Introduction

The aim of this paper is to examine the impacts of class condition on the performance of Brazilian students in basic education. The relationship between social origin and success or failure in formal education is recurrent in academic literature, in Brazil as well as around the world (^{Barbosa, 2009}; ^{Soares; Collares, 2006}; ^{Bourdieu, 2008}; ^{Bowles; Gintis, 1974}).

Despite this, even if inequalities remain at the center of public debates at the beginning of this century (^{Scalon; Salata, 2016}; ^{Piketty, 2014}), the quality of Brazilian basic education has not been progressing, as indicated by the poor results often found in evaluations such as the Program for International Student Assessment (PISA). The intersection of these phenomena justifies paying careful attention - in a continuous updating - to the determinants of our students’ performance focusing on multiple areas, one of which is class condition.

The theoretical/empirical approach of Pierre ^{Bourdieu (2008}, 2007) is at the basis of our study. As a measure of educational performance, we used the data for 2013 of the country’s main large-scale educational assessment, the Basic Education Evaluation System (Saeb). Saeb is composed of tests that assess proficiency in Portuguese and mathematics, in addition to contextual questionnaires on the lives of students and characteristics of schools and teachers.

Wide-ranging evaluations should be considered only one of the measures for assessing an education system (^{Dourado, 2007}). There are many discussions about reducing the notion of educational performance to proficiency in just two school subjects. On the other hand, this does not detract from the fact that knowing how to read and write, while interacting with a variety of texts, and knowing how to perform logical and mathematical operations are basic skills for life in society.

This paper is divided into three parts: first, an overview of the relationship between education and inequality; next, we discuss the methodology of our empirical work, presenting how we applied the concept of social class and a description of statistical techniques; finally, we discuss the research’s main results and address the weight and particularities of the influence of social class on the educational performance of Brazilian students.

Social Inequalities and Formal Education

The second half of the twentieth century saw the consolidation of a whole field of studies that started to discuss the bases of formal education, reflecting on the role that schooling plays as a central institution for the reproduction of inequalities in modern capitalist societies. Neomarxists approaches argue that the school would serve to reproduce the social relations of production. The function of the school would be to prepare a passive workforce well adjusted to the economic hierarchies of capitalism. Differences in school performance would mirror work inequalities. The school would be the means for reinforcing the ideology of the dominant classes, making young workers adapt from an early age to the dictates of capital (^{Althusser, 1980}; ^{Baudelot; Establet, 1975}; ^{Bowles; Gintis, 1974}).

It was Pierre ^{Bourdieu (2008}, 2007) who gave a broader outline to the thesis that educational systems help in the reproduction of social inequalities. For the French sociologist, society must be understood through the notion of social space, a space of social positions and power relations in dispute and movement. The basis of the notion of social space is the idea of difference, or of a gap, in a complex set of coexisting but distinct positions, “[...] exterior to one another and which are defined in relation to one another through their mutual exteriority and their relations of proximity, vicinity, or distance, as well as through relations of order, such as above, below, and between” (^{Bourdieu, 1996}, p. 18-19).

In modern times, the social space was fragmented in different fields, which started to have autonomy in relation to each other. These fields are microcosms with their own rules, in which agents dispute positions of power on the basis of the volume and composition of the capital they own. ^{Bourdieu (2008}) talks about four types of capital, of which two are determinants of the performance of agents in different fields, and, thus, of the position of these agents in the social space. Economic capital is defined in terms of income, relationship with the means of production and possession of material goods. Cultural capital is divided into three different, interrelated dimensions:

Cultural capital can exist in three forms: in the embodied state, i.e., in the form of long-lasting dispositions of the mind and body; in the objectified state, in the form of cultural goods (pictures, books, dictionaries, instruments, machines, etc.), which are the trace or realization of theories or critiques of these theories, problematics, etc.; and in the institutionalized state, a form of objectification which must be set apart because, as will be seen in the case of educational qualifications (author’s emphasis), it confers entirely original properties on the cultural capital which it is presumed to guarantee (^{Bourdieu, 2007}, p. 74).

What does the school have to do with all this? According to Bourdieu, the global cultural stock (cultural capital) of nuclear families explains the success or failure of students in educational institutions. “In all fields of culture, theater, music, painting, jazz, cinema, the students’ knowledge is so much richer and more extensive the higher their social origin” (Bourdieu, 2007, p. 45). In addition to direct contact with cultural goods and family training, young people from privileged classes inherit knowledge, tastes and cultural dispositions that fit the demands of institutionalized education.

The social space has a synchronic dimension (properties, capital) and a diachronic dimension that is shaped by modal trajectories, i.e. routes more often traveled by agents, connecting particular origins and destinations and representing the typical forms of capital appropriation (^{Bertoncelo, 2013}). The composition of the agents’ capital and the modal trajectories are the two most relevant elements to differentiate relative positions in the social space, since they form the main lines of division and conflict in modernity. The positions in the social space differ relationally in the face of a variety of social determinants (the class condition).

Social groups, and especially social classes, exist twice, so to speak, and they do so prior to the intervention of the scientific gaze itself: they exist in the objectivity of the first order, that which is recorded by distributions of material properties; and they exist in the objectivity of the second order, that of the contrasted classifications and representations produced by agents on the basis of a practical knowledge of these distributions such as they are expressed in lifestyles. These two modes of existence are not independent, even as representations enjoy a definite autonomy with respect to distributions (^{Bourdieu, 2013}, p. 111).

To socialize in this or that region of the social space means that the agent has incorporated a set of dispositions for action, a class habitus, systems of long-lasting and transposable dispositions that inform the perceptions, actions and classifications of agents in different social fields. The habitus acts as a pre-reflexive principle that coordinates the agents’ perceptions and classifications, in addition to establishing probabilities of action. Through the mediation of habitus, the generation of classifiable and classifying practices, the objective social space of the first order is transformed into a symbolic space and the practices of agents become signs of distinction. The habitus organizes the production of a homology between the symbolic space and the social space, making the hierarchy of lifestyles appear as a new expressive translation of the social space’s objective dissimilarities.

In short, the theoretical model proposed by Bourdieu implies the overlapping of three analytical schemes: the social space, the habitus and the space of lifestyles, with the habitus being the element that mediates the transmutation of objective differences into distinctive signs. It seems clear, thus, that the social space is, simultaneously, a structure of objective relationships, which is at the origin of the perception, classification and action schemes that guide practices, and a set of strategic places from which the agents struggle around the appropriation (and appreciation) of capital and the imposition of principles of classification and (di)vision of the world (author’s emphasis) (^{Bertoncelo, 2016a}, p. 75).

Social conflicts in modernity mainly revolve around symbolic disputes over access to scarce goods and resources, but this does not mean disregard for the objective conditions for survival. On the one hand, symbolic disputes are conditioned, in part, by the objective properties of social space; on the other hand, they influence the social space itself (^{Wacquant, 2007}).

^{Bourdieu (2008}, 2007) argues that the school privileges the privileged, as it does not take into account cultural disparities among students before to their entry in the school’s game. Similar methods, similar forms of evaluations and similar pedagogies are applied to people with very different class habitus, especially with regard to cultural capital. “It is, in short, the inner logic of a system whose objective function is to preserve the values ​​which are the basis of the social order” (Bourdieu, 2007, p. 56).

Agreeing or not with Bourdieu, more diverse research approaches began to guide studies on education, society and inequalities. Ethnomethodologists argue that students’ failure or success in educational establishments should be studied as practical constructions of agents. ^{Charlot (2000}) proposes that the school has to be seen as a place that goes beyond social differentiation, but that congregates and organizes itself around the transmission of knowledge.

The relationship between social conditions and schooling results is addressed in the recent works of ^{Duncan and Murnane (2011}), both Americans. They point out that the widening gap between the academic achievement of children growing up in wealthy and poor families is a reality in the United States. Between 1978 and 2008, the gap between the reading test scores of children from high and low income families grew by one third. This is reflected in a growing gap in completed schooling. In the past 20 years, the rate of wealthy children who completed college has increased by 21 percentage points, while the rate of college graduation of children from low-income families has increased by only four percentage points (Duncan; Murnane, 2011).

In Brazil, ^{Barbosa (2009}) analyzes the school institution, examining the social relations internal to the school, to study the reasons or, at least, the school factors that can be associated with the increase in social inequalities (Barbosa, 2009, p. 21). Barbosa’s research shows the strong effect of the presence of parents at home, the differences in social resources in families headed by men and women, the mother’s occupation and available time, the help received at home and the number of children in the family. Moreover, Barbosa (2009, p. 183) summarizes “[...] the effect of the school can, at least in part, reduce the effects of the social position of students on their school performance.”

In detailing the class structure of Brazilian society, ^{Costa Ribeiro (2014}) observes some decrease in the impact of social origin on Brazilian educational outcomes in recent decades. The author also warns that “[...] analyzes show that, even when we take into account the type of school, socioeconomic inequalities in opportunities remain” (^{Costa Ribeiro, 2011}, p. 79).

Methodology

This study uses the database of the Basic Education Evaluation System (Saeb), launched in 1990 by the National Institute for Educational Studies and Research Anísio Teixeira (INEP) of the Ministry of Education (MEC). Saeb is a system for sample evaluation of Brazilian students’ performance conducted every two years. Together with information on the quality of teaching, Saeb produces data that make it possible to better understand the external and internal conditions involved in learning, through questionnaires answered by students, teachers and school directors. Saeb also collect information on the schools’ physical structure and the didactic and pedagogical resources available at each school.

Saeb prepares evaluations for students from all regions and education systems. The math and Portuguese language tests are applied to students in the 5th and 9th years of basic education and to students in the 3rd year of high school, providing data on student achievement in three key moments of basic education.

Saeb is divided into a census component and a sample component. The census evaluation includes public schools with more than 20 students, from the 5th to the 9th grade of basic education. The sample evaluation involves private schools with 10 or more students in regular classes, in addition to public and private schools with 10 or more students in regular classes attending the 3rd year of high school. In order to allow more specific analyzes, another sample is added including public schools with 10 to 19 students.

Saeb editions are comparable at least since 2003, due to the similarities in sample planning. Students are divided into six groups of fundamental populations, according to the grade they are attending and the administrative affiliation of the school. A cross-classified sampling of students is used, considering that it is not possible to select student by student.

Indicators

In order to perform correspondence analysis, which demands the use of categorical variables, we built an indicator of capital volume (class condition in Bourdieu) with the levels very low, low, medium and high. The cutoff point between each category was based on the mean and more or less one standard deviation. It means that students in the very low capital volume category scored between zero and 10 points on the capital scale; students in the low category scored between 11 and 15 points; those in the medium category scored between 16 and 20 points; and students in the high capital volume category scored between 21 and 47 on the scale that indicates the volume of economic and cultural capital of students and their families. Students were distributed as follows among the different categories:

Source: The author / Saeb 2013

Graph 1: Relative Distribution of Students by Capital Volume 

To perform multivariate regression, we used individual categorical variables to indicate the objective dimension of the concept of class in ^{Bourdieu (2013}), based on economic capital and cultural capital. Five variables for economic capital and five variables for cultural capital were used. With regard to economic capital, since it was not possible able to use family income (information that young people often do not know), or the occupation of parents (information that the database does not provide), we worked with indirect measures, based on the existence of comfort goods in the students’ households (^{Buchmann; Dalton, 2002}; ^{Willms, 1992}). We used these variables and others, such as early entry in the labor market and the size of the residence.

^{Bourdieu’s notion of cultural capital (2007}) is rather polysemic and difficult to apply to quantitative research (^{Soares; Collares, 2006}, p. 621). One of the main sociological studies that quantitatively operationalized the concept of cultural capital is Paul ^{DiMaggio’s (1982}). The author focused on the access to high-culture symbolic goods, such attending theaters, exhibitions and concerts. In Brazil, it seems more appropriate to assess this cultural dimension on the basis of aspects relevant to the acquisition of academic knowledge, such as reading habits and the education of parents (Soares; Collares, 2006). In other works, we operationalized the concept of cultural capital by measuring cultural resources or by measuring different categorical variables of family education level or the student’s cultural habits (^{Caprara, 2017}, ²⁰¹⁶, ²⁰¹³).

We operationalized cultural capital here on the basis of two of its three dimensions: the capital institutionalized in the family and the capital incorporated by the student. The institutionalized cultural capital is measured by the high educational level of the students’ parents (complete college education). It is possible to think of this variable also as an incorporated cultural capital, as a kind of inheritance transmitted by the family to the students, considering that these, attending basic education, are in the process of primary institutionalization of their cultural capital, not yet possessing themselves school certifications. However, we chose to think of the parents’ high level of education as a cultural capital institutionalized in the family, as this family heritage may or may not be transmitted to the children. Incorporated cultural capital, on the other hand, is measured by the habit of reading books in general and literary books in particular, as part of the habitus acquired by the students.

Due to the large number of questions in Saeb, at first we grouped a set of variables to assess aspects of the objective dimension of the students’ class condition, schooling and faculty, in addition to demographic variables. Analyzes of these questions showed that, for the application of the chosen statistical techniques, the vast majority of the questions could be grouped into dichotomous categorical variables (such as 0 and 1), or should be transformed into binary (dummy) or ordinal categorical variables (1, 2, 3, etc.). In multivariate regressions, the dependent variables are proficiency in Portuguese language and proficiency in mathematics, measured in a scale ranging from zero to 500 points for each subject. Below we can see that both dependent variables have a normal distribution.

Source: The author / Saeb 2013

Graph 2: Histogram for Proficiency in Portuguese Language 

Source: The author / Saeb 2013

Graph 3: Histogram for Proficiency in Mathematics 

The standard deviation for both dependent variables was not very high, being much less than half the average (in Portuguese, mean of 217.38, standard deviation of 55.299 and median of 215.89; in mathematics, mean of 227, 53, standard deviation of 54.258 and median of 227.03).

Social Classes and Educational Performance in Brazil

Initially, it is interesting to look at a synthesis of the distribution of scores across different social classes. Among students from social classes with very low volume of capital (economic and cultural), 62.8% scored very low or low in Portuguese, and only 9.1% scored high. Among those from classes with high capital volume, 35.2% obtained very low or low math scores, while 29% scored high.

Table 1 shows the distribution of the volume of capital among students from different regions. In the Northeast region, only 7% of students have a high volume of capital. This region also has the highest rate of students possessing very low and low volume of economic and cultural capital - 30.2% and 43.8%, respectively. In short, 74% of students have a low volume of capital. In the South, on the contrary, we find the highest rate of students with a high volume of capital among all regions: 21.6%. The southern region also has the lowest number of students among the lower brackets of capital volume, with 7.1% in the very low bracket and 35.6% in the low bracket - totaling 42.7%.

We should also examine the association between class and scores and the effects of economic and cultural capital on proficiency in mathematics and Portuguese. Figure 1 shows the multiple correspondence analysis of capital volume, school quality and scores in Saeb 2013, for cases from all over Brazil.

Source: The Author / Saeb 2013

Figure 1: Multiple Correspondence Analysis of Capital Volume, Scores in Portuguese and Mathematics and School Quality 

Dimension 1 explained 47% of the variation in the data, while Dimension 2 explained 34.2%. The left quadrants (Dimension 1) evidenced the associations between high scores in mathematics and Portuguese language, high volume of capital and high quality of the school. On the other hand, in the right quadrants, we can observe the associations between very low scores in Portuguese and mathematics, very low volume of capital on the part of students and their families and schools of very low quality. Figure 1 also shows an opposition between these modal groups: on the one hand, high levels (of schooling, capital, scores) and, on the other, very low levels for the same categories.

The regression model that best explained the variation of the students’ scores found an adjusted R² of 0.36, indicating that 36% of the proficiency in Portuguese obtained by the evaluated students can be explained by the set of independent variables that make up the model. We did not find collinearity between the variables and the relationships were all statistically significant at the 99% level. Table 2 shows the ranking of effects of independent variables on educational performance in Portuguese language, considering the class condition of Brazilian students.

Seven of the ten variables used to operationalize class condition stood out in this model. The class variable with the greatest impact was the one indicating whether the student works outside the home, besides studying; working students obtained, on average, 15.062 points less than those who do not work. This is an economic capital variable. And this is an important piece of information, since working outside the home during basic education studies probably reveal an economic need typical of poor families. The impact of this variable makes it clear that belonging to classes that need to dedicate part of their time to work and earn money, while also studying, negatively affects scores in Portuguese language.

Another economic capital aspect, owning a computer at home, which may seem commonplace in times of widespread adoption of information technologies, showed to have a strong impact, with students who have computers at home scoring, on average, 9.685 more points on Saeb 2013 in Portuguese. Students whose family owns a car tend to score, on average, 2.352 more points. The size of the student’s residence also matters. Students who have three or more bathrooms at home obtained, on average, 4.846 points more than those who have less than three bathrooms in their homes.

Cultural capital is the other objective class dimension. Students who are regular readers of books on general topics scored, on average, 10.253 more points than students who do not read books. Those who usually read literary books obtained, on average, 2.569 more points than those who do not read this type of book. Another class variable, capable of reflecting a more cultural aspect of social class, is the mother’s education level. Students who have mothers with complete college education scored, on average, 5.649 more points than those who do not have a mother with a college degree. These effects are shown in Graph 4.

Source: The Author / Saeb 2013

Graph 4: Economic capital and cultural capital and performance in Portuguese language 

Let us now examine the particularities of the effects of economic and cultural capital on performance in mathematics. Among the models found by stepwise regression, the one that was able to better explain the variations in math scores of the students evaluated by Saeb 2013 obtained an adjusted R² of 0.30. We can see that 30% of the variation in math proficiency can be explained by this model - with statistical significance at 99% and without collinearity between the variables. Table 3 presents information for the whole of Brazil. The list of impacts of the variables goes from the greatest positive effect to the greatest negative effect.

Of the ten variables that indicate possession of capital - five of economic capital and five of cultural capital - eight were found to have important effects in multiple linear regressions: three of cultural capital and five of economic capital. The social class variable that stood out concerns students who work outside the home, besides studying. This is an economic capital variable. These students scored, on average, 11.589 less points than students who do not work, but only study. Close behind, we found that owning a computer at home is an important element. Students who have a computer at home obtained, on average, 8.211 more points than those who do not have this consumer good. Students whose families own a car tend to score, on average, 4.209 more points. The influence of social class on performance in mathematics, in economic terms, is also evidenced by the variables that indicate the students’ housing conditions. Students living in a home with three or more bathrooms tend to obtain, on average, 4.802 more points. Students living in a home with three bedrooms or more tend to score, on average, 2.670 more points.

Let us continue the examination of the influence of class on educational achievement. The mother’s education also stands out with regard to class cultural aspects. If the mother has a complete college education, students scored, on average, 4.465 more points. But the most relevant effect found is that of the variable that indicates reading habits. Students who are regular book readers tend to obtain, on average, 6.726 more points. On the other hand, students who usually read literary books obtain, on average, 2.163 more points. This allows us to observe the impact of social class in different dimensions, indicating that a larger home, the possession of a personal computer, a family with a higher education level and the habit of reading books have a direct, positive impact on the performance in mathematics in Brazilian basic education. Graph 5 shows all the effects of the economic and cultural capital variables on proficiency in mathematics.

Source: The Author / Saeb 2013

Graph 5: Economic capital and cultural capital and performance in Mathematics 

The variable that indicates working students deserves a closer look. So far, we have been treating this variable as an indicator of economic capital, considering that the need for early entry in the labor market tends to be related to the need for supplementing the family income in some degree. However, if we think about it, this is a variable that can establish a boundary between economic capital and cultural capital, as students who do not work have more free time, which thus can be spent in various activities, including cultural activities. ^{Souza (2015}) addresses this issue, tangentially, arguing that the reproduction of class inequalities involves the fact that classes possessing medium and high volumes of capital can hire low-skilled workers to perform domestic tasks, such as cleaning and maintenance, freeing up time to invest in strategies to increase capital volume. A daily life with more free time, in which students do not need to work to help their families, can reveal a connection between the possession of economic capital and the increase in cultural capital.

Among the variables of most impact on performance in Portuguese and mathematics, there are some that indicate elements apparently unrelated to social class. Their main effects are found in variables of individual life stories, in the school system, in the urban or rural location of the school, in the involvement of families that encourage the engagement in studies, in rates of school abandonment and failure and in the students’ skin color. So far, we have treated these relationships as diverse indicators, unrelated to the class condition of students and their families. It turns out that we can raise some valuable questions: who are the students who drop out of school, fail or study in rural schools, for example? How are they distributed across classes with different volumes of economic and cultural capital in Brazilian schools? What is their skin color? Do high scores, school system and encouragement to study from the family might have any association with differences in access to the two fundamental forms of capital?

Let us start by describing how students are distributed according to their volume of capital and the school system they attend. Among private school students, 52% possess a high volume of economic and cultural capital; while among public school students, only 13.5% possess a high volume of cultural capital. It seems that families with higher volumes of capital tend to invest in a private education, which may indicate that cultural capital is being institutionalized by students. This may also be linked to the assumption that private schools are better than public schools, which makes sense when they are compared with the state and municipal public systems, but not when compared with the federal system. Among private schools (2.7% of the sample), 65.2% present a high quality level. In the state and municipal systems (42.7% and 54.3% of the sample), we found 15.2% and 10.9% of schools with a high quality level. Among federal schools, 77.3% have a high quality level. In the federal system, 70.9% of students obtained high scores in Portuguese; in the private system, they were 50.4%.

Given this scenario, we performed a multiple correspondence analysis between school system, volume of capital possessed by students and their families and the scores in Portuguese and mathematics.

Source: The Author / Saeb 2013

Figure 2: Multiple Correspondence Analysis between School System, Capital Volume and Scores in Portuguese and Mathematics 

Dimension 1, which explained 48.3% of the data variation (Dimension 2 explained 35.3%), shows the existence of an association between private and federal systems and the high volume of capital possessed by students and their families, in addition to the proximity to high scores. Figure 2 also shows an opposition between these systems and the very low scores on that axis. However, this set of variables related to the volume of capital also provides a measure of economic capital, so it is necessary to verify these associations with the use of a specific variable for cultural capital. In this sense, we adopted as a proxy the categorical variables of education of the mother and father, in order to observe whether there is an association between high education levels and attending public or private schools. Figure 3 shows the multiple correspondence analysis between parent’s educational level, school system and scores of Saeb 2013 participants.

Source: The Author / Saeb 2013

Figure 3:  Multiple correspondence analysis between mother’s educational level, school network and scores in Portuguese and Mathematics 

The right quadrants (Dimension 1, which explained 43.6% of the variation) clearly show the association between students from federal and private schools and a high educational level of parents, as well as with high scores in both evaluated subjects. In Dimension 2, which explained 30.7% of the variation, we can observe the opposition between low educational level of parents (incomplete and even complete basic education), private and federal systems and high scores.

The associations found between the school system and the volume of capital/educational level of the family may indicate that studying in federal and private schools means that students are in the process of initial institutionalization of their specific cultural capital, in terms of school qualification. If this is correct, if attending a federal or private school is an indicator of this process of partial institutionalization of cultural capital by young Brazilians, our empirical results expand its scope, showing an even greater effect of the social class of students on their school performance. School level transitions also impacted Saeb scores, with students in the 9th year of basic education and in the 3rd year of high school achieving better results than those in the 5th year of basic education. Is there a possible relationship between advancing in school levels and the class condition of the agents? It is worth observing the distribution of students in terms of capital volume across the three school years analyzed. Among students in the 5th year of basic education, 61.5% are located in the low capital brackets (19.9% ​​very low and 41.6% low). Only 13.1% of the youngest students evaluated by Saeb 2013 possess a high volume of economic and cultural capital. At the end of basic education (9th grade), this scenario changes a little, but the disparity remains: 13.8% have very low volume and 39.4% have low volume of capital, totaling 53.2% of students. Only 15.3% possess a high volume of capital.

In the 3rd year of high school, in which are the students who have already overcome all the system’s barriers, we observe some important differences: only 8.6% have a very low level of access to economic and cultural capital. Adding this value to the 30.9% that possess a low volume of capital, we find 39.5% of students pertaining to the most vulnerable classes - a figure much lower than in other years. Students finishing high school also have the highest rate of access to economic and cultural capital, reaching 27.9% of them. There is a great difference between the beginning of schooling and its last stage, with regard to differences in the distribution of capital among students and their families. Figure 5 shows the multiple correspondence analysis between years, volume of capital and scores in mathematics and Portuguese, visually revealing the relations of distance and proximity between the modalities.

Source: The Author / Saeb 2013

Figure 4:  Multiple Correspondence Analysis between School Year, Capital Volume and Scores in Portuguese and Mathematics 

Dimension 1 explained 50.8% of the variance and demonstrated the association between being in the 3rd year of high school, having a high volume of capital and obtaining high scores. It also showed the opposition of these characteristics with very low scores and the 5th year of basic education, as well as with very low access to capital. This opposition is also evidenced by Dimension 2, which explains 34.8% of the variation in the data. The effect of school level transitions on the dynamics of internal struggles in schools seems to have some relation with the class condition of the students.

The involvement of the family in the students’ school life can be examined in view of the differences in access to economic and cultural capital. The effect of having parents or guardians who encourage students to study was clearly visible in the scores obtained in Portuguese and mathematics. Among students who do not receive this encouragement (only 2.4% of the sample), we find 27.8% of students with very low volume of capital and 41.8% with low volume, reaching 69.6% of students. This number is even more relevant if contrasted with the small group that has a high volume of capital and has no family encouragement to study (11.8%). Figure 5 shows the results of the multiple correspondence analysis between family encouragement, volume of capital and students’ scores.

Source: The Author / Saeb 2013

Figure 5: Multiple Correspondence Analysis between Family Involvement, Capital Volume and Scores in Portuguese and Mathematics 

Dimension 1 explained 44.5% of the variation in the data, while Dimension 2 explained 34.1%. Dimension 1 shows the associations of very low scores with the lack of family encouragement to study, as well as with the possession of very low volume of capital. In this axis, there is an opposition between these characteristics and high scores and high volume of economic and cultural capital. The non-involvement of the family in school life seems to be a more localized phenomenon affecting mainly students from the most vulnerable classes.

Two other important effects are related to dropping out of school and repetition of the school year (failure). Students who have already dropped out of school and/or have failed at least once tend to obtain worse scores than those who never did. We treat these variables as indicators of individual life stories, but we suspect that they may be linked to the students’ social class. To address this issue, starting with students who drop out of school, we ran a contingency table between the students’ volume of capital and dropping out of the school, to check how many students have already dropped out of school in the different levels of economic and cultural capital volume. Among students who have already dropped out (31% of the sample), 22.8% have a very low volume of capital and 42% have a low volume. Together, these categories represent 64.8% of the students who have already dropped out of school. It represents a very significant share of students with low and very low access to forms of capital that are essential to living in modern societies. Only 12.2% of the students who dropped out possess a high volume of economic and cultural capital.

With this evidence, we performed a multiple correspondence analysis between scores in Portuguese and mathematics, volume of capital and dropping out of school. We looked into the relations of proximity and distances between the characteristics to verify whether dropping out of school can indicate class belonging to some extent. Figure 6 presents a visualization of these categories.

Source: The Author / Saeb 2013

Figure 6: Multiple Correspondence Analysis between Dropping Out of School, Capital Volume and Scores in Portuguese and Mathematics 

Dimension 1 and Dimension 2 explained 45.7% and 34.5% of the data variation, respectively. Dimension 1 reveals the association between very low scores in mathematics and Portuguese, a history of dropping out of school and very low volume of capital. On the other hand, we can observe the association between high scores and high volume of capital. This also shows an opposition between these groups of characteristics. Dimension 2 reveals an opposition of low scores and high scores and high volume of capital. The proximity of dropping out of school to a very low volume of economic and cultural capital, together with the distance to the high volume of capital category, may signal that the episode of disruption of school attendance is more associated with classes that have reduced access to the most important forms of capital in modern societies, and that this phenomenon affects less the wealthy classes.

We carried out the same operations for the relationship between failing a school year and volume of capital. Among the students who have already failed a year (48.5% of the sample), 22.1% have a very low volume of capital and 42.8% have a low volume of capital. Only 11.1% of students who have already failed a year possess a high volume of capital. Figure 7 maps these categories.

Source: The Author / Saeb 2013

Figure 7: Multiple Correspondence Analysis between Failing a School Year, Capital Volume and Scores in Portuguese and Mathematics 

Dimension 1 explained 48.2% of the data variance, while Dimension 2 explained 34.1%. Dimension 1 shows that very low scores are associated with very low volume of capital and with the existence of at least one episode of the student failing a school year. There is also an association between high scores and high volume of capital, in addition to an opposition between these categories and possessing a very low volume of capital, obtaining very low scores and having already failed a school year.

Finally, we analyzed the relationship between social class and the skin color of the students. Figure 8 shows the results of a multiple correspondence analysis between self-declared skin color, the scores in the two disciplines evaluated by Saeb 2013 and the volume of economic and cultural capital.

Source: The Author / Saeb 2013

Figure 8:  Multiple Correspondence Analysis between Self-Declared Skin Color, Capital Volume and Scores in Portuguese and Mathematics 

Dimension 1, which explained 45.3% of the variation in the data, shows an association between high scores, high volume of capital and self-declared white skin. We can observe, in the right quadrants, an association between very low scores, very low volume of capital and self-declared black skin. The distances between these associated characteristics reveal an opposition between them, which may signal a link of skin color with social class and with educational achievement. Dimension 2 explains 34.3% of the variance, and shows a similar opposition between high volume of capital, high scores and self-declared white skin and low scores.

Among white students (34.1% of the sample), 10.9% have a very low volume of capital and 35.8% have a low volume - totaling 46.7%. Almost half of white students are among those with very low or low volume of economic and cultural capital. Whites with a high volume of capital are 20.6% of self-declared white students. On the other hand, among self-declared black students (11.3% of the sample), 19.1% and 40.0% possess very low and low volume of capital, respectively. Altogether, self-declared black students that have low or very low access to economic and cultural capital reach 59.1%. Black students who are in the high volume of capital bracket are only 11.2%. Data for brown students (48.7% of the sample) are similar to data for black students, as well as the data for indigenous students (2.7% of the sample).

Final Considerations

In this paper, we show that, in Brazil, social class appears almost always related to student performance in mathematics and Portuguese. Living in a large residence, having the habit of reading and parents with a high level of education are conditions that influence the students’ school proficiency. This demonstrates that circumstances external to the school are still very important to the life stories of students competing in the school system as a whole.

By examining the influence of social class on Brazilian students’ scores, as assessed by Saeb 2013, we show that the objective class dimension affects students’ achievements in basic education, revealing the relevance of this theme for current academic debates and educational policies. Showing the relationship between social classes and school performance does not mean that the educational problems are caused exclusively by poverty or the lack of resources on the part of students and their families, but it call attention to the fact that schooling alone has not yet been able nor is currently able to avoid reproducing social inequalities.

Studies on class inequalities and their relationship with formal education can and should continue to address the social conditions that generate and transmit the fundamental dispositions for an efficient incorporation of the logic of educational establishments. Investigating these issues requires carrying out a qualitative research agenda aimed at understanding the processes of subjectivity/morality that form the agents in our society and how these agents deal with the interactions inherent in the school environment.