISMO E RELAÇÕES RACIAIS NOS PAÍSES DA DIÁSPORA AFRICANA, 1992. Rio de Janeiro, 1992. Trabalho apresentado. [ Links ]

MORNING, A. Ethnic Classification In Global Perspective: a Cross-National Survey of The 2000 Census Round. New York: New York University, 2005. Disponível em: <>. Acesso em: 12 abr. 2017. [ Links ]

NOGUEIRA, O. Negro político, político negro. São Paulo: Edusp, 1992. [ Links ]

NOGUEIRA, O. Tanto preto quanto branco: estudo de relações raciais. São Paulo: T. A. Queiroz Editor, 1985. [ Links ]

OLIVEIRA, J. S. de. Brasil, mostra a tua cara: imagens da população brasileira nos censos demográficos de 1872 a 2000. Rio de Janeiro: ENCE: IBGE, 2003. (Textos para Discussão, n. 6). [ Links ]

OLIVEIRA, L. E.; PORCARO, R. M.; ARAÚJO, T. C. O lugar do negro na força de trabalho. Rio de Janeiro: IBGE, 1985. [ Links ]

PETRUCCELLI, J. L. A cor denominada. Rio de Janeiro: IBGE, 2000. (Textos para discussão, v. 3). [ Links ]

PIERSON, D. Negroes in Brazil. Chicago: University of Chicago, 1951. [ Links ]

PINTO, L. A. C. O negro no Rio de Janeiro: relações de raças numa sociedade em mudança. São Paulo: Companhia Editora Nacional, 1952. [ Links ]

RATTS, A.; CIRQUEIRA, D. M. Mas quem é negro no Brasil?: uma contribuição para o debate acerca das cotas raciais nas universidades brasileiras. Revista Universidade e Sociedade, Brasília, DF, ano 10, n. 46, p. 51-60, jun. 2010. [ Links ]

ROSEMBERG, F. O branco no IBGE continua branco na ação afirmativa? Estudos Avançados, São Paulo, v. 18, n. 50, p. 61-66, jan./abr. 2004. ]

SANJEK, R. Brazilian Racial terms: some aspects of meaning and learning. American Anthropologist, Washington, DC, v. 73, no. 20, p. 1126-1143, 1971. [ Links ]

SANTANA, E. B. As políticas públicas de ação afirmativa na educação e sua compatibilidade com o princípio da isonomia: acesso às universidades por meio de cotas para afrodescendentes. Ensaio: Avaliação e Políticas Públicas em Educação, Rio de Janeiro, v. 18, n. 69, p. 736-760, out./dez. 2010. [ Links ]

SCHWARTZMAN, S. Fora de foco: diversidade e identidades étnicas no Brasil. Novos Estudos CEBRAP, São Paulo, v. 55, p. 83-97, nov. 1999. [ Links ]

SENRA, N. Estatísticas desejadas (1822-c.1889). Rio de Janeiro: IBGE, 2006. (História das Estatísticas Brasileiras, v. 1). [ Links ]

SHERIFF, R. E. Como os senhores chamavam os escravos: discursos sobre cor, raça e racismo num morro carioca. In: MAGGIE, I.; REZENDE, C. B. (Org.). Raça como retórica: a construção da diferença. Rio de Janeiro: Civilização Brasileira, 2002. p. 213-243. [ Links ]

SILVA, N. V. Morenidade: modo de usar. Estudos Afro-Asiáticos, Rio de Janeiro, v. 30, p. 79-95, 1996. [ Links ]

SKIDMORE, T. Bi-racial U. S. v.s. Multi-racial Brazil: is the contrast still valid? Journal of Latin American Studies, Cambridge, v. 25, no. 2, p. 373-386, 1993. [ Links ]

TEIXEIRA M. P. A questão da cor nas relações e representações de um grupo de baixa renda. Estudos Afro-Asiáticos, Rio de Janeiro, v. 14, p. 85-97, 1987. [ Links ]

TEIXEIRA, M. P.; BELTRÃO, K. I. O eu e o outro: a alteridade próxima na declaração de cor no quesito aberto da PME 98. Rio de Janeiro: ENCE: IBGE, 2008. (Textos para discussão; n. 24). [ Links ]

WAGLEY, C. Introduction. In: WAGLEY, C. Race and Class in Rural Brazil. Paris: Unesco, 1952. p. 7-14. [ Links ]

ZIMMERMAN, B. Race relations in the arid sertão. In: WAGLEY, C. (Org.). Race and Class in Rural Brazil. New York: Columbia University: Unesco, 1952. p. 82-115. [ Links ]

2Instituto Brasileiro de Geografia e Estatística. National agency responsible for official statistics in Brazil. The first Brazilian Census was in 1872 and it already contained information about race/color population (free and slaves). The IBGE was founded in 1937 during the Estado Novo period.

3 National agency responsible for official statistics in Brazil. The first Brazilian Census was in 1872 and it already contained information about race/color population (free and slaves). The IBGE (Instituto Brasileiro de Geografia e Estatística) was founded in 1937 during the Estado Novo period.

4 PNAD – Pesquisa Nacional por Amostra de Domicílios is a National Household Survey conducted every year by IBGE.

5 PME – Pesquisa Mensal de Emprego was a Monthly Employment Survey also conducted by IBGE until February 2016.

6 It’s important to explain that when we mention PNAD, data express the Brazilian state, while PME data refers to the metropolitan region of a city.

7 In 2008, IBGE conducted a survey on race/color and origin named PCERP – Pesquisa das Características Étnico-Raciais da População Brasileira (Survey of the Ethnic-Racial Characteristics of the Brazilian Population).

8 University of Brasília (Federal District).

9 In the exact moment I was reviewing this article to publish it I read news that The Planning Ministry edited a Norm regulating the entrance of blacks candidates to the public service by Cotas. According to the norm, a commission to corroborate candidate’s race/color classification using heteroclassification or phenotypic criteria must interview all candidates. (see

10 IPEA – Instituto de Pesquisa Econômica Aplicada is a Brazilian Federal Public Foundation linked to the Ministry of Planning, Development and Management.

Received: November 05, 2016; Accepted: January 11, 2018

Informações da autora

Moema de Poli Teixeira: Assessora Plena/Departamento Acadêmico/Fundação Cesgranrio. Doutora em Antropologia – Museu Nacional/Universidade Federal do Rio de Janeiro. Contato:


Updated version of an oral presentation at the XXIX International Congress of the Latin American Studies Association (LASA). October 6-9, 2010), Toronto, Canada.

Creative Commons License  This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License, which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.