Cultural Capital and Graduate Student Achievement:
A Preliminary Quantitative Investigation

Geoffrey Moss
[email protected]

Abstract

Bourdieu’s cultural capital thesis asserts that students from privileged socioeconomic backgrounds tend to acquire a background in high culture (i.e. cultural capital) that they invest in scholastic pursuits and thereby obtain returns in the form of academic achievement and degree attainment. Cultural capital research by Bourdieu and others has examined the attainment of undergraduate and graduate degrees, but has treated graduate level education as a “black box” (i.e., such research has confined the examination of academic achievement to the undergraduate level). This study made a preliminary quantitative attempt to test the bourdieuian thesis that students from higher socioeconomic backgrounds tend to have higher levels of graduate academic achievement because they enter graduate school with cultural capital that they acquired prior to college. A convenient sample of liberal arts students attending a socioeconomically diverse graduate school was utilized as the basis for conducting a graduate student survey (n=113). An analysis of the survey data found that after controlling for relevant variables, neither parental socioeconomic status nor cultural capital (operationalized as pre-college attendance at various high culture events) had a substantively meaningful or statistically significant relationship with graduate academic achievement (operationalized as graduate grade point average). These preliminary findings suggest that Bourdieu’s thesis needs to be amended; graduate students can become academically liberated from the cultural effects of their socioeconomic origins.

Key Words: Cultural Capital, Graduate Student Achievement, Bourdieu

Pierre Bourdieu’s (1979) cultural capital thesis posits that the “high culture” of a society’s dominant socioeconomic classes (i.e. its upper class and upper middle class) plays a major role with respect to the reproduction and legitimation of socioeconomic inequality. High culture is constituted by competitive attempts to attain social distinction and sublimated pleasure via activities (e.g., attendance at art exhibits, classical plays, and classical music concerts) that involve a high level of abstraction, intellectualism, and refinement (Bourdieu 1989). Schools generally require that students possess the capacity to receive and internalize an increasingly complex body of high culture, but this capacity is inadequately supplied by K-12 schools, and by institutions of higher learning (Bourdieu 1979). Instead, students who acquire this capacity generally do so as a result of a complex and largely unplanned acculturation process that occurs within their family of origin. Such students can “invest” their cultural preparation in their scholastic pursuits, and thereby obtain “returns” in the form of greater educational achievement and degree attainment. Bourdieu thus argues that high culture functions as a form of “cultural capital” (Bourdieu 1973).

Cultural capital, like other forms of capital, is unequally distributed within western capitalist societies (Bourdieu 1973). The amount of cultural capital that students “inherit” from their family of origin is a function of their socioeconomic status (Bourdieu, 1979). The inheritance of cultural capital, along with economic and social capital, enables members of the dominant classes to reproduce their socioeconomic position (Bourdieu 1997).

The educational advantages of inheriting economic capital (e.g., being able to afford a private school or college) and social capital (e.g., admission to an elite college via family connections) are easy to recognize. The educational advantages of inheriting cultural capital (i.e., better cultural preparation for school), however, are relatively difficult to perceive. The educational effects of such advantages (e.g., higher grades) are generally attributed to individual merit (i.e., superior innate ability and/or greater individual effort). Cultural capital thus functions to support the legitimating belief that educational competition is meritocratic (Bourdieu 1979).

As access to college (and college graduation) becomes more commonplace, students from high socioeconomic status families tend to increase their investments in education (i.e. graduate school) to stay ahead of the newcomers. Between 1960 and 1997, the number of graduate degrees more than quadrupled. By 1997, one out of three postsecondary degree recipients were graduate students. (Mullen, Goyette, and Soares 2003). Graduate degrees increasingly function to exclude those whose highest degree was the Baccalaureate from high paying or prestigious occupations (Mullen, Goyette, and Soares 2003; Kingston and Clawson 1990; Cappell and Pipkin 1990).

Graduate education has become the primary educational terrain with respect to elite (i.e., upper class and upper middle class) and middle class occupational preparation. Research is needed in order to determine the extent to which such preparation is constituted by educational inequities based on socioeconomic status. Such research could attempt to debunk the myth of meritocracy with respect to graduate level education, and establish the proposition that there is a need for educational policies and programs designed to promote socioeconomic equity in the context of graduate level education.

Bourdieu recognized that graduate education is playing an increasingly significant role with respect to the reproduction of socioeconomic inequality (Bourdieu and Boltanski 1978). Cultural capital research by Bourdieu and others, however, has examined both undergraduate and graduate degree attainment,1 but has treated graduate level education as a “black box.” Such research, that is, has confined the study of academic achievement to the undergraduate level. Thus, we do not know the extent to which Bourdieu’s thesis (that socioeconomically privileged students tend to have higher levels of achievement because they have more cultural capital) applies at the graduate level.

Bourdieu’s cultural capital thesis implies that the educational effects of socioeconomic status and cultural capital do persist at the graduate level. Bourdieu’s thesis, that is, is constituted by the assertion that K-12 and undergraduate institutions fail to equalize the level of cultural capital possessed by students from unequal socioeconomic backgrounds. To the extent that this is the case, then students from lower socioeconomic origins who advance beyond the baccalaureate would tend to start their graduate level studies at a cultural disadvantage.

This disadvantage could be particularly strong with respect to graduate liberal arts education. Although the cultural capital emphasized by Bourdieu (i.e. high culture) may provide a useful background with respect to various professional specialties (e.g. Law, Education), such capital is tied most directly to the liberal arts disciplines. Those who study these disciplines beyond the undergraduate level are generally required to achieve a level of disciplinary competence that is relatively strong; those who lack a high level of cultural capital may have a particularly hard time achieving such competence.

In this preliminary study, a convenient sample of liberal arts students attending a socio-economically diverse graduate school were surveyed (n=113) to determine the extent to which Bourdieu’s cultural capital thesis can account for differences in graduate level academic achievement. To set the theoretical context for this study, this article discusses Bourdieu’s concept of cultural capital in light of related bourdieuian concepts. This article then proceeds to discuss the relevant empirical literature, present the research hypotheses, explain the methodology of the study, and present an analysis of the graduate student survey data. In its concluding discussion, this article discusses the study’s main findings, notes the study’s major methodological limitations, and suggests directions for further research.

Cultural Capital And Related Bourdieuian Concepts

Bourdieu’s concept of cultural capital has generated more research than any of his other theoretical concepts (Lamont and Lareau, 1988). Bourdieu’s overall theoretical approach to the study of social reproduction is also constituted by the related theoretical concepts of field, practice, habitus, and symbolic violence. A field is a hierarchically structured sphere of interaction where players engage in practice. While engaged in practice, players compete for resources, status, and position. Such practice is performed in accordance with an established economy of practice (i.e. the extent to which various types of cultural resources count as capital within a given field). Those who dominate a given field possess the symbolic power to change the economy of practice within that field (Bourdieu and Wacquant 1992).

To Bourdieu, the ability of an individual to succeed in and/or dominate a given field is a product of the economy of practice within that field, the individual’s capital (resources), and the individuals’ “habitus” (i.e. an individuals’ disposition to use those resources). The habitus is the mechanism through which the individual estimates his/her level of economic, social, and cultural capital as well as his/her prospects for using such capital to obtain success within various fields. The habitus is constituted by a continuous dialectical interplay between creative estimation (e.g., “If I go to college, I could probably graduate if I study really hard.”) and internalized class based dispositions (e.g., “College is not for the likes of us.”) (Bourdieu and Passeron 1977). Thus, the habitus is constituted by individual agency as well as internalized experiences of structural inequality.

Bourdieu (1989) demonstrated that the high culture that constitutes cultural capital is a product of class position; those who have substantial if not total freedom from economic pressure (i.e., the upper class), or who do work involving a substantial degree of abstract symbolic manipulation (i.e. most members of the upper middle class, and many members of the lower middle class) tend to enjoy and appreciate refined cultural products that are relatively removed from immediate material concerns (i.e. the artistic and intellectual products that constitute high culture). Those burdened with greater material pressures and work that is relatively concrete (e.g., the working class), on the other hand, tend to prefer cultural products that are directly tied to concrete practical activity (i.e. “lowbrow” culture; e.g. a realistic photograph of a car or boat). They tend to view the products of high culture as meaningless and impractical (e.g., “I don’t like abstract art…It doesn’t relate to the real world.”).

The field of Education in general, and higher education in particular has been dominated by those who favor high culture (i.e. middle and upper class professors, teachers, and administrators). As a result, this field is constituted by an economy of practice that favors high culture; the ability to enjoy and appreciate lowbrow culture is not normally constituted as cultural capital within Higher Education (Bourdieu, 1979).2 Students socialized in family environments that reject high culture (i.e. working class students) thus enter college with a relatively low level of cultural capital.

From a bourdieuian perspective, high culture is not inherently superior; it is constructed as superior within various fields through the use of symbolic power. The dominant classes institutionalize their high culture within such fields via cultural and educational practices that present themselves as universal and neutral. As a result, the curricular, linguistic, and pedagogical biases of educational institutions tend to be misrecognized by the subordinate classes and are thus viewed as inherently legitimate rather than class-based and arbitrary (Bourdieu and Passeron 1977).

The legitimation of high culture within educational institutions functions to sustain internalized class based dispositions that are focused primarily on class reproduction (i.e. dispositions to succeed within the limits of class structures) rather than class struggle (i.e. dispositions to reduce or eliminate class inequality). In western societies, such legitimation tends to replace physical violence as the primary mechanism of working class disempowerment (Bourdieu and Passeron 1977). To Bourdieu, the use of legitimating practice to disempower the working class is akin to physical violence; it constitutes violence in symbolic form (i.e. what Bourdieu termed Symbolic Violence).”

Relevant Research

Bourdieu appreciated that cultural capital is not constituted solely by high culture. He pointed out, for example, that students inherit cultural capital in the form of critical thinking skills, writing skills, linguistic skills, and scientific skills (Bourdieu, 1997). Bourdieu’s work on cultural capital and educational success, however, emphasized unequal exposure to high culture (Lamont and Lareau 1988).

The best developed, and most influential line of research concerning Bourdieu’s cultural capital thesis echoes Bourdieu’s own focus, and examines cultural capital that is grounded in high culture (see Kingston 2001; see also Lamont and Lareau 1988). A growing body of research, however, has examined cultural capital with respect to other types of educational resources. Roscigno and Ainsworth-Darnell (1999) and Teachman (1987), for example, have defined cultural capital in terms of household educational resources (e.g. books or computers).

Lamont and Laureau (1988) have argued that if capital research is to remain conceptually distinct, than cultural capital research should maintain its focus on the educational effects of high culture. Kingston (2001:8) has likewise argued that cultural capital should not be defined as a broad concept that can encompass the entire realm of learning resources. He points out that the use of the terms “culture,” “cultural capital,” or “cultural resources” “inevitably introduces constructionist connotations of arbitrariness.” In his view, a broader conceptualization of cultural capital would place those not socialized into an arbitrarily defined high culture in the same umbrella category as those who lack resources which are more objectively tied to success in modern society (e.g., study space, books in the home, use of a computer) (Kingston 2001).

A substantial body of empirical research has dedicated itself to testing Bourdieu’s cultural capital thesis. For the most part, such research has operationalized cultural capital using responses to survey questions (available within national data sets) that measure parental and/or student participation (i.e., attendance vs. non attendance) with respect to various types of high culture events. Such research has attempted to support Bourdieu’s proposition that student cultural capital can account for the positive effects of socioeconomic status on educational success (e.g., grades, degree attainment). In addition, research has attempted to support his proposition that parental cultural capital can account for differences in student cultural capital, and thereby predict educational success.

Most cultural capital research has focused on the former proposition (i.e. concerning the educational effects of unequal socioeconomic status and student cultural capital). Such studies have found that high socioeconomic status as well as student participation in high culture activities has a significant positive relationship with educational outcomes (i.e., grades and/or degree attainment) in the United States (Dumais 2002; DiMaggio and Mohr; 1985; DiMaggio 1982), Brazil (Amaral 1991), Czechoslovakia and Hungry (Boguszak, Mateju, and Peschar 1990), and Sweden (Roe 1983). Studies focused on the later proposition (concerning the effects of parental cultural capital) have found that parental participation in high culture activities has a significant but relatively modest relationship with student cultural capital and with educational outcomes in the United States (Aschaffenburg and Mass 1997; Kalmijn and Kraaykamp 1996), the Netherlands (De Graff 1989), and Germany (De Graf 1988).

Previous research on cultural capital and educational success (e.g., Kalmijn and Kraaykamp 1996) has often been constituted by measures of graduate degree attainment (i.e. via variables such as “years of schooling completed” or “highest degree earned”). It has not, as previously noted, tested Bourdieu’s cultural capital thesis with respect to graduate level academic achievement (e.g., graduate grades). In part, this may be due to the failure of national surveys that include measures of socioeconomic status and cultural capital to include indicators of graduate level achievement. Existing studies of graduate grades, moreover, have been performed solely to determine the predictive validity of the Graduate Record Exam (see, for example, House 1998); they have not been constituted by measures of cultural capital or socioeconomic status.

The non-existence of graduate achievement measures within relevant national surveys, however, does not explain why previous cultural capital researchers have not undertaken preliminary small-scale studies of graduate level achievement. A possible reason for this neglect is that they were discouraged from doing so by previous research. Until very recently, research by Mare (1980) and Stolzenberg (1994) had concluded that college graduation tends to have an equalizing effect on future academic outcomes. Mare (1980), using data from a 1964 survey of U.S. military veterans, found that the social class background had virtually no effect on the decisions of college graduates to enter graduate school. Stolzenberg (1994) replicated this finding using data from the high school class of 1972, and asserted that college graduates tend to become academically liberated from their socioeconomic origins.

Mullen, Goyette, and Soares (2003), however, challenged Stolzenberg’s “status liberation thesis” using relatively recent data (i.e. the 1992-93 cohort from the U.S. Department of Education dataset, Baccalaureate and Beyond). They found that for college graduates, the graduate entry gap with respect to parents’ educational level is significant but modest for those who enter masters programs (18 percent for those whose parents had a High School degree or less verses 22 percent for those whose families attended graduate school). They found that the gap is much wider for first professional (2.2 percent verses 7.1 percent) and doctoral programs (1.4 percent verses 5.1 percent) (Mullen, Goyette, and Soares 2003: 150).

Mullen, Goyette, and Soares (2003) findings demonstrate that college graduates, at least in terms of graduate school entry, continue to be affected by their socioeconomic background. The logical next step for empirical research is to see if the effects of socioeconomic background continue at the graduate level. Anecdotal evidence (i.e. autobiographical essays) collected by Ryan and Sackrey (1996) suggest that these effects may be substantial; liberal arts faculty members from working class families reported that while in graduate school, they felt like cultural outsiders, and had trouble adjusting to the academic demands of graduate level work.

Research Hypotheses

Following Bourdieu, this study proposes the thesis that students from higher socioeconomic backgrounds tend to get higher graduate grades because they continue to be positively affected by cultural capital that they acquired prior to college. To subject this thesis to a preliminary quantitative test, this study makes two primary hypotheses regarding the relationship between socioeconomic status, cultural capital, and graduate academic achievement among liberal arts graduate students:

Hypothesis 1: Parental socioeconomic status (parents education, occupational status, and income) has a positive effect on graduate academic achievement (operationalized as graduate grade point average), and

Hypothesis 2: The positive effect of parental socioeconomic status on graduate academic achievement occurs largely because socioeconomic status has a positive effect on cultural capital (operationalized as pre-college cultural participation).

Methodology

Sample

This study utilized a convenient sample of students enrolled in a socio-economically diverse urban graduate school located in the eastern United States. Students were asked to fill out an IRB approved Graduate Student Questionnaire while attending graduate classes in the liberal arts. 115 questionnaires were collected from students in seven graduate classes in seven different liberal arts disciplines (Art History, Anthropology, Comparative Literature, English, History, Philosophy, and Sociology). The response rate was 100% (no student refused to fill out the survey), but two questionnaires were omitted because the respondents failed to complete more than half of the survey questions. The resulting sample thus consisted of 113 graduate students.

Characteristics of the Respondents

Table 1 shows the characteristics of the respondents with respect to each variable in the analysis. Graduate grade point average (GPA), the dependent variable for this study, had a mean of 3.6. Since graduate grades typically range from A (4.0) through B (3.0), it is not surprising that the standard deviation of the dependent variable is relatively small (.32). The dependent variable, however, is constituted by a level of variation that is sufficient for meaningful statistical analysis (e.g. at the graduate level, a B represents a lower level of academic achievement than an A).

The mean undergraduate grade point average (UGPA) for the sample was 3.39 (s.d.=.35), and the mean combined Graduate Record Exam score (GRE Verbal, Quantitative, and Analytical) was 1920 (s.d.=151). The mean number of graduate credits earned by respondents was 29.07 (s.d.=21.57). 52 (46%) were humanities majors (Art History, Comparative Literature, English, History, or Philosophy), and 53 (47%) were Social Science majors (Anthropology or Sociology).

The mean age was 30.5 (s.d.=7.36). 46 (52%) were female, 48 (42.5%) were male, and 2 (1.8%) did not indicate their gender. 103 (91.2%) reported their racial or ethnic identification as white, 4 (3.5%) as multiracial, 1 (.9%) as asian, 1 (.9%) as hispanic, and four (3.5%) did not indicate their racial status.

Measures

The dependent variable (GPA) was measured by asking respondents to indicate their graduate grade point average. The control variables for this study (i.e. UGPA, GRE, Major, Age, Sex, and Race) were also measured via the use of self-report indicators. Two additional sets of self-report indicators were utilized to measure the two main independent variables of the study (i.e., parental socioeconomic status and cultural participation).

Parental socioeconomic status (SES) is a composite variable that was computed by summing the combined parental scores (mother plus father) for each of three socioeconomic indicators: mean highest educational attainment (coded as 9=doctorate through 1=some high school or less), mean income (coded as 9=over 200,000 through 1=under $25,000), and mean occupational status (coded as 9=academic or professional, 7=managerial, 5=Small Business Owner, 3=Clerical, and 1=Blue Collar). Table one indicates that the respondents exhibit a wide range of variation with respect to each of these indicators. Tests for skewness and kurtosis indicated that the sample distribution for the composite variable (SES) is normally distributed. Socioeconomic status had a mean of 32.96 and a standard deviation of 6.26.

To measure cultural capital, indicators representing respondents’ pre-college cultural participation were combined. Respondents indicated whether or not they remember attending each of the following prior to attending college: 1) an art exhibit, 2) a classical music concert, 3) a ballet, 4) a classical play, and 5) an opera. This scale is similar to scales in use within studies based on national datasets, and could enable a comparison with such studies.

Factor analysis utilizing Principle Components as the extraction method and Oblimin with Kaiser Normalization as the rotation method was performed utilizing the preceding five items. The analysis yielded one component with an eigenvalue greater than one. The factor loadings for the items were as follows: .716 (art exhibit), .736 (classical music concert), .652 (Ballet), .710 (classical play), and .662 (opera). A cultural participation scale utilizing these items was constructed in a straightforward and readily interpretable manner (i.e. by summing the number of affirmative responses to each of the five items).

The resulting zero-five scale (respondents could indicate attendance at between zero and five pre-college cultural events) had a reliability (Alpha) of .732, a mean of 1.55 (number of events), and a S.D. of 1.55. Tests for skewness and kurtosis indicated that the sample distribution for the scale is somewhat outside the bounds of normality (skewness=.818, Kurtosis=-.374). Attempts to adjust for non-normality using the Curve Estimation Procedure provided by SPSS,3 however, did not yield a statistically significant or substantively meaningful improvement in the ability of the scale to predict GPA. The analyses reported in this study will thus use the relatively straightforward and readily interpretable procedure of linear estimation.

The variables constituting the cultural participation scale imperfectly represent an underlying continuous progression from zero to high levels of pre-college cultural participation. This violates the classical OLS assumption that measures be continuous and at the ratio level. OLS regression, however, generally remains robust even with such imperfections. The option of transforming the cultural participation scale into a categorical variable (e.g., categorize people as high or low based on the distribution for the scale) was considered, but rejected on the grounds that the use of a scale might enable this study to capture greater underlying variation in cultural participation.

Prior to computing the composite variables for parental socioeconomic status and cultural participation, EM Estimation was used to impute missing data for all cultural participation and socioeconomic variables.4 EM Estimation was also used to impute missing data for the other variables in the data set prior to performing the analyses shown in Table 2 and Table 3. In table 1, only the figures for the composite variables are based on EM Estimation. The figures shown in Table 1 thus reveal the full characteristics of the sample (including missing data) with respect to the variables included in the questionnaire.

EM Estimation is a form of casewise maximum likelihood estimation that uses the Jamshidian-Bentler Algorithm followed by the Yuan-Bentler corrections for possible non-normality. Bentler and Wu (1995) have demonstrated that EM Estimation is the best available method for handling missing data. Listwise Deletion and Pairwise Deletion produce greater bias because they are based on the generally faulty assumption that the pattern of missing values does not depend on the data values (i.e. all respondents are equally likely to provide missing data for all items). Mean Imputation, moreover, is based on the generally erroneous assumption that the missing data patterns of respondents who omitted data tend to be similar to that of the typical (i.e. average) respondent who did not omit data (e.g., the assumption that those who do not report their incomes have incomes equal to the sample mean). For a full description of the EM Estimation method, see Jamshidian and Bentler (1999) and Yuan and Bentler (2000).5

Analyses

The analyses of this study tested the ability of Bourdieu’s cultural capital thesis to account for differences in graduate academic achievement. This thesis implies that a) parental socioeconomic status positively affects graduate academic achievement (i.e., graduate students from families higher in SES tend to have higher levels of graduate academic achievement), and b) parental socioeconomic status affects graduate academic achievement largely because parental socioeconomic status has a positive effect on cultural capital acquired prior to college (i.e., graduate students from families higher in SES tend to have higher levels of graduate level academic achievement because their socioeconomic background enabled them to acquire more cultural capital prior to college). After a preliminary analysis at the zero order (correlation) level, a Two-Stage Ordinary Least Squares (OLS) Regression Analysis was performed in order to control for relevant variables, and reduce the chances of making causal inferences based on spurious relationships or suppression effects. In both stages (Model 1 and Model 2), Graduate Grade Point Average (GPA) was the dependent variable.

Model 1 included the independent variable for parental socioeconomic status (SES), and control variables for demographic characteristics (Age, Race, and Gender), graduate academic status (number of graduate credits earned [Credits], Major [dummy coded as 1 for graduate humanities major, 0 for graduate social science major], undergraduate grade point average (UGPA), and total Graduate Record Exam score (GRE Verbal plus Quantitative plus Analytic). Model 1 enabled a test of Hypothesis 1- that socioeconomic status has a positive effect on graduate academic achievement (operationalized as graduate grade point average). In Model 2, the cultural participation scale (CULTPAR) was added as an independent variable. The addition of the cultural participation scale (an indicator of cultural capital) to model 2 enabled a test of Hypothesis 2 (parental socioeconomic status affects graduate level academic achievement largely via its effect on cultural capital). If both hypotheses are correct, than SES should have a positive relationship with GPA in model 1, and CULTPAR should have a positive relationship with GPA in model 2. The positive effect of SES on GPA, moreover, should be much lower in model 2 than in model 1 (i.e. if hypothesis 3 is correct, than that portion of the effect of SES on GPA that is due to shared variance between SES and CULTPAR should go down once CULTPAR is included in the model).

For both equations (Model 1 and Model 2), unstandardized beta coefficients are examined in order to determine the substantive effects of the included variables. Standardized beta coefficients are examined in order to determine the relative impact of these variables. Part correlations for the main independent variables (parental socioeconomic status and cultural participation) are utilized to compare the unique effects of these variables on the dependent variable (grade point average). Results within each model that are statistically significant (using a two tailed test) are denoted with a * (p