Presiding Officer Training module 2024 lok sabha elections
Geert Driessen (2001) IRE Ethnicity, forms of capital, and educational achievement
1. ETHNICITY, FORMS OF CAPITAL, AND EDUCATIONAL
ACHIEVEMENT
GEERT W. J. M. DRIESSEN
Abstract – Bourdieu’s cultural capital thesis is an attempt to explain how social class
influences the transmission of educational inequality. In this article, the question of
the extent to which various forms of capital also apply to ethnic minorities stands
central. On the basis of Dutch and American research findings, a model is formu-
lated and empirically tested with the aid of data from the Dutch Primary Education
cohort study. Students from four ethnic groups are included: Dutch, Surinamese,
Turkish, and Moroccan. The main variables are language and math test scores, socio-
economic milieu, and a number of capital indicators, including financial resources,
linguistic resources, parental reading behavior, and educational resources within the
family. The results show no mediating effect of resources within the various ethnic
groups. The findings also suggest that in research and practice it is relevant to not
treat ethnic groups as one homogenous group, but to differentiate between the various
groups.
Zusammenfassung – Bourdieus These kulturellen Kapitals ist ein Erklärungsversuch,
wie soziale Klassen die Übertragung von Ungleichheiten im Bildungsbereich beein-
flussen. In diesem Artikel wird untersucht, inwieweit die Theorie auch auf ethnische
Minderheiten angewandt werden kann. Auf der Basis niederländischer und amerikani-
scher Forschungsergebnisse wird ein Modell formuliert und empirisch getestet, das
Studenten aus vier ethnischen Gruppen einbezieht: holländischen, surinamischen,
türkischen und marokkanischen. Die Hauptvariablen sind Testergebnisse in
Sprache und Mathematik, im sozialwirtschaftlichen Milieu und eine Anzahl von
Kapitalindikatoren, einschließlich finanzieller Mittel, elterlichen Leseverhaltens und
Bildungsressourcen innerhalb der Familie. Die Ergebnisse zeigen, dass es ernsthafte
Schwierigkeiten bei dem Versuch der Anwendung der These Bourdieus auf ethnische
Minderheiten gibt.
Résumé – La thèse de Bourdieu sur le capital culturel tente d’expliquer la façon dont
la classe sociale contribue à maintenir l’inégalité éducative. L’auteur de l’article
examine dans quelle mesure cette théorie peut être étendue aux minorités ethniques.
Il élabore et teste empiriquement un modèle à partir de résultats scientifiques danois
et étasuniens. Ce test est administré auprès d’écoliers issus de quatre groupes
ethniques: danois, surinamiens, turques et marocains. Les variables principales
constituent les résultats aux tests linguistiques et mathématiques, le milieu socio-
économique et une série d’indicateurs décisifs dont les ressources financières, le
comportement des parents envers la lecture et le niveau éducatif de la famille. Les
résultats montrent cependant qu’il est très difficile d’appliquer la thèse de Bourdieu
aux minorités ethniques.
Resumen – La tesis del capital cultural, de Bourdieu, apunta a explicar cómo la clase
social influye sobre la transmisión de desigualdades en el nivel de instrucción de las
personas. Este artículo examina hasta que grado esta teoría también podría ser aplicada
a las minorías étnicas. Sobre la base de los resultados obtenidos con investigaciones
International Review of Education – Internationale Zeitschrift für Erziehungswissenschaft
– Revue Internationale de l’Education 47(6): 513–538, 2001.
2001 Kluwer Academic Publishers. Printed in the Netherlands.
2. holandesas y estadounidenses se está delineando un modelo sometido a un test
empírico que incluye a estudiantes de cuatro grupos étnicos: holandeses, surinameses,
turcos y marroquíes. Las principales variables son la lengua y las puntuaciones
obtenidas en pruebas de matemáticas, el entorno socioeconómico y un número de
indicatores importantes tales como recursos financieros, hábitos de lectura de los
padres y niveles de instrucción dentro de la familia. Los resultados demuestran que
se presentan serias dificultades al tratar de aplicar la tesis de Bourdieu a las minorías
étnicas.
Educational position of ethnic minorities
Just as in most West-European countries, various groups of migrants have
come to the Netherlands since World War II for mainly political and economic
reasons. These migrants can be divided into four categories: (1) Migrants from
former Dutch colonies, including the Dutch East Indies, Surinam, and the
Netherlands Antilles. As a result of their ties with the former motherland,
these migrants were already somewhat acquainted with the Dutch language
and culture. (2) So-called guest workers from such Mediterranean countries
as Spain, Turkey, and Morocco. One characteristic shared by them was their
low level of education. (3) Refugees coming from countries such as Iran, Iraq,
former Yugoslavia, and Somalia. This is a very divergent category in terms
of both language and culture. (4) Migrants from such Western countries as
Belgium and Germany with a comparable socioeconomic status.
Based on the broad criterion of “origin”, i.e., the native countries of the
migrants, their parents, and their grandparents, more than 16% of the Dutch
population of approximately 15.5 million inhabitants were non-indigenous in
1996. Of these 15.5 million, 7% were the target of the ethnic minority policy
of the Dutch government. Due to their relatively high socioeconomic status,
migrants from the Dutch East Indies are excluded from this policy, as are
514
3. migrants from Western countries. In terms of numbers, the main countries of
origin are: Surinam (282,000), the Antilles (94,000), the Moluccas (35,000),
Turkey (272,000), Morocco (225,000), the other Mediterranean countries
(164,000), and refugees (44,000) (Driessen 2000).
In the Western world, the position of ethnic minorities in education is gen-
erally a major cause of concern. Their position is characterized by low
achievement levels, little pursuit of higher types of education, widespread
truancy and disciplinary problems, and high drop-out rates (Gillborn 1997;
Rossi and Montgomery 1994). During the past few decades, numerous theories
have been developed to explain the differences observed in school careers.
Since the 1970s, a number of theories have claimed that existing social
inequalities tend to be reproduced independent of intelligence or capacity. The
educational system is assumed to occupy a key position in this process.
One of the proponents of this reproduction thesis is Bourdieu with his
culture-political theory. Central for Bourdieu is the cultural transmission of
social inequality. In education, selection occurs on the basis of a number of
arbitrary criteria – namely, differences in cultural patterns and lifestyle
characteristics. These patterns and characteristics or so-called “cultural capital”
or “cultural resources” are transmitted within one’s own milieu from
generation to generation. One can think of critical linguistic or social-cultural
competencies and affinities (e.g., interactional forms, language use, interests,
taste for art and culture, social and cultural opinions and preferences). This
habitus differs from one status group to another.
Cultural capital cannot be acquired at school and the school cannot diminish
differences in the amount of capital across groups of students. The converse
is, in fact, the case: The manner in which education is organized, the one-
sided interest of the educational system in the culture of the dominant status
groups, the fit of the existing educational system with this culture, and the
presupposition of particular codes and capacities all contribute to an expan-
sion of already existing differences. The cultural capital that education pre-
supposes on the part of its students is largely acquired during primary
socialization within the family and upbringing by parents who belong to the
dominant culture. The stronger the embedding of parents in the dominant
culture and the more oriented their child rearing is toward the dominant
culture, the more comfortable their children feel at school and the more they
benefit from that which is being taught. A gap can thus exist between home
and school as a result of a student’s access to cultural capital and belonging
to a dominant status group or not. Given that the dominant culture lies at the
core of the – hidden – educational curriculum, it is often difficult for the
children from the lower social strata to grasp the material being taught. And
this situation explains the less favorable educational position of such groups
(Bourdieu and Passeron 1977).
515
4. Research on cultural capital
Research in the Netherlands
In the Netherlands, the ideas of Bourdieu have been empirically tested by a
few researchers. De Graaf (1986) analyzed not only the influence of cultural
resources but also of financial resources (i.e., income levels). Although the
results of his study are not completely consistent, the effects of financial
resources appeared to be modest, while the effects of parental reading behavior
and passive cultural participation on the part of the student were found to
mediate the entire effect of social background on the educational levels
achieved by the students. Tesser and Mulder (1990) attempted to replicate
the study of De Graaf but could not confirm any of the findings. They did,
however, find that special significance should be attached to the “possession
of books.” Such possession indicates a specific relation to language and
perhaps a form of participation in the “culture of literacy” (cf. Carrington
and Luke 1977). Van der Velden (1991) also found the possession of cultural
capital (including reading behavior, cultural-pedagogical level of upbringing,
insight into the educational system) to almost completely explain the effects
of the educational and vocational levels of the parents on the school careers
of their children. Nevertheless, while the relevant characteristics of the family
were found to mediate part of the status effect, they did not contribute to a
better prediction of career course. The characteristics indicate, rather, which
processes appear to play a role in the observed effects.
Driessen, Mulder and Jungbluth (1999) examined a number of aspects of
cultural capital (book culture, cultural participation, parental educational
support behavior, and language choice) among an extensive sample of high
school students and came to the same conclusions as Van der Velden: The
influence of the family structural indicators (social milieu) occurs via family
pedagogical indicators (cultural capital) although the latter characteristics add
virtually nothing to the explanatory power of the first characteristics. Leseman
et al. (1995) examined the effects of family characteristics on the achieve-
ment of children in the first year of elementary school. Their analyses clearly
showed the effects of socioethnic milieu on the vocabularies of the children
to be completely mediated by the language in the home, the instructional
quality of the parent-child interactions, and the degree of social-emotional
support within this context. In addition to the language at home, the peda-
gogical conceptions and cultural-educational capital (“lifestyle characteris-
tics”) within the family appeared to constitute important links between ethnic
origin and the socialization of cognitive skills.
The research conducted during the past few decades shows not only the
educational position of ethnic minorities to be a source of concern but also
that this position may very well be related to the culturally determined values
and patterns of child rearing found within the families themselves. Among the
Dutch studies in the domain of cultural capital only a few have considered
516
5. ethnic differences. Given their current position, however, it is relevant to
evaluate the cultural-capital thesis for them as well. And in order to do this,
it may be useful to first examine the results of the cultural-capital thesis in
connection with minorities in countries with an longer tradition of migration
than in the Netherlands. A review of the literature was therefore undertaken
for research conducted within this domain in the USA.
Research in the USA
The US studies represent a broad scale of themes within diverse settings.
And although we encountered relevant initiatives and new points of view
within this research, the general utility of the research was less than we had
hoped for. This is due to not only the limited amount of specific attention paid
to “race” or “ethnicity” but also the inconsistency of the results. Most recently,
Roscigno and Ainsworth-Darnell (1999: 160) also had to conclude that “sur-
prisingly little research has focused on if and how racial and ethnic groups
differ with regard to these attributes in general, and cultural capital, in par-
ticular.”
If we examine the theoretical perspective adopted in the majority of the
studies, it can be seen that the American studies are much more oriented
toward the concept of “social capital” (Coleman 1988; Runyan et al. 1998)
when compared to the Dutch (and West-European) studies. Coleman’s thesis
of social capital pertains to the structure of the social relations between people.
Being part of a community in which educational achievement is stimulated
can itself be considered a form of capital. The classical interpretation of
cultural capital in terms of Bourdieu (i.e., such “high-brow cultural partici-
pation” as visiting the theater, concerts, and museums) is relatively infrequent
in the American studies (cf. Aschaffenburg and Maas 1997). It is striking
that cultural/social capital receives a very specific but varying definition: reli-
gious capital as the binding factor within a community (Corvig 1996), the
cultural subgroup (Sullivan 1997), or belonging to a particular ethnic group
as valuable in and of itself (Bankston, Caldas and Zhou 1997).
As already indicated, the results of the American studies are not at all con-
sistent. In the one case, either cultural or social capital is found to play a medi-
ating role; in the other, they do not. In this light, it should be noted that some
of the studies confirm the mobility thesis (namely, cultural capital as a means
of mobility for lower-class children, cf. DiMaggio 1982) but not Bourdieu’s
reproduction thesis (which is aimed at the higher social classes). With regard
to the central aim of our search, namely the significance of cultural capital
for ethnic minority groups, the findings are also disappointing. The expecta-
tion was that a country with a much longer tradition of ethnic minorities than
in the Netherlands and a country where the size and scope of the groups is
much greater than in the Netherlands would pay greater attention to such a
perspective. This did not prove to be the case. In research on cultural capital,
no explicit distinction was generally made with regard to ethnicity or the
517
6. sample was simply limited to a group of whites. The majority of the studies
cited here thus constitute an exception, but this does not mean that ethnicity
forms a central theme or explicit attention was even paid to ethnicity.
Despite this situation, we nevertheless encountered a few interesting
notions. Lareau and McNamara Horvat (1999), for example, make a distinc-
tion between the possession and activation of capital and thereby suggest
that capital only gains its value in specific settings. They point, in particular,
to the importance of student-teacher interactions and to the role that the teacher
plays within these as so-called gatekeeper (Farkas et al. 1990). “Linguistic
capital” is also shown to be of major importance in a number of studies
(Stanton-Salazar 1997), which is really not so amazing for a country with such
a wide variety of language groups. Yet another distinct approach is that of
Ainsworth-Darnell and Downey (1998) who connect cultural capital to Ogbu’s
“oppositional culture explanation” and “resistance model” (also see Kalmijn
and Kraaykamp 1996). Further attention is paid in the American studies to
the educational support activities of the parents, such as providing help with
homework and attending parent evenings or conferences (Furstenberg and
Hughes 1995). Particularly striking is the number of occasions on which
family composition is referred to as an indicator of capital (i.e., the number
of children and a single-parent versus complete family structure; Bianchi and
Robinson 1997; Downey 1995).
The manner in which ethnicity is incorporated into the research studies in
general is twofold. On the one hand, the sample may be limited to a specific
minority or immigrant group (e.g., Roscigno and Ainsworth-Darnell 1999);
on the other hand, ethnic origin may be coded in the form of dummy vari-
ables for inclusion in the analyses (cf. Teachman, Paasch and Carver 1996;
White and Kaufman 1997). The alternative of conducting separate analyses
for each of the ethnic groups and comparing the results was only encoun-
tered once, namely in the study of Okagaki and Frensch (1998). They
concluded that there are big differences among the ethnic groups in the
correlations between attitudes and behavior and educational achievement, even
after controlling for social milieu and income. Their findings suggest that
conducting separate analyses is essential.
Research questions
On the basis of the research findings, we will examine the relation between
various forms of capital and educational achievement in greater detail for a
number of different ethnic groups using recent empirical data. In doing this,
we will attempt to answer the following research questions:
• To what extent do different ethnic minority groups have access to cultural
and financial resources? To what extent does this access differ from the
access of Dutch students to such resources?
• What relations exist between the different aspects of cultural and financial
518
7. capital, on the one hand, and the educational achievements of elementary
school children, on the other hand? Does a difference exist between ethnic
minority and Dutch students in this regard?
• To what extent does the effect of social milieu on achievement appear to
be mediated by cultural and financial resources? Does a difference exist
between ethnic minority and Dutch students in this regard?
Method
Sample, instruments, and variables
For the analyses in this study, we used data collected in the school year
1994/95 for the Dutch nationally representative cohort study of Primary
Education (“PRIMA”). Of particular interest are the results of a language
and math test that the students were administered in second grade (7–8 years
of age) and information with regard to the students and their home situation
collected via the completion of a written questionnaire by the parents (see
Driessen and Vierke 1999).
A central variable in our study is the ethnic origin of the students. We
focused on (1) Dutch, (2) Surinamese and Antillean (former colonies), (3)
Turkish (guest workers), and (4) Moroccan (guest workers) families with 7531,
282, 515, and 415 students, respectively. In the present study, we were in
search of any differences between the ethnic groups suggesting a systematic
effect. For this reason, we utilized analyses in which the size of the groups
was held equal. In order to do this, the four groups were weighted with the
smallest group, the Surinamese students (N = 282), receiving a weighting of
1 and the remaining groups receiving a weighting of 282/the number of
students in the particular group. In such a manner, a total weighted N of
4 × 282 = 1128 students was obtained.
The review of the literature revealed a number of features that could
possibly be of particular importance, and the PRIMA files contained infor-
mation on a number of these features. With regard to capital, financial and
cultural resources could first be distinguished. Thereafter, the cultural
resources could be divided into linguistic resources, reading behavior, and
pedagogical family climate. We assumed that such “capital” characteristics
would play a mediating role in the explanation of language and math achieve-
ment in terms of social milieu. It was also assumed that differences in these
relations would be found across the different ethnic groups. In Figure 1, the
proposed relations between the variables are presented in the form of a basic
model. In the following, an overview of the research instruments and the
operationalization of the variables will be presented.
Tests. Two tests developed specifically for PRIMA by the Dutch National
Institute for Educational Measurement were administered to the students. The
519
8. language test consisted of 60 items and three types of problems: morpho-
logical, syntactic, and semantic. The math test consisted of 40 problems and
covered a number of math skills, including addition, ordering, structuring, and
measurement. The results of the tests are expressed as the percentage correct
answers.
Parent questionnaire. The questions addressed the socioethnic milieu and
various cultural, linguistic, and child-rearing aspects of the family situation.
Ethnic group. On the basis of the parents’ homeland we distinguished
(1) Dutch, (2) Surinamese and Antillean, (3) Turkish, and (4) Moroccan
families.
Socioeconomic milieu. To characterize the socioeconomic milieu mother’s
education, father’s education, and father’s occupation were used. Education
could range from (1) “maximum of elementary education” to (7) “university
education”, while occupation could vary from (1) “unschooled laborer” to
(6) “higher profession”.
Resources. These were first divided into financial versus cultural resources.
Thereafter, the cultural resources were further divided into linguistic resources,
reading behavior, and pedagogical family climate.
520
Figure 1. Basic model for evaluation of the cultural-capital thesis.
9. • Financial resources. The disposable income per family member.
• Linguistic resources:
– Language choice. The number of domains in which the child speaks
Dutch (as opposed to a Dutch dialect or foreign language): with mother,
father, siblings, friends.
– Language attitude. Pertains to the language situation of the parents:
“speaks no Dutch with partner”, “speaks Dutch with partner but attaches
no importance to the child doing this”, or “speaks Dutch with partner
and attaches considerable importance to the child doing this as well”.
– Dutch language mastery. The average scores of the mother and father
in four modalities: comprehension, speaking, reading, and writing. The
level of Dutch mastery in these modalities was recorded as “poor”,
“moderate”, or “good”.
• Reading behavior. For both the mother and father, the number of hours
per week spent reading books, newspapers, and magazines.
• Pedagogical family climate:
– Help with homework. Receiving help with one’s homework from one’s
mother, father, or siblings were included in the analyses as separate vari-
ables. The responses vary from (1) “never or almost never” to (3) “fre-
quently”.
– Contacts with school. These pertain to special meetings (parent evenings,
conferences) with (1) “never or almost never”, (2) “most of the time”,
or (3) “always” as the possible responses, and to consulting with the
teacher at one’s own initiative with (1) “never”, (2) “one or two times
a year”, or (3) “three or more times a year” as the possible responses.
– Talking about school. The frequency with which the parents talk with
the child about things that happen at school and could range from (1)
“less than once a week” to (4) “every day or almost every day”.
– Importance of school. The parents were asked whether they agree with
the statement that the child should go to school as long as possible; the
response categories ranged from (1) “completely disagree” to (5)
“completely agree”.
– Importance of school-appropriate behavior. The average of the scores
for variables representing the importance that the parents attach to:
“being obedient”, “working hard at school”, “neat work”, “behaving like
other children”, and “listening to what adults have to say”.
Analysis strategy
In this study, the question of whether a relation exists across and within a
number of different ethnic groups between socioeconomic milieu and the
educational achievements of children was examined. Previous research showed
a moderate relation to exist between social milieu and achievement across
groups (Driessen and Dekkers 1997; Rossi and Montgomery 1994). The
question now is whether this relation is also encountered at the level of the
521
10. family and thus within the groups themselves or simply across groups as a
result of the correlation between average scores on social milieu and achieve-
ment. In the latter case, the relation arises from some groups producing high
scores for social milieu and achievement while other groups produce low
average scores for both social milieu and achievement; within the groups
themselves, however, a relation need not exist between social milieu and
achievement.
As indicated above, a number of studies also show cultural resources to
mediate the relation between social milieu and achievement in a total analysis
across groups. In other words: When cultural resources are held constant, the
relation between social milieu and achievement disappears. In that case there
is no direct effect of social milieu on the achievement of students. Once again,
the question is whether such a mediating effect of cultural resources can be
found within specific groups or simply emerges from the relations between
the averages for the different groups considered together. In the latter case of
no mediation, neither a relation between milieu and resources nor a relation
between resources and achievement should be observed within the groups.
We initially examined the different ethnic groups for differences in the
relations of social milieu and resources to achievement. Of particular interest
was the potentially interactive effect of ethnic group and some other pre-
dictor (social milieu or resources) on test achievement. We tested for the exis-
tence of such an interaction effect by performing a number of two-way
analyses of variance with ethnic group and one of the other variables used to
predict language and math scores. If a significant interaction effect was not
found, we could conclude that the relation of social milieu and resources to
test achievement does not differ systematically for the different ethnic groups.
In that case, we could also interpret the relation of a predictor to achieve-
ment when ethnic group was held constant as the average within-group relation
between predictor and achievement. Multiple regression analyses were also
conducted to gain greater insight into the degree of mediation by resources.
Results
Differences in distributions across ethnic groups
In Table 1, an overview of the characteristics of the different ethnic groups
is presented. For this analysis, we dichotomized the predictor variables
whenever possible.
The table shows considerable differences among the four groups. With
regard to educational level, it is striking how few Moroccan and Turkish
parents have a senior secondary vocational or higher level of education. For
occupational level, a similar situation is observed: The Moroccan and Turkish
fathers show a very unfavorable occupational position, which also manifests
itself with regard to the disposable income per family member. For the
522
12. Moroccan families, this is less than half the amount for Dutch families. This
is partly due to the low occupational level of the fathers and partly due to the
large number of people in the family (namely 6.2 as opposed to 4.4 for the
Dutch families). In the Turkish and Moroccan families, very little Dutch is
spoken and learning to speak Dutch is not considered very important. The
mastery of the Dutch language is also fairly low in the Turkish and Moroccan
families – in particular the level of mastery for the mothers. In the Turkish
and Moroccan families, the parents read very little. Once again, it is partic-
ularly the mothers who score low. This is undoubtedly related to the fact that it
is primarily the mothers who are illiterate within these ethnic minority groups.
With regard to helping the children with their homework, this is more
commonly done by the mothers than by the fathers. This may be due to the
fact that fathers work outside the home and mothers are at home. Moroccan
parents help their children least with their homework. For all of the ethnic
minority groups, it is primarily the brothers and sisters who help with
homework. Within the Dutch families, this occurs considerably less often.
Older brothers and sisters in ethnic minority families can perhaps compen-
sate for the fact that their parents have little or no education and also have
little or no mastery of the Dutch language. The percentage Turkish and
Moroccan parents who always attend parental meetings is much lower than
the percentages for the Dutch and Surinamese or Antillean parents. With
respect to contact with the teacher, there are virtually no differences among
the four ethnic groups. With respect to talking about school, however, dif-
ferences are again observed. This occurs considerably less in the Turkish and
Moroccan families than in the other families. The findings with regard to the
importance attached to attending school as long as possible are quite note-
worthy: While the three ethnic minority groups virtually do not differ in this
respect, the Dutch parents score particularly low. Also with regard to the
importance attached by parents to school-appropriate behavior (“conformity”),
no great differences were observed among the three ethnic minority groups:
They all consider school-appropriate behavior to be quite important. Dutch
parents, in contrast, attach considerably less importance to such behavior.
In sum, it can be concluded with regard to the various milieu and capital
variables that Surinamese families occupy an intermediate position between
Dutch families, on the one hand, and Turkish and Moroccan families, on the
other hand, when it comes to – in particular – social milieu, financial
resources, and linguistic resources. And what about achievement? With regard
to language in particular, there are very strong differences. For language the
total average is 70, with a standard deviation of 14, for math the average is
72 with a standard deviation of 17. On average, the Dutch children get some
80% of both the language and math problems correct. The language scores
for the three ethnic minority groups and then again for the Turkish and
Moroccan children in particular are significantly lower than those for the
Dutch children. The difference for the math scores is less marked and, within
524
13. the three ethnic minority groups, there is no difference whatsoever in the
scores.
Differences in relations among ethnic groups
In Table 2, an overview of the relation between the language and math scores
for each of the predictor variables is presented. For this analysis, we divided
the continuous variables into categories and collapsed some of the non-
continuous variables together for numerical reasons.1
The numbers 1–6 in
the heading of the table refer to the (varying numbers of) categories of these
predicator variables. For every category of predictors, the average test score
for language is presented on the first line and the average test score for math
on the second line.
The information in Table 2 shows the total correlation for almost all of
the predictors with language to be significant and relevant but just relevant
for math (using the criterion that the Eta-coefficient be a minimum of 0.15,
which was seldom found to be the case). The average within-group correla-
tion was not significant for either language or math, however, and certainly
not relevant. One exception is the correlation between language choice and
language achievement, which showed a just relevant effect. The strength of
the effect of the interaction between ethnic group and predictor on achieve-
ment shows that we can interpret the average within-group correlation as indi-
cating “the relation as it is within each of the groups”: Not one of the
interaction effects proved significant. The correlations of the various predictor
variables with achievement do not, thus, vary significantly for the different
ethnic groups. Almost all of the correlations do not deviate systematically
from zero; none of the within-group correlations were found to be significant
(with the exception of language choice on language score).
The information with regard to social milieu suggests that the “total effect”
of social milieu on achievement presupposed in previous research must
possibly be seen as a level of interpretation mistake. In other words, it is
suggested that an effect exists at the level of the family while this is not the
case: Such effects exist more at the level of the ethnic groups themselves.
Dutch families have on average both a higher social milieu and in particular
a higher language achievement level when compared to the other groups. This
leads to a total positive correlation between social milieu and achievement.
Within the groups, however, such a relation is virtually nonexistent. The pre-
supposed mediation of the relation between social milieu and achievement
by resources may also rest on an error in the level of interpretation. Given
that the resources within the groups do not correlate with achievement, it
is very unlikely that one can speak of within-group mediation – that is, a
mediating effect of resources on achievement at the level of the family.
525
16. The mediating effect of resources
In order to gain greater insight into the possibly mediating effects of the
various separate resource indicators, multiple regression analyses were under-
taken with social milieu as the predictor of the language and math scores. If
we then add resources as predictors, the independent contribution of social
milieu can be identified and thereby afford insight into the meditating role of
resources. In order to simplify the analyses to some extent, the three indica-
tors of social milieu were averaged together. In Table 3, the correlations of
social milieu with the language and math scores, respectively, and the inter-
correlations between the language and math scores themselves are presented.
As can be seen, the correlations between social milieu and language
score are smaller for the different groups than the total correlation. The total
correlation is thus, in part, caused by the correlation of the group averages.
For Dutch families, one can speak of a relatively high social milieu and a
relatively high language test score for the children while the opposite holds
for the Turkish and Moroccan families (also see Table 2). The preceding holds
to a lesser degree for the math scores; in particular, because the average math
scores for the different groups differ less. We can only speak of a just relevant
correlation between social milieu and math achievement at the level of the
family for the Dutch families.
In order to gain insight into the linear relations between the variables, we
will first present an overview of the correlations of social milieu with resources
(see Table 4) and resources with the test scores (see Table 5). Our original
intention was to collapse the resource indicators into the latent constructs for
the model depicted in Figure 1. The correlations between the various indica-
tors were not sufficiently strong, however, so it was decided to include the
indicators as separate variables in the analyses.
From Table 4, it can be seen that reasonably high total correlations exist
between social milieu and both financial and linguistic resources in particular.
The correlations of social milieu with indicators of the pedagogical climate
are clearly lower or not relevant at all. It can also be deduced from the
information in this table that every total correlation consists at least in part
of the correlation between the group averages for social milieu and resources.
528
Table 3. Correlations of social milieu with language and math scores; intercorrela-
tions between language and math scores.
Language Math Language × math
Total –0.34 –0.17 0.44
Dutch –0.27 –0.23 0.41
Surinamese –0.14 –0.05 0.40
Turkish –0.02 –0.04 0.37
Moroccan –0.04 –0.05 0.40
17. In almost all of the ethnic groups, the relevant correlation is lower than the
total correlation. For the Turkish and Moroccan families in particular, the
correlations are frequently much lower than for the Dutch and Surinamese
families. There are nevertheless exceptions here as well: Within the Moroccan
families, the reading behavior of the mother and the provision of help with
homework by both the mother and father are determined to a reasonable degree
by the social milieu of the family. This is not, however, the case for the Turkish
families.
The information in Table 5 shows almost a complete absence of within-
group correlations between resources and test scores. For only the Surinamese
families were language choice (number of domains), language mastery of the
mother, and provision of help with homework by siblings found to signifi-
cantly correlate with language score. This is in line with what was concluded
on the basis of Table 2, namely that total correlations exist, in particular,
between average financial and linguistic resources and the average language
scores of the students, while the average within-group correlations are virtu-
ally nonexistent.
Note that the correlations between help with homework by the father or
siblings (and for Surinamese families: the importance of school appropriate
behavior) with achievement were, in so far as they proved relevant, negative
for the Surinamese and Turkish families. One can imagine that the provision
of help with homework by the father or siblings leads (in the causal sense)
to a decline in achievement as a result of incorrect instruction although the
alternative explanation seems more obvious, namely: poor test achievement
may prompt a father or siblings to provide help with homework. The position
of these variables as predictors of achievement can therefore be questioned.
529
Table 4. Correlations between social milieu and resources.
Total Dut Sur Trk Mor
Disposable income –0.46 –0.41 –0.39 –0.17 –0.19
Language choice –0.33 –0.16 –0.12 –0.03 –0.01
Language attitude –0.47 –0.20 –0.23 –0.13 –0.08
Language mastery, mother –0.50 –0.23 –0.31 –0.23 –0.34
Language mastery, father –0.48 –0.29 –0.35 –0.36 –0.23
Reading, mother –0.35 –0.22 –0.25 –0.08 –0.27
Reading, father –0.28 –0.30 –0.26 –0.11 –0.17
Help homework, mother –0.24 –0.02 –0.00 –0.01 –0.29
Help homework, father –0.19 –0.02 –0.06 –0.12 –0.37
Help homework, siblings –0.29 –0.10 –0.03 –0.07 –0.07
Contact school –0.20 –0.10 –0.11 –0.12 –0.06
Contact school self –0.09 –0.09 –0.12 –0.09 –0.17
Talk about school –0.22 –0.15 –0.12 –0.02 –0.07
Importance of school –0.17 –0.00 –0.02 –0.13 –0.00
School-appropriate behavior –0.28 –0.25 –0.12 –0.10 –0.03
R –0.66 –0.57 –0.54 –0.45 –0.42
19. Up to this point, we have spoken about the total correlation of social milieu
and resources with achievement. In doing this, however, the correlations
between the various resources have not been taken into consideration. In the
regression analyses presented below, these correlations were therefore con-
sidered and, using the partial correlation coefficients, the independent or direct
effects of social milieu and resources on test scores can be determined. In
Table 6, the partial correlation coefficients and the total variance explained
in the language and math scores, respectively, by all of the resources con-
sidered together are presented. The increase in the amount of variance
explained as a result of the addition of social milieu is also presented. Any
increase is the result of the direct effects of social milieu on the test
scores.
Despite the virtually complete absence of strong direct effects of resources
on test scores, one can speak of a relatively high degree of explained variance
with respect to language scores for the four groups together and within the
Surinamese group. This suggests that the resources do not correlate very
strongly with each other. That is, the explained variance consists of the
numerous small contributions of various resources rather than a limited
number of resources with a strong direct effect on achievement. Within the
ethnic groups themselves, the addition of social milieu was only found to be
of some significance for the Dutch families. That is, we can only speak of a
just relevant direct effect of social milieu on both language and math scores
within the Dutch families. In Table 7, a complete summary of the preceding
findings is presented.
The difference between the total correlation and the partial correlation indi-
cates the strength of the mediating effect of resources. We can speak of a
reasonably strong mediating effect of resources on the language scores. The
mediating effect on the math scores is quite weak but nevertheless signifi-
cant due to the large N. It can also be seen within the groups that the total
and partial correlations are only significant for the Dutch families. The medi-
ating influence of resources is virtually non-existent, however. The total medi-
ating influence of resources consists of all kinds of indirect effects of social
milieu on the test scores via the resources. The indirect effect of a particular
resource is equivalent to the product of the correlation between social milieu
and resource (see Table 4) and the independent, i.e., partial correlation between
resource and test score (see Table 6). In light of the almost complete absence
of direct effects of resources on test scores for math in particular, one can also
speak of virtually no mediating effect of resources within groups. The p-values
(under “difference”) show non-significant differences between the total and
partial correlations with the exception of the Surinamese families for which
the total correlation with the language score drops from weak (0.14) to almost
zero (0.02). The mediation occurs via disposable income, language choice,
and language mastery by the mother, which account for the largest part of
the indirect influence, namely: 0.39 × 0.10 + 0.12 × 0.19 + 0.31 × 0.13 = 0.10.
Across groups, one can speak of a mediating effect of resources in the
531
22. prediction of language scores as a result in part of the higher correlations
between social milieu and resources.
Conclusions
On the basis of the results presented here, the following conclusions can be
drawn.2
• There are no systematic differences between the ethnic groups with respect
to the total and independent effects of social milieu on test scores. The cor-
relations do not differ systematically from the very weak average within-
group correlation.
• The average direct effects of social milieu on language and math achieve-
ment are only very weak.
• Within ethnic groups, there is no mediating effect of resources. An excep-
tion is the Surinamese group with respect to the prediction of language
score, where mediating effects of disposable income per person, language
choice, and language mastery by the mother are found.
• Across groups, one can speak of a mediating effect of resources with respect
to the prediction of language score as a result of the higher correlations
between social milieu and resources.
• The results suggest that the total effect of social milieu on achievement in
ethnic groups presupposed in previous research may be a level of inter-
pretation error. The presupposition is namely that the effect exists at the
level of the family/student, but this is only very partly the case: The effect
lies more at the level of the separate ethnic groups. The presupposed medi-
ation of this relation by cultural resources also appears to rest on the same
level of interpretation error.
The research described here is fairly unique in the Netherlands. Large-scale,
quantitative research into the effects of cultural capital in which the ethnic
origin of the participants is explicitly taken into consideration is only sporadic.
Research in which separate analyses are conducted for the different ethnic
groups and the results of these analyses are then compared has not been con-
ducted. Our study must therefore be seen as an attempt to explore differences
in various aspects of capital among ethnic groups. A limitation of the study
is that the data were not specifically collected for testing the cultural capital
thesis for different ethnic groups. However, in this respect our study does not
differ from most of the studies we reviewed.
What the present study shows is that it may be relevant to split the total
sample into separate subgroups to examine the ethnic factor. Findings that
hold for the entire sample may not necessarily hold for each of the ethnic
groups separately. This may be caused by differences in the distributions of
the variables to be explained and the explanatory variables (the restriction-
of-range problem) and their inter-relations within the different groups (see
534
23. Leseman and Van den Boom 1999). With regard to Bourdieu’s reproduction
thesis, the present findings provide no confirmation. In follow-up research,
the significance of cultural capital within each of the ethnic groups should be
considered along with the extent to which the concept can be compared across
groups. The fundamental question stemming from the present study is whether
it is even possible to study the effects of cultural capital across groups in
light of the fact that little or no variation for both indicators of cultural capital
and achievement can be detected in some groups and the fact that cultural
capital is generally defined according to the standards of higher status groups.
This conclusion expands on the comments of Lareau and McNamara Horvat
(1999: 50) who observed that “Any form or type of capital derives value only
in relation to the specific field of interaction.” Our conclusion also fits with
the suggestion of Roscigno and Ainsworth-Darnell (1999: 173), namely “that
differential racial returns are a consequence of the inapplicability of the status
attainment model to nonwhites” (see also Okagaki and Frensch 1998). The
methodological problem of contamination of specific strongly cultural defined
indicators with ethnic origin also needfully requires further study. The enter-
tainment of separate indicators of capital for each ethnic group does not,
however, offer much solace for this problem as the groups become incom-
parable. A solution to this dilemma will not be easy to find, thus.
Acknowledgments
I would like to express my gratitude to the Netherlands’ Organization for
Scientific Research (NWO) for funding the project on which this article is
based. This research was made possible by grant # 575-53-013 from NWO’s
Foundation for Behavioral and Educational Sciences. I would also like to thank
Adrie Claassen and Jan Doesborgh for their help with the review of the lit-
erature and the empirical analyses, respectively.
Notes
1. For the analyses of variance, some of the categories for the predictor variables were
collapsed: the two highest categories of education “higher professional education”
and “university education”; the occupational categories “schooled laborer”and
“lower employee”; income was divided into five categories ranging from (1) “up
to 500 guilders” to (5) “1100 guilders or more”; language choice was reduced to
three categories (1) “one or no domains of use”, (2) “two to three domains”, and
(3) “four domains”; language attitude was dichotomized as (1) “considers speaking
Dutch important for the child” or (2) “does not consider speaking Dutch impor-
tant for the child”; language mastery was dichotomized as (1) “none” or (2) “all
four aspects well”; reading behavior had five categories ranging from (1) “two or
fewer hours” to (5) “eleven or more hours”; for contact with the teacher, “almost
never” was collapsed with “usually”; for talking about school, the two lowest cat-
egories “less than once a week” and “once per week” were collapsed; for impor-
535
24. tance of education, the two lowest categories “disagree completely” and “disagree”
were collapsed; and importance of school appropriate behavior had five categories
ranging from (1) < 2.6, to (5) > 3.4.
2. Multi-sample LISREL analyses were also conducted on the same data; the
results of these analyses did not differ from the multiple regression results reported
here.
References
Ainsworth-Darnell, J. and Downey, D. 1998. Assessing the Oppositional Culture
Explanation for Racial/Ethnic Differences in School Performance. American
Sociological Review 63: 536–553.
Aschaffenburg, K. and Maas, I. 1997. Cultural and Educational Careers: The Dynamics
of Social Reproduction. American Sociological Review 62: 573–587.
Bankston, C., Caldas, S. and Zhou, M. 1997. The Academic Achievement of
Vietnamese American Students: Ethnicity as Social Capital. Sociological Focus 30:
1–16.
Bianchi, S. and Robinson, J. 1997. What Did You Do Today? Children’s Use of Time,
Family Composition, and the Acquisition of Social Capital. Journal of Marriage and
the Family 59: 332–344.
Bourdieu, P. and Passeron, J.-C. 1977. Reproduction in Education, Society, Culture.
Beverly Hills, CA: Sage.
Carrington, V. and Luke, A. 1997. Literacy and Bourdieu’s Sociological Theory: A
Reframing. Language and Education 11: 96–112.
Coleman, J. 1988. Social Capital in the Creation of Human Capital. American Journal
of Sociology 94: s95–s120.
Covrig, D. 1996. Fundamentalists, Social Capital, and Children’s Welfare: A Place
for Religion in Public Education? Journal of Educational Policy 12: 53–60.
De Graaf, P. 1986. The Impact of Financial and Cultural Resources on Educational
Attainment in the Netherlands. Sociology of Education 59: 237–246.
DiMaggio, P. 1982. Cultural Capital and School Success: The Impact of Status Culture
Participation on the Grades of US High School Students. American Sociological
Review 47: 189–210.
Downey, D. 1995. When Bigger is Not Better: Family Size, Parental Resources, and
Children’s Educational Performance. American Sociological Review 60: 746–761.
Driessen, G. 1993. Achtergronden van Onderwijsprestaties. De Rol van
Gezinsstructurele en Gezinspedagogische Kenmerken. Stimulans 11: 14–17.
Driessen, G. 2000. The Limits of Educational Policy and Practice? The Case of Ethnic
Minority Pupils in the Netherlands. Comparative Education 36: 55–72.
Driessen, G. and Dekkers, H. 1997. Educational Opportunities in the Netherlands.
Policy, Students’ Performance and Issues. International Review of Education 43:
299–315.
Driessen, G., Doesborgh, J. and Claassen, A. 1999. Cultureel Kapitaal, Etnische
Herkomst en Onderwijsprestaties. Nijmegen: ITS.
Driessen, G., Mulder, L. and Jungbluth, P. 1999. Structural and Cultural Determinants
536
25. of Educational Opportunities in the Netherlands, in S. Weil, ed., Roots and Routes:
Ethnicity and Migration in Global Perspective (83–104). Jerusalem: Magnes Press.
Driessen, G. and Vierke, H. 1999. Abordar la Igualdad y la Calidad Educativas a
Través de la Investigación: Como Ejemplo un Estudio Nacional de Cohortes. Revista
de Educación 319: 37–60.
Farkas, G. et al. 1990. Cultural Resources and School Success: Gender, Ethnicity,
and Poverty Groups within an Urban School District. American Sociological Review
55: 127–142.
Furstenberg, F. and Hughes, M. 1995. Social Capital and Successful Development
among At-Risk Youth. Journal of Marriage and the Family 57: 580–592.
Gillborn, D. 1997. Ethnicity and Educational Performance in the United Kingdom:
Racism, Ethnicity, and Variability in Achievement. Anthropology & Education
Quarterly 28: 375–393.
Kalmijn, M. and Kraaykamp, G. 1996. Race, Cultural Capital, and Schooling: An
Analysis of Trends in the United States. Sociology of Education 69: 22–34.
Lareau, A. and McNamara Horvat, E. 1999. Moments of Social Inclusion and
Exclusion. Race, Class, and Cultural Capital in Family-School Relationships. Sociology
of Education 72: 37–53.
Leseman, P. et al. 1995. Gezinsdeterminanten van de Cognitieve Ontwikkeling van
Vierjarige Nederlandse, Surinaamse en Turkse Kleuters. Pedagogische Studiën 72:
186–205.
Leseman, P. and van de Boom, D. 1999. Effects of Quantity and Quality of Home
Proximal Processes on Dutch, Surinamese-Dutch and Turkish-Dutch Pre-schoolers’
Cognitive Development. Infant and Child Development 8: 19–38.
Okagaki, L. and Frensch, P. 1998. Parenting and Children’s School Achievement: A
Multiethnic Perspective. American Educational Research Journal 35: 123–144.
Roscigno, V. and Ainsworth-Darnell, J. 1999. Race, Cultural Capital, and Educational
Resources: Persistent Inequalities and Achievement Returns. Sociology of Education
72: 158–178.
Rossi, R. and Montgomery, A. 1994. Educational Reforms and Students at Risk. A
Review of the Current State of the Art. Washington, DC: US Department of Education.
Runyan, D. et al. 1998. Children Who Prosper in Unfavorable Environments: The
Relationship to Social Capital. Pediatrics 101: 12–18.
Stanton-Salazar, R. 1997. A Social Captital Framework for Understanding the
Socialization of Racial Minority Children and Youths. Harvard Educational Review
67: 1–40.
Sullivan, L. 1997. Hip-Hop Nation: The Underdeveloped Social Capital of Black
Urban America. National Civic Review 86: 235–243.
Teachman, J., Paasch, K. and Carver, K. 1996. Social Capital and the Generation of
Human Capital. Social Forces 75: 1343–1359.
Tesser, P. and Mulder, L. 1990. Cultureel Kapitaal en Schoolprestaties in het
Basisonderwijs. in C. Klaassen and P. Jungbluth, eds., Onderwijs Research Dagen
1990. Onderwijs en Samenleving (39–51). Nijmegen: ITS.
Van der Velden, R. 1991. Sociale Herkomst en Schoolsucces. Het Effect van Culturele
en Sociale Hulpbronnen op de Schoolloopbaan. Groningen: RION.
537
26. White, M. and Kaufman, G. 1997. Language Usage, Social Capital, and School
Completion among Immigrants and Native-Born Ethnic Groups. Social Science
Quarterly 78: 385–398.
The author
Geert Driessen is an educational researcher at the Institute for Applied Social Sciences
(ITS) of the University of Nijmegen, the Netherlands. His major research interests
include ethnic and social inequality in education.
Contact address: Dr. Geert Driessen, ITS, Universiteit van Nijmegen, Postbus 9048,
6500 KJ Nijmegen, the Netherlands. E-mail: g.driessen@its.kun.nl.
538