Cultural Capital, Rational Choice and Educational Inequalities

Transcription

Cultural Capital, Rational Choice and
Educational Inequalities
Alice Sullivan
August 30, 2003
Introduction
Background
The educational disadvantage suffered by children from working class families
has long been the focus of sociological attention. Traditionally, the reasons
for this disadvantage have been taken as being fairly obvious, and therefore the focus was on promoting educational reforms such as the universal
provision of free and compulsory education. The problem to be solved was
the wastage of working class talent, rather than social class differentials in
educational attainment per se (Lindsay, 1926). There is evidence that educational reforms have reduced absolute differences in rates of educational
participation between the classes (Jonsson and Mills, 1993a; Jonsson and
Mills, 1993b; Hellevik, 1997). However, it seems that the association between social class and educational attainment has remained intact despite
these reforms (Halsey et al., 1980; Shavit and Blossfeld, 1993). The failure
of educational reforms to eradicate the link between social class and educational attainment has led sociologists to focus on the question of why this
association exists.
The emergence of “culturalist” approaches can be seen as a natural consequence of the failure of educational reforms to dramatically alter the association between social class and educational outcomes. If the lifting of economic
barriers to educational participation did not eradicate social class differences
in educational outcomes, then it may be fruitful to look to cultural, rather
than (or perhaps, as well as) economic differences between the classes to
explain class inequalities in educational attainment.
One cultural difference that has been invoked to explain the social class differential in educational attainment is that of language. Bernstein focuses
on class differences in language to explain working class educational under-
i
ii
achievement. Bernstein, for instance, claims that working class people have
access to only a “restricted” language code whereas middle class people have
access to an “elaborated” code. In the restricted code “The meanings are
likely to be concrete, descriptive or narrative rather than analytical or abstract.” (Bernstein, 1973, p. 128). Bernstein claims that the school is based
on the elaborated code, in that it transmits “de-contextualised” and “universalistic” meanings. This puts working class students at a disadvantage,
because of the gap between their code, and the code of the school.
Bernstein has been misinterpreted (for instance, by Labov (Labov, 1972)) as
a deficit theorist, blaming working class children for their own educational
failure. A more relevant criticism is it is not clear that working class people
actually do have a “restricted code”, as opposed to perhaps simply having
relatively smaller vocabularies than middle class people (see Lawton (Lawton,
1968)). Neither is there any empirical evidence that the education system
is (or was) based on an elaborated code. Such evidence as there is (Cooper,
1976) points to the opposite conclusion.
Advocates of theories of “cultural reproduction” claim that the function of
the school is to legitimate economic inequalities by giving an appearance of
meritocracy, see for instance Collins (Collins, 1971),Jencks (Jencks, 1972)
and Bowles and Gintis (Bowles and Gintis, 1976).
Pierre Bourdieu is the most influential advocate of the “cultural reproduction” approach. Bourdieu’s work has strong parallels with that of Bernstein,
not least in his focus on the use of language. Bourdieu’s theory of cultural
reproduction states that the school rewards pupils who possess the cultural
traits associated with the middle classes. These include the ability to use “educated” language, as well as participation in “high-brow” cultural activities.
Another parallel between Bernstein and Bourdieu is the lack of clarity in their
prose styles, and the corresponding difficulty in deriving testable hypotheses from their theories. Perhaps because Bourdieu is the most prominent
theorist of cultural reproduction, his theory has inspired a great deal of empirical work, despite the problems inherent in attempting to operationalise
his concepts.
It should be noted that there is nothing new in the observation that the
middle class child often enjoys cultural as well as economic advantages. This
insight need not go hand in hand with an acceptance of “cultural reproduction” theory as such. Floud et. al. (Floud et al., 1956) divide the resources
associated with the home into “material” and “cultural” categories. Their
measure of cultural resources includes parents’ knowledge of the selection
iii
procedures of the grammar schools, parents’ visits to the child’s school, parents’ aspirations and preferences for the child’s education, newspapers and
magazines read and library membership. In retrospect, this measure seems
to combine some elements of both Bourdieu’s concept of “cultural capital”
and Coleman’s concept of “social capital” (Coleman, 1990).
The “cultural reproduction” approach stands in marked contrast to the “human capital” approach put forward by Becker (Becker, 1964) according to
which education makes people more productive, and this is why people with
higher levels of education are rewarded with relatively high pay. More generally, cultural reproduction is seen as being in conflict with the “rational
choice” approach, which tends to stress the importance of economic constraints in explaining social class differentials in educational attainment.
Boudon (Boudon, 1974) makes a crucial distinction between the “primary”
and “secondary” effects of stratification. The primary effects of stratification
are cultural inequalities that determine the academic abilities of pupils. The
secondary effects of stratification are the different costs and benefits that are
associated with different educational decisions for pupils from different social
classes. The secondary effects explain any social class difference in educational participation that remains once one has controlled for performance at
the previous stage. Boudon claims that the secondary effects of stratification
are more important than the primary effects, as the secondary effects are
exponential.
It seems natural that rational choice theorists should focus on key transition
points rather than on class differences in educational attainment within a
given curriculum, since choices are made at these transition points, whereas
one’s level of performance, particularly in the earliest years of schooling, can
hardly be seen as a simple matter of choice, even if the child does have some
autonomy in the question of how hard to work. However, there is no contradiction inherent in drawing upon both “cultural reproduction” and “rational
choice” perspectives to explain educational inequalities. Furthermore, the
question of whether economic or cultural factors are more important in explaining social class differentials in educational attainment can be seen as
distinct from the question of whether the primary effects of stratification are
greater than the secondary effects. It may be that economic factors have a
strong impact on performance before any key transitions have been made,
and it may also be that cultural factors affect the decisions made regarding
these transitions.
The literature I have cited so far has focused on social class inequalities,
iv
and has ignored other structures of inequality, notably gender. Traditionally, gender inequalities in educational performance were seen as natural and
unproblematic. All male samples were used in both quantitative and qualitative work, (see for instance, Halsey et. al. (Halsey et al., 1980), Willis
(Willis, 1977)) and women and girls were therefore excluded as subjects of
analysis.
More recently, although mixed samples have been collected by quantitative
researchers, the characteristics of mothers have sometimes been deemed irrelevant, and gender differentials in educational attainment have often not been
seen as worthy of explanation. Feminist ethnographic work, though taking
gender differences seriously, has often been based on an essentialist conception of gender differences as being intrinsic and fundamental. So, it has been
argued that achievement and rationality are male values, and girls will inevitably fail in an educational system which stresses these values. From the
fact that boys’ academic achievement has traditionally been higher than that
of girls it is inferred that “achievement” is a male value. So, Arnot claims
that girls have different values than boys, and girls who fail in school are
merely expressing these values. They “. . . celebrate their femininity through
a rejection of male culture that stresses the value of hierarchies. . . objective
versus subjective knowledge, and individual competition above co operation”
(Arnot, 1994, p. 95).The traditional view of gender differences has a great
deal in common with the essentialist view. Both are ahistorical, and neither can account for the fact that girls are increasingly out performing boys
educationally in many industrialised countries (Erikson and Jonsson, 1996;
Shavit and Blossfeld, 1993).
Of course, it is not surprising that girls’ educational performance has improved given womens’ increased labour market participation. However, the
fact that girls are actually outperforming boys at GCSE has caused a certain
amount of panic. A discourse of male disadvantage has emerged, according
to which, boys are failing because of a “crisis of masculinity” or because too
many teachers are women (Weiner et al., 1997). This discourse ignores the
fact that boys’ educational attainment has not deteriorated. It has improved,
but just not as much as that of girls.
v
Aims
This thesis aims to examine the mechanisms behind social class and gender
differences in educational outcomes, and to evaluate the usefulness of cultural
reproduction theory and rational choice theory in explaining educational inequalities.
Bourdieu’s theory of cultural reproduction is often vague and ambiguous (for
further discussion of Bourdieu’s theory see chapter 1. The main concept that
empirical researchers have taken from the theory is that of cultural capital.
Cultural capital can be defined loosely as those cultural traits that help people to gain educational success. Empirical researchers have often failed to
grapple with the problem of operationalising the concept of cultural capital,
preferring to use proxy measures such as parental education. A key aim of
this thesis is to clarify the concept of cultural capital. As broad as possible an
operationalisation of cultural capital will be used in order to assess which aspects of cultural participation actually constitute cultural capital in the sense
of having a pay off in terms of educational performance. Bourdieu’s theory
of cultural reproduction is also unclear on the question of why cultural participation should lead to educational success. I will assess the possibility that
relatively high levels of linguistic ability and/or cultural knowledge are the
mechanisms through which cultural participation translates into educational
success.
Rational choice theory asserts that individuals act rationally in order to fulfil
their desires in the light of their beliefs about the situation (or at least that,
on average, people behave as though this was what they were doing). In itself, this is a rather vacuous theory. If it is to yield determinate predictions,
ancillary assumptions regarding the beliefs and desires of agents must be
made. Rational choice theorists who have attempted to give explanations of
educational inequalities have made many assumptions about the motivations
of agents, but have been loath to apply any empirical test to these assumptions. Data on actions is seen as “hard” and therefore reliable, whereas data
on the beliefs and desires of agents is seen as “soft” and therefore unreliable. The doctrine of “revealed preference” has led rational choice theorists
to attempt to infer motivations from actions. The problem with this is, of
course, that one can infer any number of quite different motivations from any
given action. This is reflected in the wide variety of explanations for social
class inequalities in educational outcomes given by rational choice theorists.
This thesis will attempt to directly examine whether pupils’ beliefs (about
their academic abilities) and attitudes towards education vary according to
vi
social class and gender. Plans for further education will be analysed in order to assess whether significant social class and gender differences remain
once GCSE performance has been controlled for, and in order to attempt to
explain any such differences.
Various versions of rational choice and cultural reproduction explanations of
educational inequalities have been both advocated or rejected on the basis
of inadequate empirical evidence or even no empirical evidence at all. A
common problem has been the tendency of researchers to attempt to answer
questions using the data that is to hand, regardless of whether this data can
actually provide a good test of hypotheses derived from the theory to be
assessed. I have tried to collect appropriate data to answer the questions
I am posing here. My sample consists of 465 pupils in their final year of
compulsory education in four English schools.
From the point of view of examining educational differentials, an advantage of the contemporary British context is that the first crucial educational
“branching point” has been pushed back to the end of compulsory education, age 16. This means that in analysing results in the GCSE exams, the
whole ability range is covered, and one does not risk conflating the “primary
effects” of stratification with the “secondary effects” in Boudon’s sense.
The GCSE (General Certificate of Secondary Education) exams are taken by
all pupils at the end of their compulsory schooling, when they are approximately 16 years old. The GCSE examination was introduced in England
and Wales in 1985, replacing CSEs and O levels. CSEs were of a lower academic standard than O levels, and were designed for pupils in the secondary
modern schools. (Before 1965, Britain had a tripartite system of secondary
education, and children were allocated to grammar, technical, or secondary
modern schools at age 11. Subsequently, local authorities gradually introduced the comprehensive system, and there are now only 162 selective state
schools in Britain). The introduction of the GCSE means that all fourteen to
sixteen year olds, regardless of academic ability, follow a similar curriculum
(although there is some choice of subjects). Typically, students study 8 or 9
subjects, and are assessed partly by examination and partly by coursework.
Pass grades range from G to A*. (The A* grade was recently introduced due
to fears that there was not enough differentiation at the top level). More generally, there has been an increasing focus on credential outcomes within the
education system. A national curriculum has been introduced, with standard
tests for seven, eleven and fourteen year olds, and the publication of “league
tables” of schools’ results in these tests, along with increased parental choice
vii
of school, has put great pressure on schools to focus on improving test and
examination results overall.
I will attempt to answer the following key questions.
1. Cultural Reproduction and GCSE performance
• Are cultural practices transmitted from parents to children?
• What is cultural capital and how important is it for educational
attainment and participation?
• Which elements of the dominant culture are actually associated
with educational success?
• To what extent can differences in cultural participation account
for social class differences in educational outcomes?
• Are there gender differences in cultural participation, and, if so,
do these help to explain girls’ superior performance compared to
boys at GCSE?
• If cultural participation is associated with educational performance,
why is this so?
2. Rational Choice and Post-16 Decisions
• How important are the “secondary effects” of stratification at the
key transition from GCSE to further education? Is it the case, for
instance, that working class pupils are less likely to choose to do
A levels once GCSE performance has been controlled for?
• If there are significant secondary effects of stratification, how does
the size of these effects compare to the size of social class differences in GCSE performance?
• To what extent is subject choice associated with pupils’ sex?
• How can we explain any social class and gender differences in
educational decision making, controlling for GCSE results?
• Do attitudes towards education and beliefs about ability differ
according to social class and gender?
• Do occupational aspirations differ according to social class and
gender?
• Are occupational aspirations associated with educational decisions, controlling for other relevant variables?
viii
I have divided the key questions above into those associated primarily with a
cultural reproduction approach and with initial academic performance (or the
primary effects of stratification) and those associated with a rational choice
approach and with educational decision making (or the secondary effects of
stratification). Although this is a convenient classification, the comment I
made above that the primary effects of stratification may well not be entirely
cultural, nor the secondary effects entirely economic, will not be forgotten in
the analysis to follow.
Outline
Introduction
Bourdieu’s Theory of Cultural Reproduction In this chapter I will
outline and discuss Bourdieu’s theory, and both Bourdieu’s attempts to apply
empirical evidence to this theory, and the work of other researchers on this
theme. It will be argued that previous operationalisations of the concept
have been inadequate, and that a broader operationalisation of the concept
of cultural capital is necessary.
Rational Choice Theories of Educational Inequality This chapter
will discuss rational choice theory, and the various attempts to apply rational choice theory to the question of educational inequalities. I will argue
that rational choice theorists have made various assumptions regarding the
motivations of agents, but that these assumptions are in need of empirical
test.
Methodology This chapter will describe my operationalisation of concepts, questionnaire design, sample, and data collection process.
Cultural Capital and Educational Attainment This chapter includes
an empirical analysis of the link between parents’ cultural participation and
pupils’ cultural participation, and between pupils’ cultural participation and
GCSE attainment. I will show that reading and television viewing habits
are the only types of cultural participation that are associated with GCSE
ix
attainment, once other relevant factors have been controlled for, and the
effect of these variables is mediated entirely by pupils’ cultural knowledge
and vocabulary.
Students’ Beliefs and Desires This chapter will assess the question of
whether students’ attitudes to education and beliefs about their own academic abilities vary according to social background and gender. I will conclude that pupils’ attitudes to education vary only slightly by social class,
and that this effect is entirely mediated by school success as measured by
GCSE results. However, pupils’ attitudes do vary by gender, and both the
social background and gender of pupils affect their perception of their own
abilities, with service class and male pupils’ overestimating their own abilities
as compared to non-service class and female pupils.
Students’ Decisions This chapter will address the question of social class
and gender differences in rates of educational participation post-16. Social
class has no direct effect on participation in further education for my sample.
So, in this case, the ‘secondary effects’ of stratification seem to be nonexistent. However, parental education does have a significant association
with the decision to do A levels, controlling for pupils’ GCSE results. Girls
are significantly less likely to choose hard science subjects than are boys, and
this seems to be partly explained by girls’ relatively lower assessment of their
own abilities.
Conclusions This chapter will summarise my results and provide some
suggestions for future research.
Contents
1 Cultural Reproduction
1
1.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
1.2
Bourdieu’s theory . . . . . . . . . . . . . . . . . . . . . . . . .
2
1.2.1
Cultural Capital . . . . . . . . . . . . . . . . . . . . .
2
1.2.2
Habitus . . . . . . . . . . . . . . . . . . . . . . . . . .
9
1.2.3
Political Action . . . . . . . . . . . . . . . . . . . . . . 13
1.2.4
Science and Language . . . . . . . . . . . . . . . . . . 15
1.3
1.4
Empirical Evidence on Cultural Capital . . . . . . . . . . . . . 17
1.3.1
Bourdieu’s Own Evidence . . . . . . . . . . . . . . . . 17
1.3.2
Other Research on Cultural Reproduction . . . . . . . 19
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2 Rational Choice Perspectives
24
2.1
The Utility Model . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2
Murphy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.2.1
Formalising Murphy . . . . . . . . . . . . . . . . . . . 29
2.3
Boudon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.4
Gambetta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
x
CONTENTS
xi
2.4.1
Formalising Gambetta . . . . . . . . . . . . . . . . . . 34
2.5
2.6
Breen and Goldthorpe’s Model . . . . . . . . . . . . . . . . . . 34
2.5.1
Applying the Utility Model to Breen and Goldthorpe . 38
2.5.2
Breen and Goldthorpe’s Assumptions . . . . . . . . . . 41
2.5.3
A Generic Model? . . . . . . . . . . . . . . . . . . . . . 44
2.5.4
Breen and Goldthorpe’s attempt to explain empirical
changes . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3 Methodology
3.1
3.2
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.1.1
Cultural Capital . . . . . . . . . . . . . . . . . . . . . 50
3.1.2
Rational Choice Theory . . . . . . . . . . . . . . . . . 51
Survey Design . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.2.1
3.3
3.4
50
Pilot . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Questionnaire Design and Coding . . . . . . . . . . . . . . . . 56
3.3.1
Cultural Capital Questions . . . . . . . . . . . . . . . . 56
3.3.2
Viewing Habits . . . . . . . . . . . . . . . . . . . . . . 66
3.3.3
Formal Culture and Music
3.3.4
Parents’ Cultural Capital . . . . . . . . . . . . . . . . . 70
3.3.5
Validity of Cultural Capital Questions . . . . . . . . . 71
3.3.6
Rational Choice Questions . . . . . . . . . . . . . . . . 73
3.3.7
Reliability of Rational Choice Questions . . . . . . . . 74
3.3.8
Parents’ Social Class and Qualifications . . . . . . . . . 75
. . . . . . . . . . . . . . . 68
Sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
CONTENTS
xii
3.4.1
Administration of questionnaire . . . . . . . . . . . . . 78
3.5
Response Rates . . . . . . . . . . . . . . . . . . . . . . . . . . 80
3.6
Ethics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.7
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
4 Educational Attainment at GCSE
83
4.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
4.2
Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.3
4.2.1
Parental Cultural Capital . . . . . . . . . . . . . . . . 84
4.2.2
Pupils’ Cultural Capital . . . . . . . . . . . . . . . . . 85
4.2.3
Language and Knowledge . . . . . . . . . . . . . . . . 88
4.2.4
Gender . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
4.2.5
GCSE attainment . . . . . . . . . . . . . . . . . . . . . 95
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
5 Beliefs and Desires
111
5.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
5.2
Rational Choice Theories . . . . . . . . . . . . . . . . . . . . . 113
5.3
Research Questions . . . . . . . . . . . . . . . . . . . . . . . . 115
5.4
Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
5.5
5.4.1
Attitudes to Education . . . . . . . . . . . . . . . . . . 116
5.4.2
Beliefs About Ability . . . . . . . . . . . . . . . . . . . 121
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6 Pupils’ Choices
6.1
129
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
CONTENTS
xiii
6.2
Research Questions . . . . . . . . . . . . . . . . . . . . . . . . 131
6.3
Occupational Aspirations . . . . . . . . . . . . . . . . . . . . . 132
6.4
Attitudes to Jobs . . . . . . . . . . . . . . . . . . . . . . . . . 134
6.5
Occupational Aspirations and Values . . . . . . . . . . . . . . 140
6.6
6.7
6.5.1
Occupational values . . . . . . . . . . . . . . . . . . . . 141
6.5.2
Occupational Aspirations . . . . . . . . . . . . . . . . . 149
Educational Choices . . . . . . . . . . . . . . . . . . . . . . . 152
6.6.1
Choosing to stay on in different courses . . . . . . . . . 152
6.6.2
Further Qualifications . . . . . . . . . . . . . . . . . . 157
6.6.3
A Levels . . . . . . . . . . . . . . . . . . . . . . . . . . 158
6.6.4
Subject Choices . . . . . . . . . . . . . . . . . . . . . . 161
6.6.5
Gender and Subject Choice . . . . . . . . . . . . . . . 163
6.6.6
GNVQs . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
7 Conclusions
171
7.1
Cultural Reproduction . . . . . . . . . . . . . . . . . . . . . . 172
7.2
Rational Choice . . . . . . . . . . . . . . . . . . . . . . . . . . 173
7.3
Gender . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
7.4
Family Size and Structure . . . . . . . . . . . . . . . . . . . . 178
7.5
Suggestions for future research . . . . . . . . . . . . . . . . . . 179
7.6
General Theoretical Conclusions . . . . . . . . . . . . . . . . . 181
A Correlations
184
B Questionnaire
187
CONTENTS
C Educational Attitudes Scale
xiv
204
C.1 Reliability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
D Coding
207
List of Figures
2.1
A generic utility model . . . . . . . . . . . . . . . . . . . . . . 26
2.2
Murphy’s model . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.3
Breen and Goldthorpe’s model . . . . . . . . . . . . . . . . . . 39
3.1
Histogram of knowledge score . . . . . . . . . . . . . . . . . . 60
3.2
Histogram of passive vocabulary score . . . . . . . . . . . . . . 61
3.3
Histogram of active vocabulary score . . . . . . . . . . . . . . 62
3.4
Histogram of language score . . . . . . . . . . . . . . . . . . . 63
3.5
Histogram of reading cultural capital score . . . . . . . . . . . 65
4.1
Parents’ cultural participation by parents’ class . . . . . . . . 85
4.2
Parents’ cultural participation by parents’ educational level . . 87
xv
List of Tables
2.1
Perceived Utilities in Breen and Goldthorpe’s Model . . . . . . 40
3.1
Knowledge of cultural figures . . . . . . . . . . . . . . . . . . 59
3.2
Coding of Books . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.3
Pupils’ reading habits . . . . . . . . . . . . . . . . . . . . . . . 65
3.4
Coding of television programmes . . . . . . . . . . . . . . . . 67
3.5
Formal culture and music . . . . . . . . . . . . . . . . . . . . 69
3.6
Social class breakdown for each school . . . . . . . . . . . . . 78
4.1
Pupils’ cultural participation . . . . . . . . . . . . . . . . . . . 86
4.2
Language score . . . . . . . . . . . . . . . . . . . . . . . . . . 89
4.3
Cultural knowledge scores . . . . . . . . . . . . . . . . . . . . 92
4.4
Gender differences in cultural participation . . . . . . . . . . . 94
4.5
GCSE results . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.6
GCSE results using SOC class schema . . . . . . . . . . . . . 100
4.7
English GCSE results . . . . . . . . . . . . . . . . . . . . . . . 103
4.8
Maths GCSE results . . . . . . . . . . . . . . . . . . . . . . . 105
5.1
Pupils’ attitudes to education — mean scores . . . . . . . . . 117
xvi
LIST OF TABLES
xvii
5.2
Pupils’ attitudes to education — analysis of variance . . . . . 118
5.3
Pupils’ attitudes to education using bivariate parental class
and education . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.4
Pupils’ attitudes to education controlling for GCSE results . . 119
5.5
Pupils’ self-assessed ability . . . . . . . . . . . . . . . . . . . . 121
5.6
Pupils’ self-predicted grades . . . . . . . . . . . . . . . . . . . 122
5.7
Pupils’ self-predicted grades and actual grades . . . . . . . . . 122
5.8
Self-assessed ability by gender . . . . . . . . . . . . . . . . . . 123
5.9
Self-predicted grades by gender . . . . . . . . . . . . . . . . . 124
5.10 Pupils’ self-predicted grades and estimation of own abilities . . 124
6.1
Occupational aspirations by gender . . . . . . . . . . . . . . . 133
6.2
Occupational aspirations by parental class . . . . . . . . . . . 135
6.3
Occupational values by gender . . . . . . . . . . . . . . . . . . 136
6.4
Factor analysis of occupational values . . . . . . . . . . . . . . 138
6.5
Occupational value categories by gender . . . . . . . . . . . . 139
6.6
Occupational values by parental class . . . . . . . . . . . . . . 139
6.7
Models of desire for intellectual stimulation . . . . . . . . . . . 142
6.8
Material values . . . . . . . . . . . . . . . . . . . . . . . . . . 144
6.9
Control and respect . . . . . . . . . . . . . . . . . . . . . . . . 147
6.10 Job aspirations . . . . . . . . . . . . . . . . . . . . . . . . . . 150
6.11 Educational intentions and gender . . . . . . . . . . . . . . . . 153
6.12 Educational intentions and social class . . . . . . . . . . . . . 153
6.13 Reliability of A level choices . . . . . . . . . . . . . . . . . . . 154
6.14 Reliability of further education choices . . . . . . . . . . . . . 154
LIST OF TABLES
xviii
6.15 Pupils’ intention to puruse further education . . . . . . . . . . 155
6.16 Pupils’ intention to pursue A-levels . . . . . . . . . . . . . . . 159
6.17 A level subject choice by gender . . . . . . . . . . . . . . . . . 161
6.18 A level subject choice by class . . . . . . . . . . . . . . . . . . 162
6.19 GNVQ subject choice by gender . . . . . . . . . . . . . . . . . 163
6.20 Pupils’ choice of any hard science A levels . . . . . . . . . . . 164
6.21 Proportion of hard sciences
. . . . . . . . . . . . . . . . . . . 166
6.22 GNVQ subject choices . . . . . . . . . . . . . . . . . . . . . . 167
C.1 Attitudes to education scale . . . . . . . . . . . . . . . . . . . 205
C.2 Factor analysis of attitudes to education . . . . . . . . . . . . 206
Chapter 1
Bourdieu’s Theory of Cultural
Reproduction
1.1
Introduction
This chapter will examine Bourdieu’s theory of cultural reproduction, and
the use empirical researchers have made of this theory. Bourdieu’s work
must be seen in the context both of the debate on class inequalities in educational attainment and of broader questions of class reproduction in advanced
capitalist societies. Bourdieu is interested in the link between original class
membership and ultimate class membership, and in the way in which this
link is mediated by the educational system.
According to Bourdieu, the education systems of industrialised societies function in such a way as to legitimate class inequalities. Success in the education
system is facilitated by the possession of cultural capital and of higher-class
habitus. Lower class pupils do not in general possess these traits, so the
failure of the majority of these pupils is inevitable. This explains class inequalities in educational attainment. However, success and failure in the
educational system is seen as being due to individual gifts (or the lack of
them). Therefore, for Bourdieu, educational credentials help to reproduce
and legitimate social inequalities, as higher class individuals are seen to deserve their place in the social structure.
The first part of this chapter will consist of a general discussion of Bourdieu’s
theory of education, with particular reference to the concepts of cultural
1
CHAPTER 1. CULTURAL REPRODUCTION
2
capital and habitus. I will argue that the concept of habitus is theoretically
incoherent and has no possible use for empirical researchers. The concept
of cultural capital, on the other hand, while not constructed in the most
precise and lucid manner possible by Bourdieu, is substantive enough to be
potentially extremely useful to empirical researchers. The second section
of this chapter will therefore assess some of the empirical work concerning
cultural capital and the problems of operationalising the concept.
1.2
Bourdieu’s theory
1.2.1
Cultural Capital
Introduction to Cultural Capital
Bourdieu states that cultural capital consists of familiarity with the dominant culture in a society, and especially the ability to understand and use
“educated” language. The possession of cultural capital varies with social
class, yet the education system assumes the possession of cultural capital.
This makes it very difficult for lower class pupils to succeed in the education
system.
“By doing away with giving explicitly to everyone what it implicitly demands of everyone, the educational system demands of
everyone alike that they have what it does not give. This consists
mainly of linguistic and cultural competence and that relationship
of familiarity with culture which can only be produced by family
upbringing when it transmits the dominant culture.” (Bourdieu,
1977a, p. 494)
Bourdieu claims that, since the educational system presupposes the possession of cultural capital, which few students in fact possess, there is a great
deal of inefficiency in “pedagogic transmission”. This is because students
simply do not understand what their teachers are trying to get across. For
Bourdieu, this is particularly apparent in the universities, where students,
afraid of revealing the extent of their ignorance “. . . minimize the risks by
throwing a smoke-screen of vagueness over the possibility of truth or error.”
(Bourdieu and Passeron, 1990, p. 114)
3
But despite the fact that lower class pupils are seriously disadvantaged in
the competition for educational credentials, the results of this competition
are seen as meritocratic and therefore as legitimate. In addition, Bourdieu
claims that social inequalities are legitimated by the educational credentials
held by those in dominant positions. This means that the educational system
has a key role in maintaining the status quo.
“. . . it [education] is in fact one of the most effective means of
perpetuating the existing social pattern, as it both provides an
apparent justification for social inequalities and gives recognition
to the cultural heritage, that is, to a social gift treated as a natural
one.” (Bourdieu, 1974, p. 32)
In sum, Bourdieu’s view is that cultural capital is inculcated in the higher
class home, and enables higher class students to gain higher educational credentials than lower class students. This enables higher class individuals to
maintain their class position, and legitimates the dominant position which
higher class individuals typically goes on to hold. Of course, some lower class
individuals will succeed in the educational system, but, rather than challenging the system, this will strengthen it by contributing to the appearance of
meritocracy.
Bourdieu can be criticised for not being precise enough about exactly which
of the resources associated with the higher class home constitute cultural
capital, and how these resources are converted into educational credentials.
However, I think that the concept of cultural capital is substantive enough
to be operationalised, although Bourdieu does not make it at all obvious
how this should be done. The empirical evidence Bourdieu gives for the
link between class and cultural capital and cultural capital and educational
success is rather weak. Neither does Bourdieu give any evidence for the view
that educational credentials serve to legitimate class inequalities.
Bourdieu’s emphasis on the non-material resources possessed by the higherclass household is to be welcomed. We have evidence (see (Halsey et al.,
1980)) that the dramatic fall in the material costs to families of education due
to educational reforms, such as the universal provision of free and compulsory
secondary education, have not diminished the degree of association between
class origins and educational attainment. This suggests that the educational
advantage which higher class parents pass on to their children may not be
entirely caused by economic factors, and that the notion of cultural capital
is therefore worthy of serious attention. One does not have to accept all the
4
trappings of Bourdieu’s grand theory in order to acknowledge that cultural
factors may be important in generating inequalities in educational outcomes.
Cultural Capital vs. Other Forms of Capital
Bourdieu has been accused of giving too much weight to symbolic relations
at the expense of material ones (Willis, 1983). Yet Bourdieu refers to economic capital and social capital (social relationships and networks) as well
as symbolic and cultural capital (Bourdieu, 1989, p. 230). I will try to assess
how important Bourdieu thinks cultural capital is in relation to other forms
of capital. Bourdieu is characteristically unclear on this point.
“Apart from the fact that the increase in the proportion of holders
of the most prestigious academic qualifications among the ruling
classes may mean only the need to call upon academic approval
in order to legitimate the transmission of power and privileges is
being more and more felt, the effect is as though the cultural and
educational mechanisms had merely strengthened or taken over
from the traditional mechanisms such as the hereditary transmissions of economic capital, of a name or of capital in terms of
social relationships. . . ” (Bourdieu, 1977a, p. 496)
Have cultural and educational mechanisms “merely strengthened” traditional
mechanisms or reproduction or have they “taken over from” such traditional
mechanisms? Bourdieu slides from the former to the latter claim as if there
were not much to choose between them. And as if that was not vague enough,
Bourdieu is not actually claiming to describe reality, but merely states that
“the effect is as though” this claim were true. In fact, (the first part of
the sentence implies) educational credentials may not be necessary to secure
privileges at all, but only to legitimate them. In short this passage is quite incoherent. We are left with no clear idea of Bourdieu’s view of the importance
of cultural and educational capital in the transmission of privileges.
It is easy to find passages in Bourdieu in which the role of educational credentials in social reproduction is stressed. For instance,
“In ever more completely delegating the power of selection to the
academic institution, the privileged classes are able to appear to
5
be surrendering to a perfectly neutral authority the power of transmitting power from one generation to another. . . ” (Bourdieu and
Passeron, 1990, p. 167)
In this passage, the power of the educational system as a mechanism of
social selection is “ever more complete”. Bourdieu seems to be stating that
the importance of educational capital is supreme and increasing. However,
elsewhere Bourdieu admits the importance of inheriting economic capital,
and of social networks which are extremely valuable in the job market.
“The same academic qualifications receive very variable values
and functions according to the economic and social capital (particularly the capital of relationships inherited from the family) which
those who hold these qualifications have at their disposal. . . ” (Bourdieu, 1977a, p. 506)
Bourdieu goes further than this, stating that possessors of economic capital
are “able to do without” cultural capital, “. . . since academic qualifications
are a weak currency and possess all their value only within the limits of the
academic market.” (Bourdieu, 1977a, p. 507)
For cultural capital to be an important mechanism of social reproduction it
must be the case, not just that cultural capital facilitates the acquisition of
educational credentials, but that educational credentials are an important
mechanism by which wealth and power are transmitted. Bourdieu focuses
on the first of these relationships at the expense of the latter, and this may
account for the ambiguity in his views on the subject. But, on balance, I think
that although Bourdieu often stresses the importance of cultural capital, the
claim that he fails to recognise the importance of other forms of capital, and
the ultimate primacy of economic capital, is unfair. In particular, Bourdieu
clearly recognises that cultural capital forms a far more important part of
the strategy of social reproduction for some sections of the higher classes
(in particular, the professional classes), than for others (such as capitalists).
It must be acknowledged, though, that his statements on the subject are
inconsistent.
The Cultural Arbitrary
In addition to cultural capital, Bourdieu introduces the supplementary concept of the cultural arbitrary, which poses an additional obstacle to lower
6
class educational attainment. Bourdieu does not define the concept of the
cultural arbitrary. However, he states that:
“In any given social formation the legitimate PA 1 , i.e. the PA
endowed with the dominant legitimacy, is nothing other than the
arbitrary imposition of the dominant cultural arbitrary insofar
as it is misrecognized in its objective truth as the dominant PA
and the imposition of the dominant culture. . . ” (Bourdieu and
Passeron, 1990, p. 22)
Since Bourdieu uses the term “cultural arbitrary” without defining it, it
is not clear precisely what he means when referring to arbitrariness, or to
what extent he sees the cultural skills demanded and transmitted by the
educational system as arbitrary.
In some cases, the educational standards described by Bourdieu are clearly in
some sense arbitrary. For instance Bourdieu claims that lower class students
who achieve a degree of academic success by dint of hard work, face the
obstacle that their achievement may be deemed to be too hard won, and not
natural enough. In the educational system:
“. . . application becomes pedantry and a respect for hard work
grinding, limited pettiness, with the implication that it is intended
to compensate for lack of natural talents.” (Bourdieu, 1974, p.
59)
This aristocratic disdain for lower class attempts to appropriate higher class
culture leads to a peculiar set of values in higher education. Namely:
“. . . a tendency to prefer eloquence to truth, style to content.”
(Bourdieu, 1967, p. 335)
Bourdieu backs up this claim by reference to university examination reports
(Bourdieu and Saint-Martin, 1974). Bourdieu claims that the criteria of
university examiners reflect the values of the dominant classes, and that the
more vague the demands of the examiners are, the less chance lower class
pupils will have of adhering to these demands.
1
Pedagogic Action
7
Bourdieu’s comments on the theme of academic values are plausible. But
one must ask how important the cultural arbitrary is in contributing to class
inequalities in educational attainment. Although Bourdieu’s argument is
rather compelling in relation to the evaluation of work in the arts and humanities departments of universities, it does not have the same force when
applied to the sciences or to primary and secondary schools. At least Bourdieu provides no evidence that “eloquence” is preferred to “truth” in university physics exams. In addition, it seems to me that the National exams
taken by school children (at least in this country) are examined using clear
and explicit criteria. This problem reflects a general tendency of Bourdieu’s
to focus on universities rather than on schools. This can only detract from
his arguments since it means that Bourdieu is dealing with a population from
which the lower classes have already been largely eliminated.
I am not suggesting that Bourdieu sees every element of the cultural capital
transmitted in the home and the education transmitted in the school as
arbitrary. For instance Bourdieu notes that:
“. . . logical and symbolic mastery of abstract operations and, more
precisely, mastery of the laws of transformation of complex structures [are] a function of the type of practical mastery of language
and the type of language acquired in the home. . . ” (Bourdieu and
Passeron, 1990, p. 58)
Presumably the skill of abstraction, and the complex language which it requires, is not “arbitrary” simply because it is associated with the higher
classes.
So, how do we decide which educational values and practices are arbitrary,
and which valid? Bourdieu states that:
“The sociological theory of PA distinguishes between the arbitrariness of the imposition and the arbitrariness of the content
imposed, only so as to bring out the sociological implications of
the relationship between two logical fictions, namely a pure power
relationship as the objective truth of the imposition and a totally arbitrary culture as the objective truth of the meanings imposed. . . There is no PA which does not inculcate some meanings
not deducible from a universal principle (logical reason or biological nature. . . )” (Bourdieu and Passeron, 1990, p. 9–10)
8
In relation to the content of teaching, “arbitrariness” is opposed to “objective truth” and “meanings deducible from a universal principle”. This is
confusing, since “objective truth” and “meanings deducible from a universal
principle” are not the same thing. If anything which is not objectively true
is therefore “arbitrary”, does this mean that subjects (e.g. the humanities)
where we can never be confident of objective truth are always utterly arbitrary? Or is the aim of truth enough to justify a discipline? What about
subjects such as music or woodwork which do not aim at truth?
Bourdieu’s notion of the “cultural arbitrary” is unclear. It is not possible
to determine to what extent he is arguing that the dominant culture and
the educational values that serve it are no better than any other culture. Is
there an element of cultural relativism in his argument? It is very important
to distinguish between the view that many lower class pupils may find highbrow literature (for instance) hard to understand because of the distance
between ordinary language and literary language, and the view that highbrow literature is irrelevant to working class pupils because they come from
a different culture. The latter view is explicitly based on the seemingly egalitarian notion that all cultures are ‘equal but different’, but implicitly, it is
based on the insulting view that lower class children are inevitably incapable
of appreciating any aspect of highbrow culture.
Bourdieu does not make a clear enough distinction between those parts of
the dominant culture which are in some way snobbish (i.e. exclusive for
exclusivity’s sake) and arbitrary, and those which are universally valuable
but not universally accessible.
So, Bourdieu gives some interesting examples of arbitrary values in education,
but does not give a precise definition of what constitutes arbitrariness in this
context. Such a definition is essential if we are to distinguish between those
elements of the dominant culture which should be taught in schools, and those
which should be removed from the curriculum. Furthermore, it seems clear
that lower class pupils would be disadvantaged by a lack of cultural resources
even if the content of educational syllabuses and assessments were utterly
rational. A sophisticated grasp of language alone would be a huge advantage
in just about any conceivable educational system. Given this, the cultural
arbitrary should surely be relegated to a minor role in any explanatory theory
of class inequalities in educational attainment.
1.2.2
9
Habitus
Introduction to Habitus
The notion of habitus is central to Bourdieu’s thought, yet it is never clearly
defined. I will try to elucidate the concept, before going on to criticise it.
Like cultural capital, habitus is transmitted within the home. However,
whereas cultural capital consists of the possession of legitimate knowledge,
habitus is a set of attitudes and values, and the dominant habitus is a set of
attitudes and values held by the higher classes.
“In fact, each family transmits to its children, indirectly rather
than directly, a certain cultural capital and a certain ethos2 . The
latter is a system of implicit and deeply interiorized values which,
among other things, helps to define attitudes towards the cultural
capital and educational institutions. The cultural heritage, which
differs from both points of view according to social class, is the
cause of the initial inequality of children when faced with examinations and tests, and hence of unequal achievement.” (Bourdieu,
1974, p. 43)
So, habitus works with cultural capital to the disadvantage of lower class
students. A major components of the dominant habitus is a positive attitude
towards education.
“. . . the system of dispositions towards the school, understood as
a propensity to consent to the investments in time, effort and
money necessary to conserve and to increase cultural capital.”
(Bourdieu, 1977a, p. 495)
So is habitus just a set of attitudes, directed primarily towards education
and culture? Sometimes Bourdieu seems to suggest that the dominant habitus consists of more than this — that it includes (or at least gives rise to)
competence in specific social settings, including for instance:
“. . . the practice of the games and sports of high society or the
manners and tastes resulting from good breeding. . . ” (Bourdieu,
1977a, p. 506)
2
Ethos is an element of habitus.
10
So what does Bourdieu mean by a “set of dispositions”? His definition in
(Bourdieu, 1977b, p. 214) is unhelpful. He states that the word disposition
encompasses three meanings:
1. “the result of organising an action”.
2. a “way of being” or “habitual state”.
3. a “tendency”, “propensity” or “inclination”.
If Bourdieu insists on using the word habitus to signify three such different
things, it would be helpful if some attempt was made to indicate to the reader
which sense is intended each time the term is used. Needless to say, Bourdieu
does not do this. No wonder then that the concept of habitus is condemned
by Nash as “ambiguous and overloaded” (Nash, 1990, p. 446). Unlike cultural
capital, there has rarely been any attempt by empirical researchers to use
the concept of habitus. This is because the concept is too nebulous to be
operationalised.
Structure and Agency: the Role of Habitus
Given the messiness of the concept of habitus, one might ask why Bourdieu
introduces it into his theory at all. The answer is that Bourdieu thinks that
the concept of habitus solves a fundamental problem in sociology — the
conflict between structure and agency.
Bourdieu attacks crude structuralism on the grounds that “certain structuralists” see “agents as the simple ‘supports’ of structures invested with
the mysterious power of determining other structures.” (Bourdieu, 1977a, p.
487)
However, Bourdieu also criticises methodological individualism. Certain
“atomistic” mobility researchers are singled out for attack on the grounds
that they do not recognise that social mobility can coexist with stable class
structures. Bourdieu protects himself here by failing to name the researchers
he is referring to, and I cannot think of any mobility researcher who fails
to recognise this simple point. In fact, Bourdieu’s comment reveals a crude
view of methodological individualism (and perhaps also of structuralism) neither of which he defines, of course, despite the fact that they mean so many
different things to different people.
11
According to Bourdieu, if we wish to avoid the dichotomy between individualism and structuralism:
“This means that our object becomes the production of the habitus, that system of dispositions which acts as mediation between
structures and practice; more specifically, it becomes necessary to
study the laws that determine the tendency of structures to reproduce themselves by producing agents endowed with the system of
predispositions which is capable of engendering practices adapted
to the structures and thereby contributing to the reproduction of
the structures” (Bourdieu, 1977a, p. 487)
Following (though not acknowledging) Boudon, Bourdieu notes that working
class students are more likely to drop out of the educational system than
higher class students, even if we control for previous achievement. He claims
that this is a more important mechanism of selection than exam failure.
“Thus, previous performances being equal, pupils of working-class
origin are more likely to ‘eliminate themselves’ from secondary
education by declining to enter it than to eliminate themselves
once they have entered, and a fortiori more likely not to enter than
to be eliminated from it by the explicit sanction of examination
failure.” (Bourdieu and Passeron, 1990, p. 153)
Bourdieu claims that this phenomenon can be explained in terms of the lower
class habitus. The habitus is in some way formed by the objective chances
of success shared by the class. The habitus in turn determines the actions of
the members of the class.
“. . . the principle underlying the production of the most durable
academic and social differences, the habitus — the generative
unifying principle of conducts and opinions which is also their
explanatory principle, since at every moment of an educational or
intellectual biography it tends to reproduce the system of objective
conditions of which it is the product.” (Bourdieu and Passeron,
1990, p. 161)
But if attitudes are due to a particular habitus, how does this habitus form,
and how is it absorbed by individuals? Bourdieu is characteristically vague:
12
“. . . everything happens as if parental attitudes towards their children’s education. . . were primarily the interiorization of the fate
objectively allotted (and statistically quantifiable) as a whole to
the social category to which they belong.” (Bourdieu, 1974, p. 33)
Again we have the non-committal phrase “everything happens as if”. Bourdieu is not suggesting that parents consult statistics on class differences in
survival rates in education before making decisions on their children’s education. So how exactly is their “fate” “interiorized”? The process becomes
more mysterious the deeper one probes:
“. . . the negative predispositions towards the school which result
in the self-elimination of most children from the most culturally unfavoured classes and sections of a class — such as selfdepreciation, devaluation of the school and its sanctions, or a resigned attitude to failure and exclusion — must be understood as
an anticipation, based upon the unconscious estimation of the objective probabilities of success possessed by the whole category, of
the sanctions objectively reserved by the school for those classes
or sections of a class deprived of cultural capital.” (Bourdieu,
1977a, p. 495)
The objections to this are obvious. Firstly, How can an estimation be unconscious? If habitus is not generated by conscious individuals, where does it
come from? Secondly, even if an individual knows the objective probabilities
of success possessed by the whole category, why does she not recognise that,
by changing her attitude to the educational system, she can escape the fate
of the rest of her category?
Thirdly, Bourdieu seems to be arguing that people’s behaviour is the result of
accepting the “objective probabilities” of future success. However, as Jenkins
points out, “Something which happens at time ‘x’ cannot be accounted for by
the likely state of affairs — as predicted by statistics — at the time ‘x + 1’.”
(Jenkins, 1992, p. 81) Expectations about the future must be based upon the
present. The actions based on these expectations create social reality, rather
than “objective probabilities” creating expectations which lead to action.
It might be argued that it is uncharitable to interpret Bourdieu as putting
forward an explanation of current events in terms of future events. But even
if we interpret Bourdieu more kindly as arguing that lower class pupils do
13
not pursue demanding educational options because they are aware of the
current tendency of the class as a whole not to pursue such options, it must
be admitted that this is a feeble explanation. If we were happy to accept
explanations of the characteristics of individuals which simply refer us to the
characteristics of the group of which these individuals are members, without
explaining these characteristics, there would be little need for sociology.
In sum, the notion of habitus utterly fails in Bourdieu’s stated purpose of
avoiding both structuralist determinism and “atomism”. The notion of habitus is completely deterministic leaving no place for individual agency or even
individual consciousness.
Yet Bourdieu denies the charge of determinism on three counts. Firstly, the
same habitus will produce different practices in different social fields. Secondly, the habitus can be changed by changed circumstances. Thirdly, the
habitus may be controlled by the “awakening of consciousness and socioanalysis”(Bourdieu, 1990, p. 116).
The first two points depend on forces external to the social group and are
therefore no defence against determinism. The speciousness of these arguments can be illustrated by the fact that the same charge will produce
different motion in different electric fields - which hardly shows that electromagnetism has a role for individual freedom. The third point is rather vague
- Bourdieu does not suggest how we are supposed to awaken our (presumably
slumbering) consciousnesses.
So, habitus is a vague concept, which fails to fulfil the role that Bourdieu
claims for it within his theory. Bourdieu may be right that the attitudes towards education associated with different social classes affect the tendency of
individuals within those classes to pursue education. But this is barely worth
saying if we cannot be any more precise. So the concept of habitus adds nothing to Bourdieu’s analysis of class inequalities in educational attainment. As
for the claim that the concept of habitus transcends the dichotomy between
structure and agency, this is totally unjustified.
1.2.3
Political Action
A criticism which is linked to Bourdieu’s determinism is the view that Bourdieu does not allow for working class cultural production, and hence has
nothing to say about a radical politics of education. For instance, Giroux
14
states that Bourdieu’s theory:
“. . . is a theory of reproduction that displays no faith in subordinate classes and groups, no hope in their ability or willingness
to reinvent and reconstruct the conditions under which they work
and learn.” (Giroux, 1982, )
One could simply retort that sociology should never be driven by a demand
for “faith” and “hope” — that is the role of dubious metaphysics. However, it
is fair to ask whether Bourdieu suggests any way of improving the education
system.
In fact, Bourdieu does propose an education system which would do justice to
lower class pupils — namely “universal pedagogy”. This form of pedagogy:
“. . . would take nothing for granted initially, would not count as
acquired what some, and only some of the pupils in question had
inherited, would do all things for all and would be organised with
the explicit aim of providing for all with the means of acquiring
that which, although apparently a natural gift, is only given to the
children of the educated classes. . . ” (Bourdieu, 1974, p. 38)
This means that cultural capital would be taught explicitly in the school,
rather than simply transmitted within the higher class home. However, Bourdieu does not suggest that “universal pedagogy” could be implemented now,
since:
“. . . only a school system serving another system of external functions and, correlatively, another state of the balance of power
between the classes, could make such pedagogic action possible”
(Bourdieu and Passeron, 1990, p. 127)
In my view Bourdieu is right to argue that the education system must be seen
in terms of a wider system of social relations, and that changes to the education system alone cannot transform those relations. However, while Giroux
exaggerates the autonomy possessed by the educational system, Bourdieu
perhaps underplays it. Although Bourdieu is right to argue that the possibilities for educational reform, and for the effectiveness of such reforms are
15
limited by the balance of power between the classes, this does not mean that
the idea of universal pedagogy has no implications for educational reformers.
If the lack of cultural capital is an important cause of lower class educational
disadvantage, then this should be taken on board by schools. In particular,
schools should put more emphasis on the explicit teaching, not just of basic
literacy, but of a sophisticated grasp of language. If Bourdieu’s theory were
correct, this would certainly increase the absolute level of attainment of lower
class pupils (a worthwhile aim in itself) and might well have some success in
eroding class inequalities in attainment.
1.2.4
Science and Language
I have pointed out that Bourdieu fails to express his theory clearly. In my
view, this failure is bound up with Bourdieu’s rejection of what he describes
as a “. . . positivist conception of science. . . ”(Bourdieu, 1990, p. 19–20). Of
course, Bourdieu does not define what he means by positivism. Instead,
he uses the common ploy of denouncing all research that attempts to test
hypotheses empirically as positivist without actually saying what he thinks
is wrong with this type of methodology. The rejection of the importance of
deriving hypotheses from a theory and attempting to test these hypotheses
allows Bourdieu to be unapologetic in the use of poorly defined concepts.
“Especially in the Anglo–Saxon tradition, people criticise the researcher for using concepts that function as signposts pointing
to phenomena that are worth examining but that often remain
obscure and vague, even if they are suggestive and evocative.”
(Bourdieu, 1990, p. 40)
Against such Anglo–Saxon criticisms, Bourdieu asserts that, because the
social world is complex, theories about it must be complicated, and must be
expressed in complicated language.
“I think that, literary and stylistic qualities apart, what Spitzer
says about Proust’s style is something I could say about my own
writing. He says, firstly, that what is complex can only be said in a
complex way; secondly, that reality is not only complex, but also
structured. . . if you want to hold the world in all its complexity
and at the same time order and articulate it. . . you have to use
16
heavily articulated sentences that can be practically reconstructed
like Latin sentences. . . ” (Bourdieu, 1990, p. 51–52)
The absurdity of this argument is easily shown. Firstly, the aim of science
is not to “hold the world in all its complexity”, and the history of science
tells us that a simple theory will be preferred to a more complicated theory
if the simpler theory has equal or superior predictive power, e.g. Copernicus’
defeat of Ptolemaic astronomy. Furthermore, it simply is not true that a
difficult concept or theory must be expressed in difficult language.
In my view, the real purpose served by the obscurity of Bourdieu’s prose is
to protect his own work from refutation. Bourdieu’s strategy in dealing with
criticism is to claim that his critics have not understood his work, and to
imply that his critics are just jealous because they are not as clever as him.
“. . . they criticise not my analyses, but an already simplified, if
not maimed, representation of my analyses. This is because they
invariably apply to them the very modes of thought, and especially
distinctions, alternatives and oppositions, which my analyses are
aimed at destroying and overcoming.” (Bourdieu, 1990, p. 107)
The point that the critic may not agree that Bourdieu has succeeded in
destroying such oppositions is ignored by Bourdieu, who never deals with
specific criticisms in a direct way.
When a criticism is made of Bourdieu, the explanation for this is always to
be found in the inadequacies of the critic. So, behind “positivist methodology” lies an “epistemology of resentment” which allows its advocates to
“prohibit others from doing what they themselves are unable to do, so that
they can impose their own limits on others.”(Bourdieu, 1990, p. 35) “Positivism” is simply “. . . a tradition often appealed to by the most mediocre of
researchers in order to “pare the lion cubs’ claws”, as Plato put it — in other
words, to disparage and reduce the creations and innovations of the scientific
imagination.” (Bourdieu, 1990, p. 40)
So, although Bourdieu declares a “headlong, rather crazy commitment to
science” (Bourdieu, 1990, p. 26), his rejection of scientific values is made
plain. Furthermore, his impenetrable prose style should not be seen simply
as an irritation for the reader, but rather as being closely bound up with
this rejection of scientific values, since clarity makes a theory amenable to
testing, whereas obscurity protects it from falsification.
1.3
17
Empirical Evidence on Cultural Capital
This section consists of a survey of some of the empirical work relating to
Bourdieu’s theory. Most of this work focuses on the link between cultural
capital and educational attainment. I will conclude that the evidence is
mixed, largely due to widely varying operationalisations of cultural capital.
Evidence on the link between educational attainment and social reproduction
and mobility will also be examined. There has been less focus on this part
of Bourdieu’s theory, but such evidence as there is suggests that educational
capital is more a vehicle of social mobility than of social reproduction.
1.3.1
Bourdieu’s Own Evidence
Bourdieu is adamant that he does not engage in theory for its own sake, and
that empirical work is central to his enterprise.
“Let me say outright and very forcefully that I never ‘theorise’,
if by that we mean engage in the kind of conceptual gobbledegook. . . that is good for textbooks and which, through an extraordinary misconstrual of the logic of science, passes for Theory in
much of Anglo-American social science. . . There is no doubt a theory in my work, or, better, a set of thinking tools visible through
the results they yield, but it is not built as such. . . It is a temporary construct which takes shape for and by empirical work.”
(Waquant, 1989, p. 50)
Unfortunately, the claim that Bourdieu’s theoretical framework is subordinate to the needs of empirical research is not backed by the evidence provided
by Bourdieu regarding cultural reproduction.
For Bourdieu’s theory to be backed empirically, he would need to show that:
1. Parental cultural capital is inherited by children.
2. Children’s cultural capital is converted into educational credentials.
3. Educational credentials are a major mechanism of social reproduction
in advanced capitalist societies.
18
Of course, Bourdieu does not deny that privilege can be inherited through
means other than the acquisition of educational credentials. Inheritance of
property, and occupational advantage gained through social networks are
obvious examples of this. So, Bourdieu’s theory is not refuted by empirical
evidence that there is no absolute correspondence between credentials and
occupational outcomes (see for instance Dale and Pires 1984). However, it
is crucial to Bourdieu’s theory that cultural capital actually does facilitate
educational success, and that educational success actually is associated with
occupational advantage, even if this is only a means of legitimating class
inequalities.
Bourdieu claims that (1) and (2) are shown:
“. . . by the fact that, among the pupils of the grandes écoles, a very
pronounced correlation may be observed between academic success
and the family’s cultural capital measured by the academic level of
the forbears over two generations on both sides of the family. . . ”
(Bourdieu, 1977a, p. 497)
Bourdieu is not entitled to assume that a high parental level of education
reveals a high level of parental cultural capital. In fact, Bourdieu’s use of
parental educational credentials as a measure of cultural capital begs the
question of whether educational credentials simply constitute “. . . embodied
cultural capital that has received school sanctioning.” (Bourdieu and Boltanski, 1981, p. 145). In addition, the use of bivariate analyses is crude. Clearly,
a simple association between two variables is not convincing evidence of a
causal relationship. Bourdieu fails to show that parental cultural capital
is inherited by the children, and that this is the mechanism through which
higher class pupils tend to attain higher educational credentials than lower
class pupils. His evidence is quite consistent with educational privilege being passed down through mechanisms other than cultural capital, such as
parental encouragement and material resources.
Bourdieu also presents evidence that both social class and educational attainment are strongly associated with participation in cultural activities such as
book reading and buying, and cinema, theatre, concert and museum attendance. (Bourdieu and Boltanski, 1981, p. 490–2). However, on their own,
these figures do not really back up Bourdieu’s theory. They do not constitute
evidence that participation in cultural activities is the mechanism by which
middle class parents ensure good qualifications for their children.
19
In sum, Bourdieu assumes much of what he sets out to prove. It is circular
to treat educational level as a proxy for cultural capital if one is trying to
assess whether cultural capital does in fact help to determine the educational
levels reached by individuals.
1.3.2
Other Research on Cultural Reproduction
Bourdieu’s concept of cultural capital is not clearly defined, and it is not
particularly surprising that it has been operationalised in various different
ways. I have argued that Bourdieu’s own operationalisation of the concept is
quite inadequate. Yet Bourdieu is not the only author to use parental education as a proxy for cultural capital. For instance, Halsey, Heath and Ridge
(Halsey et al., 1980) use this proxy, as do Robinson and Garnier (Robinson
and Garnier, 1985) and Jan o. Jonsson (Jonsson, 1987).
Since Bourdieu’s definition of cultural capital is not precise, it is not clear
what an “authentic” operationalisation would consist of. However, Bourdieu
does explicitly state the importance of linguistic competence. Cultural “competence” and “familiarity” can reasonably be interpreted as knowledge of and
participation in the dominant culture. Despite this, previous investigations
of cultural capital have not included data on linguistic ability, and DiMaggio
and Mohr (DiMaggio, 1982), (DiMaggio and Mohr, 1985) and (Mohr and
DiMaggio, 1995) are unusual in using data on cultural knowledge. Data
on cultural activities other than reading has often tended towards highly
exclusive activities such as gallery attendance, which are foreign to a large
proportion even of the middle and upper classes. For example, De Graaf (De
Graaf, 1986) uses a measure of the number of visits per month to museums,
galleries, concerts, theatres and historical buildings. In general, surveys include data on either pupils’ or parents’ cultural participation, but not both.
Most commonly, the proxy of parental education is used instead of data on
parental cultural capital, although this proxy clearly begs the question of
whether occupational status and educational attainment actually do reflect
the possession of culturalcapital. Given that researchers have operationalised
the concept of cultural capital in different ways, it is not surprising that empirical studies of the effect of cultural capital on educational attainment have
varied in their conclusions. As well as those already mentioned, note (Crook,
1997), (Egerton, 1997), (Graetz, 1988), (Kalmijn and Kraaykamp, 1996),
(Katsillis and Rubinson, 1990) and (Savage and Egerton, 1997).
Which cultural attributes should be seen as constituting capital cannot be
20
determined without empirical investigation, since the term cultural capital
implies an analogy with economic capital, and therefore, a return. The return
on cultural capital takes the form of educational credentials and, ultimately,
occupational success. Therefore, in subsequent chapters, I will develop a
broad operationalisation of cultural capital in order to examine which elements actually yield returns in the sense of contributing to educational
success.
If participation in cultural activities does lead to academic success, one may
ask why this should be. It may be suggested that the culture of the school
reflects the dominant culture. This could occur if teachers are prejudiced
in favour of pupils who display “cultured” traits, and therefore give them
higher grades (Farkas et al., 1990). This view is perhaps most relevant in
the US, where grades awarded by teachers are an important outcome of
schooling. It is a less plausible explanation in nations such as Britain, where
the key outcome of schooling is the results gained in national examinations.
Alternatively, the dominant culture could be ingrained in the curriculum.
However, it has been pointed out that, although this may be true of France,
there is little emphasis on highbrow culture in schools in countries such as
Britain, the Netherlands, and the US (De Graaf et al., 2000).
An alternative explanation is that participation in cultural activities leads to
the development of knowledge or skills, which in turn enable pupils to succeed
at school. For instance, one might expect reading novels to contribute to both
linguistic competence and cultural knowledge. Crook (Crook, 1997) and de
Graaf et. al. (De Graaf et al., 2000) follow de Graaf (De Graaf, 1986; De
Graaf, 1988) in breaking cultural capital into two constituent parts, reading
and beaux arts participation. Beaux arts participation refers to participation
in formal cultural activities outside the home, such as gallery, theatre and
concert attendance. Both Crook (Crook, 1997) and de Graaf et. al. (De
Graaf et al., 2000) find that reading is associated with academic success
whereas beaux arts participation is not, and infer from this that the effect of
cultural capital on educational attainment is due to the “educative resources”
such as analytic and cognitive skills which are developed by reading, rather
than to the communication of status via participation in formal culture.
However, this inference may be questioned, since one could argue that participation in beaux arts may contribute to the development of skills and
knowledge, or that pupils’ reading is as likely to prejudice teachers in their
favour as is participation in other cultural activities. Therefore, as a further
test of this hypothesis, it will be useful to test pupils on the sorts of abilities
21
and knowledge that may be developed through cultural participation, in order to see whether these skills are in fact the means through which cultural
participation promotes educational success.
In sum, many researchers examining cultural capital have used what data
was available to them, even though this data has not been ideally suited to
the purpose. In my view, it is far preferable to begin with an exploration of
the theory of cultural reproduction, and of the mechanisms through which
cultural capital may operate, in order to develop a sound operationalisation
of the concept of cultural capital, and then to collect appropriate data.
1.4
Conclusions
Bourdieu’s project is extremely ambitious, and I have argued that, on many
scores, it is a failure. I have stated that habitus is an unhelpful concept. It
has some intuitive plausibility, but is at once too all-inclusive and too vacuous
to be of any use to empirical researchers. I have also argued that Bourdieu’s
claim that the notion of habitus solves the conflict between structure and
determinism on the one hand and agency and individualism on the other
is quite unjustified. In fact Bourdieu’s theory has no place not only for
individual agency, but even for individual consciousness.
Even in the area where I have been most sympathetic, namely the concept
of cultural capital, I have argued that the concept is not clearly defined. In
addition, I argued that the related concept of the cultural arbitrary is of very
limited usefulness. Bourdieu has some valuable insights into arbitrary practices in higher education. However, he does not distinguish clearly enough
between standards which are prejudicial to lower class pupils and students
because they are arbitrary, and standards which are prejudicial to lower class
pupils and students because they do not have the resources to meet those
standards.
Bourdieu’s theory is vaguely expressed throughout. One might even accuse
him of what he accuses undergraduates — a tendency to “. . . minimise the
risks by throwing a smoke-screen of vagueness over the possibility of truth
or error.”
However, I have defended Bourdieu against what I took to be unfair criticisms. Bourdieu has been accused of emphasising the importance of cultural
capital at the expense of economic capital. Although it is easy to see how
22
one would get this impression from some passages in Bourdieu, on balance I
think that Bourdieu is aware of the prime importance of economic capital
Bourdieu has also been accused of not having anything to say about a radical
politics of education. This is true in that Bourdieu does not see the education system as independent from the rest of society, and therefore sees the
transformative power of radical educational policies as rather limited. However, I believe that Bourdieu is, to some extent, correct in his pessimism.
The education system has a degree of autonomy, but critics of Bourdieu such
as Giroux exaggerate this autonomy somewhat naively. However, Bourdieu
may be accused of understating the autonomy of the educational system, and
of ignoring the policy implications of the theory of cultural capital.
But my main defence of Bourdieu is that despite the huge flaws in the grand
structure of his theory, some of his ideas have the potential to be very fruitful. In particular, the concept of cultural capital is one that has inspired
empirical researchers. However, many of the empirical findings on cultural
capital contradict one another. This may be partly due to the fact that these
studies were carried out at different times in different countries. It may be
that cultural capital is more important in some countries than in others, or
operates differently in different countries at different times. However, in my
view the main reason for the variable findings presented here is the different
methodologies used in each study, and in particular, the array of different
operationalisations of cultural capital that are used.
Bourdieu himself uses parental educational attainment as a measure of cultural capital. In my view this is a crude indicator of the concept, and little
improvement is made on it by most of the studies cited here. Bourdieu describes cultural capital as a set of activities and competencies associated with
the higher classes. From the examples Bourdieu gives, these seem to include
participation in high culture, but also a sophisticated grasp of language, and
specific cultural skills and knowledge. In order to test the link between cultural capital and academic attainment thoroughly, these different elements
of cultural capital should be distinguished, in order to find which (if any)
elements of cultural capital are important.
The link between educational attainment and occupation is not examined
empirically by Bourdieu, and he is not consistent in the strength of the role
he wishes to claim for educational credentials in maintaining social reproduction. Given the evidence showing that the association between class origins
and destinations persists even when educational credentials are controlled
for (Marshall et al., 1997), an overriding role should not be claimed for ed-
23
ucational credentials. Evidence from France (Robinson and Garnier, 1985)
suggests that educational capital is a mechanism for social mobility rather
than social reproduction. Therefore, Bourdieu may have overstated the role
of educational credentials in social reproduction, at least in his stronger statements on this subject.
The work that I will go on to describe in subsequent chapters uses measures
of as broad a range of cultural resources associated with the higher class
home as possible. I will use measures of both parents’ and pupils’ cultural
participation, and of pupils’ linguistic ability and cultural knowledge in order to examine whether any element of cultural capital is important in the
transmission of educational advantage.
Chapter 2
Rational Choice Perspectives
Rational choice theory aims to explain human behaviour on the basis that
people act rationally to achieve their aims in the light of their beliefs about
the situation. This of course leaves much scope for disagreement among
rational choice theorists. In this chapter, I will examine the application of
rational choice theory to the question of inequalities (it must be acknowledged
that the inequalities addressed are largely class inequalities) in educational
attainment, with reference to the work of Murphy, Boudon, Gambetta and
Breen and Goldthorpe.
Breen and Goldthorpe (Breen and Goldthorpe, 1997) give a formalisation of
their model. This technique is useful in highlighting the key variables and
their effects. I will describe a generic “utility” model, which I will argue
covers the models of decision making about post compulsory education of all
the authors mentioned here (with the partial exception of Murphy). I will
go on to discuss Breen and Goldthorpe’s formalisation in detail. I will argue
that Breen and Goldthorpe fail in their stated aim of giving an explanation
that does not depend on social class differences in preferences, although they
give a common mechanism for the formation of such preferences.
2.1
The Utility Model
Breen and Goldthorpe give a formalisation of their model. I will now describe
a basic “utility” model, which I will use to analyse the work of all the authors mentioned here, and in particular, to discuss Breen and Goldthorpe’s
24
CHAPTER 2. RATIONAL CHOICE PERSPECTIVES
25
formalisation in detail. Let me stress that this model is only intended as
an aid to examining other models of educational decision making, not as an
explanatory model in its own right.
The utility model is a model of the decision to stay on in education after
having already achieved a certain level of education. For the sake of argument
let us say it is the decision to stay on in the sixth form having passed GCSE’s.
Suppose that each individual attaches a “utility” to each of the following
possible educational outcomes:
1. Staying on in education and “passing” (i.e. achieving some minimum
level chosen by the researcher, not the individual).
2. Staying on in education and “failing”.
3. Leaving education.
By utility I mean the value of the educational outcome in life after education
(in the job market — but also possibly in the marriage market or for other
social or individual purposes). The utility of the outcome should be contrasted with the “cost” of education, by which I mean costs incurred in the
course of trying to gain the qualification, set off against any benefits of doing
the course (as opposed to benefits of success in gaining the qualification), for
example, direct financial costs, earnings forgone, like or dislike of the course,
approbation/disapproval of peers and family etc.).
Let us denote the utilities of each of the possible outcomes by U1 , U2 and
U3 respectively, and denote the cost of education by C. Suppose that the
individual believes that they would have probability π of passing were they
to stay on in education. This can be represented diagrammatically as shown
in figure 2.1.
It should be noted that these utilities, costs and probabilities are subjective
— i.e. they are those perceived by the individual.
It is reasonable to say therefore that the expected net utility (utility − costs)
if an individual chooses to stay on is given by:
(expected utility if they pass)(probability of them passing)+
(expected utility if they fail)(probability of them failing)−(costs of education)
26
Pass .r
..... U
..........
1
..........
..........
......
..........
..........
..........
..........
.
.
.
.
.
.
.
.
.
....
..........
...........
.......... ...................
..........
..........
..........
..........
.
.
.
.
.
.
.
.
.
..........
..
..........
..........
..........
..........
.
.
.
.
.
.
.
.
..........
.
.......
.
.
..........
.
.
.
.
.
.
.
..........
.......
.
.
.
.
.
.
..........
.
.
.
.
...............
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
......
Stay on
r
r
Fail
r U2
rU
Leave
3
Figure 2.1: A generic utility model
which in turn is given by:
π(U1 ) + (1 − π)(U2 )C = π(U1 − U2 ) + U2 − C
whereas the expected net utility if they choose to leave is given by
U3 .
Thus if one introduces a parameter p given by
p = (expected utility of staying on) − (expected utility of leaving)
we can say that:
p = π(U1 − U2 ) + U2 − C − U3 .
p is interpreted as follows: if p is greater than zero, the individual expects
more net utility from staying on than from leaving education; if p is less than
zero, the reverse is true. Note that this is as far as one should attribute any
meaning to p. It is only its difference from zero that is relevant and not its
size. Thus we could equally well have defined p by:
(expected utility of staying on)
expected utility of staying on) + (expected utility of leaving)
and then interpreted p as follows: if p is greater than one half, the individual
expects more net utility from staying on than from leaving; if p is less than
one half, the reverse is true. Of course, if one uses this alternative definition
of p the results obtained will be the same — however, the formulae will
27
be superficially more complex. Therefore, I would like to use the original
formula for p, but in deriving Breen and Goldthorpe’s equations I must use
the second formula. Nevertheless, the differences are purely superficial. 1
To complete the model one now assumes that if an individual has p greater
than 0 they will stay in education and if p is less than zero they will leave
education.
As it stands, this model, of course, makes no predictions. One would have to
describe how the various parameters vary across class, gender etc. in order to
make predictions. However, one can observe how these variables must vary
in order to account for social inequalities.
If we make the reasonable assumption that for all individuals U1 > U2 (i.e.
that there is greater utility in passing than failing) then it must be true that
for some individuals U2 − C − U3 is negative. Otherwise p will always be
positive and so it will be in no-one’s interest to leave education. In words,
if one refers to the difference between the utility of staying on and failing
and the utility of leaving as the guaranteed utility of education then it must
be the case for some people that the costs of education are greater than the
guaranteed utility of education. (Of course, if the costs of education are
positive and the guaranteed utility of education is negative then this will
automatically be the case.) One can therefore describe the primary problem
for rational choice theorists as explaining why the costs of education outweigh
the guaranteed utility of education for working class pupils more often than
is the case for middle class pupils.
(Note however, that although the above condition is necessary, it is not sufficient to explain class differentials. It would be possible for the expectation
1
The name p for this variable is best thought of as standing for “parameter”. I call
it p because Breen and Goldthorpe do. However, with their definition, p always takes
values between zero and one. Given, on top of this, its name it is tempting to think of p
as representing some probability — perhaps the probability of the student staying on in
education. However, there is no justification whatsoever for thinking that this arbitrary
formula represents any probability at all. Similarly, the fact that one student may have a
larger value of p than another student does not even in imply that it is more in the interest
of the first student to stay on. At times in their paper, Breen and Goldthorpe make both
of these misinterpretations. For example, they prove that, on the basis of their model, p
is always greater for service class than working class students and deduce that “children
from middle class backgrounds will more strongly prefer (in the sense of perceiving it to
be in their best interests) to remain in school to a further level of education rather than
leave” (Breen and Goldthorpe, 1997, p. 12). Given the arbitrary nature of the formula for
p such a deduction is totally unjustified. Similarly Breen and Goldthorpe’s argument on
pages 287–289 rests on the implicit assumption that p is a probability.
28
of success in exams to be high enough for working class pupils to compensate
for a difference between the costs and guaranteed utility of education.)
2.2
Murphy
Murphy takes issue with the view that class differentials in educational attainment have anything to do with inequality of opportunity (Murphy, 1981;
Murphy, 1990). Murphy states that the equation of class differentials in educational attainment with class inequality is due to a failure to take class
differences in educational aspiration seriously. In Murphy’s view, working
class youth simply demand lower levels of education than do middle class
youth. This difference in the levels of demand for education is not due to
inequality of opportunity, but is simply a matter of taste — i.e. preferences
differ by social class. Murphy claims that the refusal to acknowledge this
cultural difference between the social classes is a form of liberal ethnocentrism.
It is clear how a rational choice approach could lead to such a view. Agents
act to maximise the fulfilment of their preferences, given their beliefs about
the situation. Class differentials in educational attainment have not narrowed
significantly, despite the removal of obstacles that stood in the way of working
class pupils in the past (for instance, through the provision of universal free
education, compulsory up to age 16). Therefore, it might be argued, working
class pupils must simply prefer lower levels of educational attainment to those
demanded by middle class pupils.
However, this version of the rational choice approach leaves several issues
unexamined.
Firstly, to what extent are class differentials in attainment actually due to
decisions made by pupils, and to what extent do unequal resources (cultural
or economic) determine educational ability? It seems implausible that class
differences in initial attainment (at GCSE for instance) are due to rational
action by pupils. Of course Murphy could claim that even young children
make a decision on whether to work hard at school or not, and that the
primary effects of stratification are due to working class children making a
rational choice not to work hard at school, since they dislike schoolwork.
However, in my view, the younger the children we are considering, the more
dubious it is to present their actions in this way. The problems of a lack of
information on the future benefits of education, and of weakness of will when
29
immediate effort is demanded for a distant payoff, are compounded massively
in the case of five year old ‘decision-makers’.
Secondly, what causes the class differential in demand for education? Murphy
claims that working class pupils demand less education than middle class
pupils because they simply dislike the educational process more than middle
class pupils. But there are other possible explanations for the difference in
the amount of education demanded by pupils from different social classes,
even at the same level of demonstrated ability. Working class pupils may
have lower occupational aspirations than middle class pupils do, and may
therefore feel less need for qualifications. Another possibility is that working
class pupils may have a lower level of belief in the instrumental value of
education in attaining a desired occupation than middle class pupils do, or
they may evaluate their chances of success in further and higher education
modestly as compared to middle class pupils. Murphy does not give any
reason for preferring his explanation to these alternatives.
2.2.1
Formalising Murphy
It should be noted that unlike other rational choice theorists of education,
Murphy does not use Boudon’s distinction between the primary and secondary effects of stratification. Murphy sees the effort made by pupils during
their school careers as a result of decisions made on the basis of preferences.
This contrasts with the more usual approach taking tested ability or exam
results at the first key educational “branching point” as a given primary effect. For Murphy, ability is a function of effort, and effort is a function of
social class. Working class pupils make less effort in school simply because
they dislike education. Thus they see the costs of education (for Murphy,
“costs” always include psychic costs) as being higher than do middle class
children — moreover they see the costs of education as a function of effort,
i.e. working class pupils dislike making an effort in school more than middle
class pupils do.
Thus Murphy’s model of educational decision making pre-GCSE can be represented pictorially as shown in figure 2.2. π1 denotes the probability of
passing if the pupil makes an effort, π2 denotes the probability of passing if
the pupil does not make an effort. U1 denotes the utility of passing, U2 the
utility of failing, and C is the cost of making an effort.
One can read off from this that the expected net utility from making an effort
π1
30
Pass .r
..... U
..........
1
..........
..........
......
..........
..........
..........
..........
.
.
.
.
.
.
.
.
.
..
..........
..........
...... ....................
......
..........
..........
......
.
.
.
.
.
..........
.
..........
......
..........
......
.
.
.
.
..........
.
....
.
..........
.
.
.
.
..........
....
.
.
.
..........
.
.
....
.
.
.
.
.
.
....
.
.
.
.
.
....
.
.
.
.
.
....
.
.
.
.
.
..
......
......
......
......
...........
......
......
......
......
......
......
......
......
......
......
......
....
......
..........
......
..........
......
..........
......
..........
.
.
.
.
.
......
.
.
.
.
......
......
..........
......
..........
......
..........
......
...... ....................
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
......
Make an effort, cost C
r
1 − π1
Fail
r U2
r
π2
Pass r
U1
r
Not make an effort
1 − π2
Fail
r U2
Figure 2.2: Murphy’s model
is given by:
π1 U1 + (1 − π1 )U2 − C
and that the expected net utility from not making an effort is given by:
π2 U1 + (1 − π2 )U2
So the expected utility from making an effort minus the expected utility from
not making an effort is given by:
p = (π1 − π2 )(U1 − U2 ) − C
This does not completely fit the generic model.
Nevertheless, Murphy’s theory as applied to post GCSE decisions does fit the
model. Murphy takes the values of the utilities to be independent of social
class and explains the difference between the classes by
1. The difference in (psychic) costs. These differences in costs are a function of differences in preferences, which in turn are a function of class.
Working class pupils find education more unpleasant than middle class
pupils do.
31
2. The difference in expected chances of success (π) post 16 is also a
function of social class via ability. However, it should be borne in mind
that for Murphy the social class difference in ability is an indirect consequence of the class difference in costs at the previous educational level.
This is because working class pupils make less effort at the previous
educational level.
2.3
Boudon
Boudon explains class differentials in educational attainment with reference
to two mechanisms — the primary and secondary effects of stratification.
The primary effects of stratification are cultural inequalities that determine
the academic abilities of pupils. The secondary effects of stratification are
the different costs and benefits that are associated with different educational
decisions for pupils from different social classes.
For Boudon, the benefits associated with each educational option vary with
social class because ambition is relative to the social starting point of an
individual. So, a working class child who wants to be a lawyer must be more
ambitious than a middle class child who wants to be a lawyer. Therefore, high
prestige educational options may be essential in avoiding social demotion for
middle class pupils, whereas working class pupils can avoid social demotion
without pursuing such options. This leads to middle class pupils being more
likely to pursue such options than working class pupils at any given level of
ability.
For his argument to work, Boudon must make two assumptions. Boudon
does not make these assumptions explicit, but Breen and Goldthorpe do.
Firstly, we must assume that people’s priority is to avoid social demotion,
rather than to pursue social mobility. Otherwise, prestigious educational
options might be more attractive to students from working class backgrounds
than to students from middle class backgrounds, since the social distance
likely to be travelled as a result of successful completion of a prestigious
course will be far greater for the working class student.
Secondly, we must assume that failure in a high prestige option is believed
to be more likely to lead to social demotion than not attempting to pursue
such an option.
32
Boudon also states that the social costs of taking an educational option may
vary by social class. He gives one example of this.
“Thus not choosing a prestigious curriculum may represent a high
social cost for a youngster from a middle-class family if most of
his friends have chosen it; but choosing the same course may
represent a high cost for a lower-class youngster if most of his
friends have not.” (Boudon, 1974, p. 30)
This sort of explanation rests on an extreme and fallacious version of methodological individualism — the view that the explanation of a group’s behaviour
is identical to the sum of the explanations of the behaviour of each individual in the group. In this case, this leads to the explanation of each child’s
behaviour in terms of the behaviour of the rest of the group. This leaves us
none the wiser as to why the working class child finds that all his friends
are taking a low-prestige curriculum in the first place. (This is not to deny
that there may be a contextual effect such that a critical mass of working
class children taking a particular option may influence the remainder of the
group in favour of taking that option.) More importantly, it is easy to come
up with hypothetical costs and benefits which “explain” pupils’ decisions in
terms of rational choice theory. But if we are to derive concrete hypotheses
from rational choice theory, assumptions must be made about the beliefs and
desires of agents, and these assumptions should be made explicit. Boudon’s
failure to make his assumptions explicit seems to derive from a failure to
realise that assumptions are necessary. This problem may stem from the
version of rational choice theory that Boudon espouses in (Boudon, 1974),
which is a “utilitarian” rather than a “desire-fulfilment” approach.
“Now, the ‘rationality theory’ that was developed in chapter 2 is
incompatible with the value theory. The latter assumes that people
— at least some people — behave against their interests because
of the values they are committed to, whereas the former assumes
that people behave according to their interests in the sense that
they attempt to maximise the utility of their decisions. Of course
the ‘rationality factor’ and the ‘value factor’ cannot be added to
each other, since they are contradictory.” (Boudon, 1974, p. 111)
If we phrase rational choice theory in terms of the efficient fulfilment of
desires, it is clear that any hypotheses we form must be based on assumptions
about beliefs and desires. Boudon may think that in defining rational choice
33
in terms of utility maximisation he is avoiding this problem. In fact, it
just means that we have to make assumptions about what makes people
happy, and these assumptions will be, if anything, harder to back up than
assumptions about what people want. Value driven (or wertrational) action
only contradicts a form of rational choice theory that assumes that the desires
acted on by an agent must be individualistic and hedonistic.
So, if rational choice hypotheses must be based on some view of beliefs and
desires, and if values are just a special kind of belief, then Boudon is wrong
to claim that explanations of action in terms of the values of a group are
incompatible with rational choice theory.
Since Breen and Goldthorpe’s work is an extension and formalisation of
Boudon’s, it is not necessary to formalise Boudon separately.
2.4
Gambetta
Gambetta (Gambetta, 1987) notes that the decisions of working class students are more sensitive to poor academic performance than those of middle
class pupils. However, Gambetta disputes Boudon’s explanation of this phenomenon in terms of differing costs and benefits of education across social
classes. Gambetta recognises that, although a working class child may find
the decision to drop out of school early to be a low cost option, since it does
not entail social demotion, whereas for a middle class child it would be a
high cost option, since it would probably entail social demotion, correspondingly, the relative benefit to a working class pupil of taking a high prestige
educational option should be much higher than it would be for a middle
class pupil, since the social distance travelled is much further. Gambetta
prefers an explanation in terms of class differences in norms and preferences
for education.
“. . . it could either be that relatively more subjects in the middle
class feel a greater normative pressure to resist the temptation to
abandon school after a failure or, on the other hand, that relatively
more subjects in the working class do not attach as high a value
to education.” (Gambetta, 1987, p. 173)
Of course, class differences in norms and preferences are a possible alternative
explanation to Boudon’s. However, Gambetta fails to provide evidence that
34
norms regarding education vary according to social class.
Gambetta does not pursue a hard-line rational choice approach to the study
of educational decision making. Instead, he is happy to borrow from structuralism and sub-intentional theories. In my view, although Gambetta states
that he is aware of the dangers of an eclectic approach, which may “fail to
offer simplifying insights”, (Gambetta, 1987, p. 167) this does not always
stop him from falling into these pitfalls. Gambetta’s conclusion is that many
mechanisms are at work.
“Educational decisions are the result of three main processes: of
what one can do, of what one wants to do and, indirectly, of the
conditions that shape one’s preferences and intentions. They are
the result partly of causality and partly of intentionality. It seems
hardly the case that decisions are generated by either of these two
forces alone.” (Gambetta, 1987, p. 169)
This seems hardly a very interesting conclusion.
2.4.1
Formalising Gambetta
Gambetta can easily be fitted into the generic model. Although he describes
many mechanisms which may cause class differentials, all of the mechanisms
which can be seen as part of a rational choice approach relate primarily to
the cost, C. Specifically in explaining why the costs exceed the guaranteed
utility from education he emphasises social pressures which increase the cost
of education for working class pupils. He does not take the approach of
suggesting that U3 > U2 for working class pupils. In common with Boudon
and Goldthorpe, Gambetta sees perceived chances of success, π, as a function
of ability which in turn is a function of class. This also contributes to class
differentials.
2.5
Breen and Goldthorpe’s Model
Breen and Goldthorpe’s stated aim is to explain the following empirical phenomena:
35
“(i) increasing educational participation rates; (ii) little change in
class differentials in these rates; and (iii) a recent and very rapid
erosion of gender differentials in educational attainment levels.”
(Breen and Goldthorpe, 1997, p. 275)
Breen and Goldthorpe aim to explain these empirical phenomena as the
product of rational choices by families. They especially wish to avoid giving
cultural or normative differences any place in this explanation. They construct their model formally in order that it should have testable implications.
Breen and Goldthorpe follow Boudon’s (Boudon, 1974) distinction between
primary and secondary effects of stratification, and concur in Boudon’s claim
that it is the secondary effects (i.e. the different costs and benefits that are
associated with different educational decisions for pupils from different social
classes) that play the crucial role. They make explicit Boudon’s implicit
assumptions that:
1. People’s priority is to avoid social demotion, rather than to pursue
social mobility
2. Failure in a high prestige option is believed to be more likely to lead
to social demotion than not attempting to pursue such an option.
This second assumption means assuming either that it is in fact the case
that failure in a high prestige option is more likely to lead to social demotion
than not attempting such an option, or assuming that although this is not
in fact the case, people mistakenly believe that it is. In the latter case, the
mistaken belief will itself require explanation. The former assumption on the
other hand is open to empirical attack. Payne (Payne, 1987) examines NCDS
(National Child Development Survey) data on young people who reached 16
in 1974 and finds no disadvantage in staying on in education post-16, even
to students who gain no further qualifications.
“. . . those who left at 17 or 18 with qualifications no better than
those of minimum age leavers suffered no long term disadvantage
in comparison with the latter, despite their loss of potential work
experience, and some groups had lower unemployment rates in
the long term than minimum age leavers with equally good qualifications.” (Payne, 1987, p. 425)
36
In addition, there is evidence that an uncompleted college degree is not
useless in the labour market (see Becker (Becker, 1964) and Layard and
Psacharopoulos (Layard and Psacharopoulos, 1974)).
Breen and Goldthorpe explicitly assume that values, norms and beliefs regarding education do not vary by social class, and that classes differ in terms
of two factors only:
1. Average ability
2. Resources
Breen and Goldthorpe focus on three factors that will be taken into account
by parents and children.
1. The cost of remaining in school
2. The likelihood of success in continued education
3. The value that is attached to educational outcomes. This is determined
by beliefs about the chances that each outcome gives of access to each
social class. (Although these beliefs need not be accurate according to
Breen and Goldthorpe, they must not be systematically distorted by
social class.)
Their model contains three social classes. The service class, which consists of
professionals, administrators and managers; the working class; and the underclass, which consists of the unemployed and “those with only a precarious
place in the labour market and in only the lowest grades of employment if not
unemployed” (Breen and Goldthorpe, 1997, p. 281).
Three mechanisms are postulated for the secondary effects of stratification.
The first and most important of these is what Breen and Goldthorpe call
‘relative risk aversion’. Again, in line with Boudon, Breen and Goldthorpe
state that aspirations must be seen as relative to the social starting point of an
individual. As I stated earlier, Breen and Goldthorpe explicitly assume that
the priority in all classes is avoiding downward social mobility. So middle
class pupils will prefer prestigious educational options more strongly than
working class pupils, due to the greater value middle class pupils attach to
middle class occupational outcomes.
37
The second mechanism is due to the different levels of educational ability
shown by pupils in different social classes. Pupils’ own knowledge of their
ability is likely to shape the subjective probability they attach to success in
a particular curriculum.
The third mechanism is that middle class families have more resources to
devote to their children’s education. So, more middle class than working
class families can allow their children to continue in education for longer.
Breen and Goldthorpe’s work is an advance on Boudon’s approach, in that
it makes explicit assumptions, and these have testable implications. The
main one of these is that before pursuing a particular curriculum, working
class children will require a higher probability of success than middle class
children. This view would have to be revised if either:
1. Working class pupils do not require a higher level of perceived success
than middle class children before pursuing a particular curriculum; or
2. Pupils’ expectations of success are formed differently according to social
class, rather than being formed purely according to objective evidence.
However, one would need to gather subjective data on expectations to test
Breen and Goldthorpe’s view — and this is something that rational choice
theorists have not generally been inclined to do. Or, as Manski puts it:
“The standard economic model assumes that a youth’s schooling
choice c is a function f (.) of his or her expected returns to schooling r; that is c = f (r). Suppose that one wishes to learn the decision rule f (.) mapping expectations into choices. If one observes
the choices and expectations of a sample of youth, then one can
infer the decision rule. But if one observes only the choices of
these youth, then clearly one cannot infer f (.).” (Manski, 1993,
p. 44–45)
One problem with Breen and Goldthorpe’s model is that one cannot infer
from it how pupils from ‘underclass’ backgrounds will behave, since there is
no risk of social demotion for these pupils. One could argue that these pupils
will have low aspirations since they come from humble backgrounds, but
equally, one could argue that they will be able to take ambitious educational
options since they face no risk of social demotion if they fail.
38
I would also argue that ‘relative risk aversion’ is a misleading term to describe
Breen and Goldthorpe’s model. Risk aversion refers to a preference for a
large probability of a small gain rather than a small probability of a large
gain (e.g. I am risk averse if I prefer a 90% chance of winning $50 to a 10%
chance of winning $5,000). This concept of risk aversion does not apply to
Breen and Goldthorpe’s model. On this model, a service class pupil will
pursue an educational option on the grounds that it represents their best
chance of remaining in the service class, even if this means a greater risk
of demotion to the underclass than not pursuing this option would entail.
So, perhaps Breen and Goldthorpe’s model is more accurately described as
one of ‘loss aversion’ than of ‘risk aversion’. Pupils perceive themselves as
already having a particular social class status (i.e. that of their parents), and
they will risk everything to maintain that status.
In addition, I dispute Breen and Goldthorpe’s claim that their model does
not rest on an assumption of class differences in preferences. It is clear that,
on Breen and Goldthorpe’s model, the working class attach a higher value
to a working class outcome than do the service class, and so on. Thus,
preferences for each outcome class vary by class of origin.
2.5.1
Applying the Utility Model to Breen and Goldthorpe
I wish to derive Breen and Goldthorpe’s formulae for p from the utility model
described above. The formulae for the service class pupil and for the working
class pupil are as follows:
pS =
πα + (1 − π)β1
πα + (1 − π)β1 + γ1
π + (1 − π)(β1 + β2 )
π + (1 − π)(β1 + β2 ) + (γ1 + γ2 )
Where the various variables are defined by:
pW =
1. π is the perceived chance of ‘passing’ given that one has stayed on
2. α is the probability of entering the service class given that one has
‘passed’.
3. β1 is the probability of entering the service class given that one has
‘failed’
39
4. β2 is the probability of entering the working class given that one has
‘failed’
5. γ1 is the probability of entering the service class given that one has not
stayed on
6. γ2 is the probability of entering the working class given that one has
not stayed on. The situation is summarised diagrammatically by figure
2.3 (figure 1 in (Breen and Goldthorpe, 1997)).
α.............................................r S
..........
..........
Stay on
Pass ..r..................................................................................................................................................r
........
W
π ...............................................
1
−
α
...
.
.
.
.
.
.
.
.
.
........
....
..........
..........
..........
..........
.............................
.
.
.
.
.
.
.
.
.
..........
..........
..
..........
..........
..........
..........
..........
.
..........
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..........
..
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
.......... ...................
..........
................................................................................................................
................
..........
........
..........
........
..........
........
..........
........
..........
........
..........
........
..........
........
..........
..........
........
..........
........
.....
........
........
........
........
........
........
........
........
........
........
........
...
........
..........
........
..........
........
..........
........
..........
........
..........
.
.
.
.
.
.
.
.
.
........
....
........
..........
........
..........
........ ...................
..................................................................................................................
..........
..........
..........
..........
..........
..........
..........
..........
..........
..........
........
r
r
rS
β1
1−π
β2
r
Fail
1 − β1 − β2
rU
rS
γ1
r
Leave
1 − γ1 − γ2
rW
γ2
rW
rU
Figure 2.3: Breen and Goldthorpe’s model
The central assumption of Breen and Goldthorpe’s ‘relative risk aversion’
model is:
“. . . families in both classes alike seek to ensure, so far as they
can, that their children acquire a class position at least as advantageous as that from which they originate or, in other words they
seek to avoid downward social mobility.” (Breen and Goldthorpe,
1997, p. 283)
We can express this in terms of utilities as shown in table 2.1:
For service class
Perceived utility of service class
occupations
Perceived utility of working class
occupations
Perceived utility of underclass
occupations
40
1
For working
class
1
0
1
0
0
Table 2.1: Perceived Utilities in Breen and Goldthorpe’s Model
If one applied this to members of the underclass they would perceive all
occupational classes as being of equal utility.
Breen and Goldthorpe would certainly resist the application of the concept of
utility to their work. In particular they would resist the view that they give
an explanation that at root relies on different perceptions of utility between
social classes. However, as we shall see, their model can be derived from the
above table.
Suppose that the probabilities of entering the various classes given the three
educational outcomes are as described in figure 1 of Breen and Goldthorpe:
Then one can say:
U1S = α,
U1W = 1,
U2S = β1 ,
U2W = β1 + β2 ,
U3S = γ1
U3W = γ1 + γ2
where the superscripts S and W refer to the service and working class respectively, and the U ’s refer to the expected utility of each possible educational
outcome (labelled 1 [staying on and passing], 2 [staying on and failing] and
3 [leaving]). One can now write the expected utility from staying on, which
we recall is:
π(U1 ) + (1 − π)U2
and the expected utility from leaving which is U3 . Thus for service class
pupils the parameter p is given by:
pS =
πα + (1 − π)β1
πα + (1 − π)β1 + γ1 )
41
and similarly for working class pupils it is given by:
pW =
π + (1 − π)(β1 + β2 )
.
π + (1 − π)(β1 + β2 ) + (γ1 + γ2 )
These are Breen and Goldthorpe’s formulae 1 and 2 (Breen and Goldthorpe,
1997). Thus, their model fits into the utility model completely. This reveals
the implicit reliance of their model on the differences between utilities as
perceived by each class. Therefore Breen and Goldthorpe have clearly failed
in their stated aim of providing an explanation of class differentials which
does not ultimately rely on class differences in preferences — i.e. they have
failed to avoid a ‘cultural’ or ‘normative’ account.
2.5.2
Breen and Goldthorpe’s Assumptions
Further to the table of utilities, it is still necessary to make some assumptions
about the parameters π, α, β etc. if one is to extract any predictions from
the model.
Breen and Goldthorpe make four assumptions about the parameters in their
model, which I will now discuss:
1. α > β1 ,
α > γ1
2. γ1 + γ2 > β1 + β2
3.
γ2
γ1
≥ 1,
γ2
γ1
≥
β1
β2
4. α > 0.5
The first assumption is uncontroversial — it simply states that remaining at
school and succeeding affords a better chance of access to the service class
than does remaining at school and failing or leaving school.
The second assumption is most important to the conclusions drawn by Breen
and Goldthorpe. It states that remaining at school and failing increases the
chances of entering the underclass as against just leaving school. Breen and
Goldthorpe make no attempt to justify this rather striking assumption with
empirical evidence. As I pointed out earlier, Payne’s (Payne, 1987) analysis
of NCDS (National Child Development Survey)data seems to contradict the
assumption in the case of the UK.
42
Assumptions 3 and 4 are less central to Breen and Goldthorpe’s model than
assumption 2. However, the dubiousness of these assumptions still deserves
to be pointed out.
Assumption 3 is stated as:
“Those who leave school immediately have a better chance of entry
into the working class than the service class. This may or may not
be the case among those who remain at school and fail though, if it
is, their odds of entering the working class rather than the service
class are no greater than for those who leave school immediately”
The first part of this assumption constrains the relative sizes of the respective
classes. For instance, the assumption would be false in the case where 70% of
the population is middle class, and 30% working class, and only 30% of the
population stay on in education at a given decision point. In this example,
even if all 30% who stay on at school end up in the service class, more than
half of those who leave school immediately must also. (What we mean by
‘passing’ and ‘failing’ are also highly relevant here.) Since assumption 3 can
only be true given a favourable combination of class structure and staying
on rate (and even then the assumption will not necessarily be true) it would
seem reasonable to demand some empirical justification for the assumption.
This is not provided. It seems very unlikely that assumption 3 would hold for
higher level educational decisions, such as whether or not to go to university.
The second part of assumption 3 fails to take into account the possibility
that those who fail a course of study may still gain labour market benefits
above and beyond what they would have gained from any work experience
they missed out on due to study.
Assumption 4 is stated as
“Staying on at school and passing the examination makes entry
to the service class more likely than entry to the working class.”
The plausibility of the assumption depends on how we define a pass or fail.
If passing the examination means getting at least 3 Cs at A level, the assumption seems plausible. If however, we define a pass as one E grade at A
level, then the assumption that this makes entry to the service class more
likely than entry to the working class becomes questionable.
43
Given the expressions for expected utilities we derived above, we can rewrite
these assumptions as:
1. U1S > U2S ,
U1S > U3S .
2. U3W > U2W
3.
U W −U3S
U3S
≥ 1 ≡ U3W − U3S > U3W ≡ U3W > 2U3S and
U W −U3S
U3S
≥
U2S
U2W −U2S
4. U1S > 0.5
In words:
1. For the service class, the utility of staying on and passing is greater
than the utility of staying on and failing. For the service class, the
utility of staying on and passing is greater than the utility of leaving.
2. For the working class, the utility of leaving is greater than the utility
of staying on and failing.
3. The utility of leaving is more than twice as big for the working class
as for the service class. Putting the additional assumption into words
does not make it any clearer than it is as a formula.
4. For the service class, the utility of staying on and passing is greater
than 0.5 (note that 1 is the maximum available utility) Assumptions
three and four only look more contrived and implausible when phrased
in terms of utilities.
One striking fact about the assumptions once they are phrased in terms of
utilities is that they contain an asymmetry between the working and service
classes. In order to avoid this asymmetry, one would have to add the following
assumptions.
1*. For the working class, the utility of staying on and passing is greater
than the utility of staying on and failing. For the working class, the utility
of staying on and passing is greater than the utility of leaving.
2*. For the service class, the utility of leaving is greater than the utility of
staying on and failing.
Assumptions 1 and 1* are certainly reasonable assumptions. It is clear that
assumption 2 is the driving force behind Breen and Goldthorpe’s argument.
44
However, clearly assumptions 3 and 4 (which are unfounded) must also play
a key role in their argument.
2.5.3
A Generic Model?
Let us examine Breen and Goldthorpe’s claim that their model is generic.
2
“The model that we present is intended to be generic: that is, as
one applicable in principle to the entire range of decisions that
young people may be required to make over the course of their
educational careers as regards leaving or staying on or as regards
which educational option to pursue.” (Breen and Goldthorpe,
1997, p. 279)
It is a doubtful empirical question whether assumptions 2, 3, and 4 hold
when the decision in question is whether to stay on at school post 16. It is
clear, however, that assumptions 2, 3 and 4 cannot hold when applied to all
educational decisions.
Assumption 2 (that staying on and failing increases the chance of entering the
underclass as against leaving) would seem not to hold in the case of US college
students. As I pointed out earlier, research shows that individuals with
uncompleted college degrees fare better in the labour market than individuals
who did not go to college. Of course, there is always a difficulty in evaluating
such research. (For instance, it may be that individuals who attempt college
but do not graduate are simply more able than those who do not, and would
have done better in the labour market even if they had not gone to college.)
But the risk of entering the underclass will certainly decrease as we consider
more ambitious educational options, becoming insignificant when we consider
decisions such as whether to train to be a barrister having completed an
undergraduate degree.
Assumption 3 states that those who leave immediately have a better chance
of entry into the working class than the service class. Again, this is clearly
2
In private communication, John Goldthorpe has clarified to me that the word ‘generic’
is intended in an extremely weak sense here, and that the model is only applicable to
decisions which lead to class differentials, and even then, applying the model to the full
range of decisions would necessitate changing the outcome classes. However, I do not think
that this weak usage of the word ‘generic’ is at all obvious from the text, so perhaps it
may be useful to explain why a stronger claim would be false.
45
not true of more advanced options, such as the decision of whether to leave
education after one’s undergraduate degree or to take a postgraduate course.
Assumption 4 claims that staying on and passing the examination makes
entry to the service class more likely than entry to the working class. This
assumption would be implausible in the case of less prestigious options. E.g.
consider the decision of whether to stay on to take a vocational qualification
such as a GNVQ or leave the education system.
There is a special problem with decisions where more than one alternative is
considered simultaneously. For example, the decision of whether to stay on
in the sixth form to do A-levels or leave might arguably meet assumptions 3
and 4. However, one must question whether any individual actually makes
this decision. I would argue that the decisions made by real individuals are
much finer e.g. the decision to do A-levels or a GNVQ. In this more relevant
example condition 4 might be plausible in that passing A-levels (if by this we
mean getting reasonably good grades) might make the service class the most
likely destination. It is not necessarily clear however that someone who does
a BTEC has a better chance of entry to the working class than to the service
class. If Breen and Goldthorpe’s model is to apply to this decision, then this
must be the case by assumption 3. Suppose for the sake of the exposition
that this is the case. Since, by assumption 3, it is true that the probability
of entering the working class is greater than the probability of entering the
service class if one takes a BTEC, and assuming the probability of failing a
BTEC is not too high — it is not at all unlikely that a pupil who passes a
BTEC also has a greater probability of entering the working class than the
service class. Now consider the case of a pupil who decides between taking
a BTEC and leaving education. For this decision assumption 4 reads: the
probability of the pupil entering the service class if they do a BTEC is at
least one half. This is of course impossible if the probability of them entering
the working class is greater than the probability of him entering the service
class. Thus assumptions 3 and 4 cannot apply across the range of educational
decisions. Furthermore they do not apply to realistic educational decisions.
Goldthorpe claims that:
“The assumption that the educational decision problem has three
possible outcomes, each of which results in direct entry to the
labour market, is innocuous” (Breen and Goldthorpe, 1997, p.
287)
46
If this statement is taken to mean that applying this assumption to the full
range of educational decisions will be unproblematic, then it is certainly false.
In addition, extending assumptions 3 and 4 to a model containing many social
classes would be highly problematic.
So, Breen and Goldthorpe’s model would have to be adapted considerably
to apply to the full range of educational decisions. In my view, this makes
the claim that it is a generic model somewhat misleading.
2.5.4
Breen and Goldthorpe’s attempt to explain empirical changes
Increase in Participation in Education
The first empirical phenomenon Breen and Goldthorpe try to explain is the
increase in participation in education by children of both working and service
class origins. They explain this by the fact that the relative costs of education
have declined over time in all economically advanced countries. However,
they go on to say that:
“. . . in so far as education is regarded as a ‘positional’ good, p
could be expected to rise steadily simply as a consequence of educational expansion itself.” (Breen and Goldthorpe, 1997, p. 294)
By this they presumably mean that increased rates of educational participation have made extended educational participation increasingly necessary for
individuals who wish to enter the service class. For an earlier formalisation
of this, highly plausible, view see (Halsey et al., 1980, p. 113-116).
However, according to Breen and Goldthorpe’s model the working class is indifferent to the distinction between a service and a working class occupation.
Thus this change would not necessarily affect them. Indeed this change could
occur simultaneously with a growth in the chance of entering the underclass
if one leaves education. As argued previously assumption 2 then requires a
greater chance of entering the underclass if one pursues education and fails.
This would lead to the working class being less likely to persist in education.
Thus the only explanation of rising participation in education that can be
derived from Breen and Goldthorpe’s model is that of decreasing costs.
47
Persistence of Class Differentials
The second phenomenon that Breen and Goldthorpe attempt to explain is
the fact that class differentials in educational participation have remained
stable despite this increase in overall participation rates.
Given the fact that the underclass (if this is, as they suggest, to represent
the unemployed and those with a tenuous position in the labour market)
has grown over the past two decades, and given Breen and Goldthorpe’s
assumption 2 — which states that the risk of falling into the underclass is
greater if one fails in education than if one chooses to leave — it seems likely
that the working class (if they are trying to minimise their chances of falling
into the underclass) would be even less inclined to risk pursuing further
education, all other things being equal. This contradicts the empirical fact
that the working class has increasingly pursued further education as have
the service class. Thus, to account for empirical facts, one is forced to reject
either assumption 2 or the assumption that the working class’ educational
decisions are driven by the desire to minimise their chances of falling into
the underclass.
Gender
Breen and Goldthorpe state that in the past, women’s class position did not
depend strongly on their educational qualifications. Therefore, Breen and
Goldthorpe’s, model could not be applied to women pre 1970s. However,
due to increased labour market participation by women this is no longer the
case and the model increasingly applies to women.
Breen and Goldthorpe seem to claim that their model explains the decrease
in gender inequality in educational participation. However, in the light of the
statement that the model did not apply to women in the past, it is hard to
see how this could be the case. They can only explain why the participation
of women is currently similar to that of men. They cannot use their model
to explain the change. Clearly, Breen and Goldthorpe’s model is designed
to explain class differences, and the idea that a key aim of their paper is
to explain the erosion of gender differences in educational participation is
laughable. This is a clear example of gender being seen as an obligatory “addon” in sociological analysis, rather than being taken seriously as a category
of analysis.
2.6
48
Conclusions
Different rational choice theorists have very different approaches to the explanation of educational inequality, as they do to other questions. Murphy
uses class differentials in preferences as an explanation of class inequality in
attainment, whereas Boudon claims that rational choice theory is in conflict
with theories that postulate differences in preferences and values between
social classes. Gambetta states that structural constraints that limit agents’
scope for choice also limit the explanatory force of rational choice theory. In
my view, the rational choice assumption cannot be in direct conflict with any
claims about beliefs, desires or constraints, since rational choice is a claim
about a decision rule, not about the beliefs, desires or constraints which are
factored into the decision. No determinate hypotheses can be derived from
the decision rule alone, and the failure to fully realise this seems to lead to
some confusion among rational choice theorists. I have applied a generic
model to the various approaches considered here, and I hope this has been
helpful in bringing out the key differences.
I have given particular attention to Breen and Goldthorpe’s model of educational decision-making. My main conclusions on this model are as follows:
Breen and Goldthorpe describe their model as being one of “risk-aversion”,
but it is better understood as a model of “loss-aversion”. Breen and Goldthorpe fail to give an explanation of class differentials in educational participation rates that does not rely on class differences in preferences, although
the preferences of each social class do arise from a common mechanism (the
desire to avoid social demotion) on their model. In fact, Breen and Goldthorpe’s model implies that an individual’s class of origin affects the value
they attach to each destination class, such that individuals from service class
origins attach a higher value to achieving a service class “destination” than
do individuals from working class origins.
The assumptions on which Breen and Goldthorpe’s model is based are highly
problematic, and are not backed up by empirical data. For instance, the
assumption that remaining in education and failing increases the chance of
entering the underclass as compared to just leaving education is crucial to the
model, but what empirical evidence is available seems to contradict rather
than support this view.
Breen and Goldthorpe’s model would have to be adapted enormously if it
was to apply to all decisions about educational participation or to all class
49
schemas. Therefore, to describe the model as “generic” is misleading.
Breen and Goldthorpe attempt to explain empirically observed changes in
educational participation rates in terms of their model. This attempt is
unsatisfactory even from the perspective of their model.
Crucially, none of the authors cited here has sought to provide empirical
evidence for the assumptions they have made regarding the motivations of
actors. In subsequent chapters I will attempt to bring empirical evidence
to bear on the questions of whether beliefs and attitudes regarding education vary according to social class and gender, and how class and gender
differentials in educational decision making can be explained.
Chapter 3
Methodology
3.1
Introduction
In previous chapters, I have discussed the attempts of two theoretical approaches — cultural reproduction theory and rational choice theory — to
explain educational inequalities. I have argued that neither of these approaches has been subjected to adequate empirical test. This chapter will
describe the survey that I have carried out, and the relevance of this empirical
work to the theoretical questions at hand. I will discuss the operationalisation of key concepts such as cultural capital, subjective beliefs about ability,
and attitudes towards education. I will also describe the sample, and the
practical questions involved in conducting the survey. (The survey questionnaire is included in appendix B).
3.1.1
Cultural Capital
According to Bourdieu’s theory of cultural reproduction, cultural resources
associated with the middle-class home facilitate the acquisition of educational credentials (see for example (Bourdieu and Passeron, 1990; Bourdieu,
1977a)). However, I have argued that the concept of cultural capital is ill
defined, and has been operationalised in various, sometimes rather arbitrary
ways.
In order to find out which, (if any), cultural resources are important in determining educational attainment, I will give as broad as possible an opera50
CHAPTER 3. METHODOLOGY
51
tionalisation of cultural capital.
Most surveys of cultural capital focus on culturally élite activities, such as
attendance at galleries and classical concerts. I asked these conventional
questions, but I also asked about more popular activities, for example, which
TV programmes pupils watch. This is because questions on participation
in extremely high-culture activities can only distinguish between a tiny minority who engage in these activities and the rest — leaving the bulk of the
population undifferentiated. This point is particularly important when you
consider that I am asking these questions to 16-year-olds.
In addition to questions about cultural activities, my questionnaire included
tests of vocabulary and cultural knowledge. These items were included because they represent a possible mechanism through which cultural participation may affect educational attainment.
Initially, I planned to ask pupils questions about the education system itself,
on the grounds that information about different educational options might
be an important advantage conferred by the middle class home. (This might
be seen either as an element of cultural capital, or as an advantage in decision
making on the rational choice model, or both.) Unfortunately, I was forced
to abandon this aim on the grounds that my questionnaire would be overlong
if I tried to do too much. However, I still think that this is an important
area for research, (this view was reinforced when one girl asked me what a
university was).
3.1.2
Rational Choice Theory
A type of explanation in competition with the cultural capital approach is
the rational choice approach. For example, Murphy claims that working-class
children simply do not demand as much education as middle class children do
(Murphy, 1981). But why do working-class children “demand less” education
than middle-class children — i.e. why are they less likely to stay on at school
post-16 and go on to university? Possible explanations include:
• Working class pupils’ perceived probabilities of success in education are
low as compared to middle class pupils of similar ability.
• In comparison to middle-class pupils, working-class pupils undervalue
education either as a means to success in the job market or as a good
in itself
52
• Breen and Goldthorpe’s view that middle class pupils attach a stronger
value to gaining middle class jobs than do working class pupils (Breen
and Goldthorpe, 1997).
Rational choice theorists have typically shied away from using “soft” data on
pupils’ motivations, relying instead on the doctrine of revealed preference,
which claims that individuals’ motivations are most reliably revealed by their
actions. An obvious problem with this view is that the pattern of pupils’
observed choices is compatible with any of the above explanations of this
pattern. I will assess these competing explanations using data on students’
beliefs, attitudes, plans, and aspirations.
In order to assess the possibility that working class pupils’ perceived probabilities of success in education are low as compared to middle class pupils
of similar ability, it was necessary to collect data on pupils’ subjective view
of their own academic abilities, as well as data on their actual academic performance (GCSE results). This has also allowed me to assess the possibility
that beliefs about ability are distorted by gender.
In order to assess the possibility that working class pupils place a lower value
on education than do middle class pupils, I collected data on pupils’ attitudes
towards education. This has also allowed me to examine the question of
whether there are gender differences in attitudes towards education.
For Breen and Goldthorpe’s view (Breen and Goldthorpe, 1997) to be supported, it would have to be the case that social class differences in educational
plans are explained by social class differences in occupational aspirations.
Therefore, I asked pupils about their occupational aspirations.
3.2
Survey Design
I surveyed pupils in their final year of compulsory schooling (i.e. “year 11”
pupils, about 16 years old) in England, in 1998.
I chose to survey year 11 pupils because I wished to follow up on the actual
GCSE results attained by pupils and on what they ended up doing after their
GCSEs. If I had used a younger year group, I would have had to wait too
long for this data. If I had used an older year group, a large proportion of
pupils, those who do not stay on in the 6th form, would have been excluded
from the sample.
53
Before GCSEs
I surveyed children on:
• parents’ occupation and educational level
• family structure and siblings
• parents’ cultural participation
• pupils’ cultural participation
• pupils’ cultural knowledge
• pupils’ active and passive vocabulary
• pupils’ expectations of exam grades
• pupils’ intentions for future study
• pupils’ occupational expectations and aspirations
• self-evaluated intelligence
• pupils’ attitude towards the intrinsic and instrumental value of education.
Initially, I aimed to survey parents. This is because a theory of cultural reproduction cannot be tested using only information on pupils’ cultural capital. A correlation between pupils’ cultural capital and attainment does not
support a theory of cultural reproduction unless there is also a correlation
between pupils’ and parents’ cultural capital. However, I decided that surveying parents was not a realistic option, since few would have volunteered
to participate. Therefore I have surveyed pupils on their parents’ characteristics. The cultural activities pupils were asked about regarding their parents
include reading (and number of books in the home), newspapers taken, type
of music and radio stations listened to, “formal culture” participation, and
the subjects discussed by parents in the home. It would have been ideal to
get information on parents directly from the parents themselves, as responses
from pupils may be less accurate than responses from parents. For instance,
it seems possible that pupils with high levels of cultural participation may
over report their parents’ cultural participation. However, note that de Graaf
et. al.(De Graaf et al., 2000) find that, in a survey of cultural practices in
54
the Netherlands, respondents own cultural practices have no effect on their
reporting of their parents’ cultural practices. At any rate, the best available
option was to get information on parents’ cultural activities from the pupils.
After GCSEs
I have collected information on the actual GCSE results gained by the pupils.
As far as possible, I have collected information on the pupils’ actual educational outcomes — whether they continued in education, and if so, what
course of study they followed. However, the level of detail available on this
question varied from school to school, and one school failed to provide the
information altogether, despite making a commitment to do so at the start of
the project, and despite repeated requests from me. Therefore, in my analyses of pupils’ decisions, I have used the pupils stated intentions of educational
participation as an outcome variable.
A table showing the associations between the key variables is shown in appendix A.
3.2.1
Pilot
The survey was piloted in order to assess the questionnaire. In particular,
I needed to determine whether the questionnaire could be completed in the
time available, even by pupils’ of relatively low ability, whether all the items
on the questionnaire could be readily understood by pupils and whether the
tests of cultural knowledge and active and passive vocabulary were pitched at
approximately the right level. In general, I needed to confirm that all items
that required a spread of responses in order to discriminate between individuals actually did so and that as far as possible, no item caused confusion or
offence.
I piloted my questionnaire at an Oxfordshire comprehensive, with a small
sample (16 pupils), on July 18 1997. I used year 10 students rather than year
11 students. This was because year 11 students would have been too busy
with their GCSE’s at this stage in the year to participate. I introduced the
questionnaire, explaining that it asked them about various things, including
what they did in their spare time, what they thought about studying, and
what they wanted to do in the future. I stressed the confidentiality of the
questionnaire. Since I did not intend to follow up on GCSE results for these
55
students, or what they did after their GCSEs, there was no need for them to
identify themselves on the questionnaire, either by name or number, so the
responses were completely anonymous.
The pupils completed the questionnaire without much difficulty, and did not
object to any of the questions. Only a few changes to the questionnaire
were necessary. The only parts of the questionnaire that pupils had serious
difficulties with were those intended to test active vocabulary.
I had taken an open-ended sentence completion test from Denis Lawton’s
Social Class, Language and Education (Lawton, 1968). This test involves
completing sentences so that they make sense, using a specific number of
words and without breaking into more than one sentence.
Example: The admiral went. . . . . . . . . . . . . . . . . . . . . . . . . . . .(add ten
words)
The admiral went to China six times when he was in the navy.
This test was originally used to test the ability to formulate sentences among
15-year-old boys, in order to compare the complexity of grammatical structures used by middle and working class boys. This test was too hard for
the pupils in my sample — most of them managed at best a few correct responses. Even those pupils who gave correct responses to most of the items
did not attempt to use complex grammatical structures — and, one may ask,
why on earth should they, as they were not instructed to attempt to do this.
It is possible that social class differences in performance on this test could
reflect the ability to know what the assessor is going to reward in the absence
of explicit instructions on this (i.e. a knowledge of the rules of the game) as
much as superior writing skills. In addition, the results of this test would
have been difficult and time consuming to analyse.
Another test of active vocabulary from Lawton (Lawton, 1968) that I tried
in the pilot was to ask pupils to write a short essay on one of the following
subjects “home”, “school” or “my life in 10 years time”. Many pupils did
not even attempt this, and those that did tended to write very short pieces
indeed. Furthermore, the results of this test would have been perhaps even
more cumbersome to analyse than those of the previous test.
I have replaced these tests with my own test of active vocabulary, which is
described in 3.3.1. Pupils have not found this test too difficult, and it has
56
the advantage that they can give up after thinking of as many synonyms as
they can. In addition, this test can be scored as a simple mark out of five
for each question.
The other changes I made were minor. For instance, there were a few words
that some pupils did not understand. Doing the pilot allowed me to identify
words and phrases that were too difficult in order to change them for the
main survey. This was important, as I could never have predicted some of
the words pupils found difficult. However, this did not always work out as
planned. For instance, some pupils did not understand the word shine as in
“I would like a job that will allow me to shine”. I replaced the word “shine”
with “excel”, only to find that several pupils did not understand that word
either.
The pilot school is predominantly working class, with low GCSE results,
(in 1996, 21.5% of its pupils achieved 5 A–C grades). This made me feel
reasonably confident that, if these pupils could complete the questionnaire
easily, others should have no difficulty either. I found that the pupils’ scores
on the passive vocabulary and cultural knowledge tests were approximately
normally distributed, which was encouraging, although one cannot say too
much on the basis of such a small sample.
3.3
Questionnaire Design and Coding
In constructing my questionnaire, I drew from various previous surveys of
young people designed to measure various factors such as cultural participation, language ability, aspirations and attitudes towards school, (these include: PROJECT TALENT, British Social Attitudes: Young People’s Survey
SCPR, 1994 and National Child Development Survey, ESRC Data Archive,
1993).
3.3.1
Cultural Capital Questions
I have surveyed pupils on a broad range of possible components of cultural
capital.
1. Cultural knowledge
57
• Tested knowledge of famous cultural figures.
2. Language
• Active and passive vocabulary test scores.
3. Activities
• Reading: type and amount of books read, library use, newspapers
read.
• Television: type of TV programmes watched.
• Music: type of music listened to, playing an instrument.
• Participation in “public” or “formal” culture: art gallery, theatre
and concert attendance. (Film attendance was included in the
questionnaire, but excluded from the analysis, since only 7 respondents had watched films that could be categorised as in any
sense “high-brow”.)
Cultural Knowledge
The test of cultural knowledge is intended to measure pupils’ familiarity with
the dominant culture. This test is of course not intended to reflect all aspects
of a pupil’s cultural knowledge. However, it at least provides us with some
indication of cultural knowledge, something that has been lacking in most
previous research on cultural capital. The test consists of asking pupils to
categorise 25 famous cultural figures according to whether these figures are
associated with politics, music, novels, art or science. There were equal numbers of cultural figures in each of these categories. The figures were chosen
to reflect both past and contemporary culture. The items were pitched at
different levels of difficulty in order to give a scale that could differentiate
well between individuals in the sample according to their level of cultural
knowledge. Some figures were intended to be easily identifiable by practically anyone, while others were intended to be more difficult. Pupils were
instructed not to guess if they did not know the answer, and in constructing
a score, a deduction of 0.25 was made for false responses. (The figure of
0.25 simply reflects the fact that, for each question, there were five possible
responses other than “don’t know”. Therefore, a pupil who knew none of
the answers, but guessed consistently, could expect to get, on average, one
in five correct. A deduction of 0.25 for each false response would balance
this out, leaving the pupil with a score of zero.) Making a deduction for false
58
responses is standard practice in scoring multiple choice questions, and is important in the light of the possibility that boys may be more likely than girls
to guess in multiple choice tests (Gipps and Murphy, 1994). The questions
were introduced as follows:
Cultural Knowledge
Each of the following names is a person you may have heard of. For
each person listed, which do you associate him or her with most out
of the following categories: politics, music, novels, art or science? If
you do not know, do not guess, just tick “don’t know”.
The names of the people and the number of correct responses for each are
summarised in Table 3.1. These results show that, in the case of novelists,
the modern figures were far less well known than pre 20th Century ones
such as Dickens and Austen. Similarly, among those associated with music,
only Mozart was well known, and the least known figures were Gershwin
and Miles Davis — those least likely, perhaps, to be categorised as élite
rather than popular culture. The artists buck this trend in as much as
Picasso is correctly identified as an artist by 72% of respondents. However,
the most recent artist (Warhol) was the least known, despite his obvious
association with popular culture, not least through the Velvet Underground.
Only in the field of politics were contemporary and recent figures the best
known, perhaps unsurprisingly given the role of the media, and in the case
of JFK, an eponymous film that had been on at the cinema shortly before I
administered my questionnaire. Only 25% could correctly categorise Marx.
Of the scientists, Einstein stood head and shoulders above the rest as the
only scientist that could be correctly identified by a majority of respondents
(although it still seems a little shocking that one in ten failed to do so).
The distribution of pupil’s scores is shown in Figure 3.1. This shows an
approximately normal distribution with a wide spread of scores indicating
that the test was pitched at about the right level of difficulty.
Passive vocabulary
Passive vocabulary consists of those words that an individual can understand,
but cannot necessarily actually use in their own writing or speech. The test
of passive vocabulary used in the questionnaire was a conventional multiple
Albert Einstein
Galileo
Marie Curie
Louis Pasteur
Stephen Hawking
Charles Dickens
Jane Austen
Virginia Woolf
Graham Greene
Martin Amis
Bill Clinton
John F. Kennedy
Mahatma Gandhi
Gordon Brown
Karl Marx
Wolfgang Mozart
Rachmaninov
Johannes Brahms
Miles Davis
George Gershwin
Vincent van Gogh
Pablo Picasso
Claude Monet
Rembrandt
Andy Warhol
59
Percentage
correct response
90
31
27
21
21
89
76
23
7
5
94
89
47
33
25
81
18
15
12
10
85
72
49
29
23
Table 3.1: Knowledge of cultural figures
100
60
frequency
90
80
70
60
50
40
30
20
10
0
0
2 4 6 8 10 12 14 16 18 20 22 24 26
Figure 3.1: Histogram of knowledge score
choice “sentence completion” test (see for instance (Levy and Goldstein,
1984)). This test is derived mainly from the National Child Development
Study — SWEEP III (NCDS, 1993) (this test was administered to 16 year
olds in 1974). 14 out of the 15 questions I used were derived from the 35
questions used by NCDS. I could not use the full NCDS test, as it would
have taken too long. I found many of the questions originally used by NCDS
ambiguous, i.e. in my view, there was more than one correct answer to many
of the questions. For example:
In the days when people lived in caves and rough shelters there were
no fields or farms; men got their food by hunting animals and by
(finding, planting, gathering, growing, eating) berries and wild fruit.
I would guess that gathering was the required response, but in what sense
is finding actually wrong, given that this word makes a perfectly good and
sensible sentence? Not wishing my test to be an exercise in second-guessing
the tester, I was careful to select only those questions which, in my view, had
only one correct answer — an example is shown below. 15 such questions
were asked and the scores were added to give each pupil a mark out of 15.
The distribution of pupil’s marks for this test is shown in figure 3.2
61
Underline the word that correctly completes the sentence. If you do
not know which word is correct, just move on to the next question.
• Having had her expectations so much raised, it was very (realistic, discrediting, uplifting, disconcerting, discriminating) to
have them suddenly shattered.
60
frequency
50
40
30
20
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Figure 3.2: Histogram of passive vocabulary score — 41 were coded as missing
Active vocabulary
Active vocabulary consists of those words that an individual can actually use,
rather than simply understand. When piloted, tests of active vocabulary that
have been used elsewhere proved unsatisfactory (see section 3.2.1 for details).
Therefore, I developed a new test of active vocabulary. This test demands
that pupils provide several synonyms for each of five words given.
62
Synonyms are words that mean the same or approximately the same
as each other. For example happy and cheerful are synonyms.
For each of the following questions think of as many synonyms as you
can for each word (stop at five for each word).
1. Small
2. Stupid
3. Angry
4. Sad
5. Odd
Pupils were given a mark for each genuine synonym they provided thus giving
a total mark out of 25 for active vocabulary. The results are summarised in
Figure 3.3. Again there is a reasonable spread of results suggesting that this
test was pitched at about the right level of difficulty.
80
frequency
70
60
50
40
30
20
10
0
0 2 4 6 8 10 12 14 16 18 20 22 24
Figure 3.3: Histogram of active vocabulary score — 91 were coded as missing
The passive and active vocabulary scores were summed to give an overall
language score for each pupil. This composite score will be used in subsequent
statistical models as a measure of pupil’s linguistic ability. The distribution
is shown in Figure 3.4. This distribution is approximately normal.
60
63
frequency
50
40
30
20
10
0
1 4 7 10 13 16 19 22 25 28 31 34 37 40
Figure 3.4: Histogram of language score — 34 were coded as missing
Reading Habits
Previous surveys of cultural capital have generally asked individuals how
much they read, and sometimes also whether they are a member of a public
library, and how many books they have at home. I asked these questions,
but also asked pupils what books and newspapers they read.
The section of the questionnaire concerned with pupils’ reading habits asks
pupils to state the titles and authors of any books they have read recently
that are unconnected with their schoolwork, and to give the names of their
favourite authors if they have any. Clearly, the responses to these questions
must be categorised according to the “cultural capital” content of the books.
The categorisation I used is shown in table 3.2. The starred categories are
those for which credit for cultural capital was given.
When unsure of the category a book fell into, I used the Book Review Digest
database. This is a database of reviews from 100 English language journals
such as the Times Literary Supplement and the New York Review of Books,
from 1983 to the present. This allowed me to check out any title or author
to see whether they had received reviews from the quality press, and also,
to read those reviews in order to determine the genre of the book. Given
the role of prestigious journals such as the Times Literary Supplement in
conferring legitimacy on high culture, this seems like a reasonable way of
determining the cultural status of contemporary books.
The percentages of pupils who had recently read a book by an author in
each category and who had a favourite book or author belonging to a given
category is shown in Table 3.3. In deciding whether a book counts as high
Category
Children’s books
Teenage books
Factual low culture
Adult low culture — including
much horror, science fiction,
fantasy and romance and some
factual books, e.g. travel
*Factual high culture books
*Classic
*Modern high culture — 20th
century
*Contemporary high culture —
books of the sort that receive
reviews in the quality press
Other
64
Examples
Roahl Dahl, Enid Blyton
Judy Blume, Point Horror series
Guinness book of records, books
about cars, etc.
Stephen King, Terry Pratchett,
Danielle Steele
Science, humanities, some
biographies, etc.
Dickens, Austen
Orwell, Graham Greene
Salman Rushdie, Martin Amis
Table 3.2: Coding of Books
or low culture I have erred on the side of charity, since anything outside
the lowest culture is quite rare. For instance, a fairly sophisticated science
fiction author like Philip K. Dick would count as high culture. For those
pupils that stated the authors and/or titles of any books, teenage books and
adult low culture predominated. When coding the data, I saw that the same
authors came up very regularly (Stephen King, Terry Pratchett, Danielle
Steele). “High culture” books (the four starred categories) were included in
the responses of only a small minority.
Pupils were given a “reading score”, consisting of points awarded for: reading
books unconnected with schoolwork once a month (1 point), reading such
books once a fortnight or more (2 points), library membership, having any
favourite author, having a “high brow” favourite author, “high brow” recent
reading, and reading a broadsheet newspaper. This gives a score out of 7.
The distribution of this score is shown in Figure 3.5.
Percentage with
a favourite
author in this
category
Classic
Contemporary high culture
Modern classic
Factual high culture
Adult low culture
Factual low culture
Teenage
Childrens
Other
None
65
Percentage who
had read a book
in this category
5.4
9.0
6.5
3.7
26.0
8.8
12.0
4.9
15.1
37.8
4.1
3.2
2.2
0.2
18.3
0.0
5.2
8.0
3.9
66.9
Table 3.3: Pupils’ reading habits
frequency
100
80
60
40
20
0
0 1 2 3 4 5 6 7 8
Figure 3.5: Histogram of reading cultural capital score
3.3.2
66
Viewing Habits
Television viewing was included in the questionnaire because this is an important part of the “cultural consumption” of most people. It also has the
advantage (like books and unlike theatre, art galleries, etc.) of not being any
more inaccessible to those who do not live in London than to those who do.
Pupils were asked “Which TV programmes do you watch regularly? Name
as many as you can.” Having asked pupils to list the television programmes
they watched, I was faced with the task of categorising these programmes
according to their cultural capital content. This is by its nature a somewhat
subjective task, and one feels less confident in categorising TV programmes
than categorising books, as the categories for TV programmes are not so well
established and accepted. There is no debate about which fiction counts as
“classic” (even if there is debate about whether a classic author is actually
any good). Even modern fiction is put into pretty clear categories by such
indicators as the publisher and the colour of the book’s spine, and whether it
is reviewed in the quality papers. The initial judgements about these books
(made by publishers, journalists, etc.) were perhaps no less subjective than
those I am now making about TV programmes. However, I do not have to
construct the categorisation of books myself. This categorisation is in a sense
a social fact. Since TV programmes are not publicly categorised in the same
way as books, there is more scope for disagreement with my categories.
I looked through some of the questionnaires and watched a bit of any programme that I could not categorise initially. The coding I used, and the
frequencies and proportions of pupils who had watched a programme in each
of the categories, are shown in Table 3.4.
I have divided the factual programmes according to their subject matter,
“science”, “arts” and “politics/humanities”; and according to their level of
sophistication. By “level of sophistication” I mean how in depth the programme is and how much attention it requires of the viewer.
The judgements that I have made about factual programmes are, of course,
open to question, but it is my judgements about non-factual programmes
that I am more concerned about. I think that people are more likely to
disagree in their judgements on non-factual programmes, as it is hard when
considering these programmes not to be swayed by the question of whether
or not one finds the programme enjoyable. The only non-factual category
that I have divided into sophisticated/non-sophisticated is comedy. I have
tried not to judge this distinction according to how funny I find particular
Category
*Science — sophisticated
*Science —
unsophisiticated
*Arts — sophisticated
*Arts — unsophisticated
*Politics, current affairs,
society and humanities —
sophisticated
*Politics, current affairs,
society and humanities —
unsophisticated
Other factual
*Literary adaptations
*Comedy — sophisticated
Comedy — unsophisticated
*Educational —
sophisticated
*Educational —
unsophisticated
Drama
Science Fiction
Soaps
Teen/kids
Sport
Examples
QED, Horizon, Equinox
Wildlife programmes
South Bank Show, Late
Review, Arena, Without
Walls
Film ’98 Bookworm
Newsnight, Panorama,
World in Action, Everyman
The news (except
Newsnight), Witness, The
Big Story Have I got News
for You, Rory Bremner
Crimewatch, Watchdog,
Food and Drink
Tom Jones, Dance to the
Music of Time, Pride and
Prejudice
Frasier, The Simpsons,
King of the Hill
Friends, Men Behaving
Badly, Only Fools and
Horses, Mr Bean
Open University
The learning zone
Casualty, ER, The Bill
Star Trek, X Files, Red
Dwarf
Brookside, Eastenders,
Neighbours
Hollyoaks, Blue Peter, The
Fresh Prince of Bel Air
football, athletics
Table 3.4: Coding of television programmes
67
Frequency
9
28
Percentage
2.0
6.2
3
0.7
4
11
0.9
14.7
66
14.7
99
22.0
1
0.2
133
29.6
228
50.8
4
0.9
3
0.6
177
126
39.4
28.1
307
68.4
313
69.7
130
29.0
68
programmes. Instead, I have tried to focus on the subtlety of the humour,
the cultural references used, and the vocabulary used. To illustrate, let me
take two programmes, Frasier and Shooting Stars. In Frasier, the humour
is based on dialogue. Sophisticated vocabulary is used. In parodying the
pretensions of the main characters, “high-brow” cultural references are made
(to Freud, the opera, etc). Shooting Stars, on the other hand, relies on
slapstick to a large degree. The vocabulary used is not sophisticated. The
cultural references are popular (often to pop music).
I considered the possibility of dividing drama into the categories of highbrow and popular. Clearly, a drama such as The Singing Detective would
fit into the former category, with things like The Bill and Casualty in the
latter. However, going through the TV schedules, I found that there were no
high-brow drama series on at the time I was conducting the survey (if one
excludes literary adaptations, which form their own category).
As in the case of books, and for the same reason, when in doubt about
whether to categorise a programme as low or high culture, I have erred on
the side of high culture. For instance, I have categorised Have I got News
for You as unsophisticated politics/current affairs, rather than as unsophisticated humour, thus giving credit for cultural capital. Having broken certain
categories down into “sophisticated” and “unsophisticated”, I found that
the frequencies of pupils watching these programmes were too low to justify
this subdivision. For instance, only 3 pupils stated that they watched sophisticated arts programmes, and 4 that they watched unsophisticated arts
programmes. So, in the final coding, these categories were collapsed. Pupils
were given a “TV viewing” score, consisting of points awarded for watching
any programme in each of the following categories: science, arts, politics,
educational, literary adaptations, sophisticated comedy, i.e. a mark out of
six.
3.3.3
Formal Culture and Music
Pupils were asked:
69
How often do you do each of these spare time activities (not organised
by the school) — (Often, Sometimes, Hardly ever, Never).
1. Going to art galleries or museums
2. Going to see plays
3. Going to classical concerts
4. Playing an Instrument
5. Listening to classical music
As expected, none of these activities was very popular among my sample.
Activity
Often
Art galleries and
museums
Plays
Classical concerts
Plays Instrument
Classical listening
Hardly
Ever
33.6
Never
0.9
Sometimes
13.8
4.9
0.4
14.5
3.9
17.0
4.5
14.0
9.6
38.5
16.9
13.2
16.8
39.6
78.1
58.3
69.7
51.7
Table 3.5: Formal culture and music
The responses for all these items are heavily skewed towards the “never” end
of the scale. This backs up my view that such questions are not a good way
of differentiating between 16 year old respondents (if indeed they are a good
tool for differentiating between levels of cultural participation of the adult
population).
Since more than half the respondents stated that they never went to art
galleries or museums, any other response (often, sometimes, or hardly ever)
was given one point for cultural capital. This pattern was followed for the
other items. In the case of going to plays, more than half responded “hardly
ever” or “never”, so a point was given for “often” or “sometimes”.
3.3.4
70
Parents’ Cultural Capital
Parents’ cultural capital was measured on a scale composed of items on:
the number of books in the home, whether a broadsheet newspaper is read,
subjects discussed in the home, participation in cultural activities, type of
music listened to and radio stations listened to.
The questions on subjects discussed in the home were as follows.
Which of the following have you heard your parents discuss? —
(Often, Sometimes, Never).
1. Art
2. Politics
3. Books
4. Science
5. Current Affairs
A point for cultural capital was given if the response put the respondent in
the top less than 50% of the sample.
The questions on activities were as follows:
71
For each of these activities, do your parents do it often, sometimes,
rarely or never?
• Going to art galleries or museums
• Reading novels
• Reading non fiction
• Going to see plays
• Going to concerts
• Playing a musical instrument
• Evening or daytime classes
The same pattern was followed in giving points for cultural capital — for each
item, a point was awarded if the level of participation put the respondent in
a minority, at the higher end of the spectrum.
Pupils’ were asked “Do your parents listen to the radio? If so, which station(s) do they listen to?” A mark for cultural capital was given for any one
of the following stations: Radio 3, Radio 4, Classic FM, Jazz FM, and the
World Service.
Pupils were asked “If your parents listen to music, what kind(s) of music do
they listen to?” Responses that fell within the categories of classical or jazz
were given a point for cultural capital.
Points for cultural capital were also given for having two or more cases of
books in the home, and for taking a broadsheet newspaper.
All the above items were summed to give an overall measure of parents’
cultural capital — a score out of 16.
3.3.5
Validity of Cultural Capital Questions
Criterion validity concerns whether a measure accurately reflects the concept
that it is intended to reflect. Of course, one major problem with assessing
72
the validity of a measure in sociology is that the theoretical concepts one is
dealing with are rarely precisely defined. This may be seen as a particular
problem in attempting to construct a measure of cultural capital. In addition, the assessment of criterion validity requires an independent measure
of the sociological construct in question. I certainly have no independent
measure of the cultural capital of my sample of pupils. It could perfectly
well be claimed that my operationalisation leaves out important elements of
cultural capital, or that it includes items that do not really reflect cultural
capital. The vagueness of the concept leaves plenty of scope for disagreement. However, I would argue that my aim is not simply to measure some
received and theoretically pre-defined notion of cultural capital and then attempt to correlate it with achievement. Rather, I aim to explore the notion
of cultural capital empirically. In order to examine whether the different
cultural resources that I have distinguished belong together in the same concept, and deserve the name of capital one must ask whether each of them is
associated with attainment. It will also be interesting to examine whether
individuals who score highly on one element of cultural capital tend to also
score highly on the other two. In other words, does possession of cultural
capital usually mean possession of a general culture comprising knowledge,
lifestyle and language (as Bourdieu’s writings would seem to suggest) or is
possession of cultural capital more fragmentary than this? (These questions
reflect construct validity rather than criterion validity).
There are specific problems that may be anticipated with some of the questions I have asked to assess possession of cultural capital. Pupils may have a
tendency to give the response that they think will be seen as desirable rather
than a completely honest response. For instance, pupils may overestimate
the frequency with which they participate in certain activities such as reading, listening to classical music, etc. When naming the TV programmes they
watch regularly, they may favour more high-brow programmes, even if they
do not really watch them regularly. (One girl exclaimed, “Panorama! I’ve
seen that once” while filling in this question.)
However, although it is possible that the level of these elements of cultural
capital will be exaggerated by my survey, this does not mean that one must
expect systematic distortion between social groups. E.g. It does not mean
that one must expect girls to exaggerate more than boys, or middle class
pupils to exaggerate more than working class pupils, or vice versa. Of course,
I cannot rule out systematic distortion (perhaps because girls are more eager
to please than boys, or because middle class pupils see high brow activities
as more desirable than working class pupils do). Even if there is system-
73
atic distortion between social groups, this would not imply distortion in the
patterns of association between variables.
I computed scores for parental cultural participation and pupils’ cultural
participation, giving a point for each item on the questionnaire that was
designed to be an indicator of cultural capital. The reliability coefficient
Cronbach’s α (Carmines and Zeller, 1979) for the parents’ (16 item) cultural
participation scale is α = 0.84. For the pupils’ (11 item) scale α = 0.67. Note
that the items these scales are composed of are not all on the same scale,
so the reliability statistic can be seen as a baseline estimate. In the case of
pupils, test scores for active and passive vocabulary and cultural knowledge
were also calculated. For the cultural knowledge score (25 items), α = 0.87.
The active and passive vocabulary scores were summed to create an overall
vocabulary score (20 items), for which α = 0.84. All the above scales are
approximately normally distributed. (Of course, these figures only reflect
internal reliability, or the level of association between items on the scale.
This is not to be confused with criterion validity).
3.3.6
Rational Choice Questions
In order to assess various rational choice theories of educational inequalities,
it is crucial to examine the beliefs and motivations of pupils.
Pupils’ beliefs about their own abilities were assessed using two questions,
“What grades do you think you are likely to get in your GCSEs?”, and “In
general, how do you rate your academic abilities as compared to other pupils
at your school?”(these were both multiple choice questions).
Pupils’ attitudes to education were assessed using a scale of 16 statements
about studying. Pupils were asked to give a response ranging from “agree
strongly” to “disagree strongly”. Items reflected both the labour market
value of education, e.g. “The more qualifications you get, the better the job
you are likely to get”, and the intrinsic or personal value of education e.g. “I
enjoy studying”.
Pupils were asked both “What sort of job would you like to be doing in 10
years time”, and “What sort of job do you think you will actually be doing
in 10 years time”. These were open response questions. The coding scheme
for this question was determined by the pupils’ responses. (See appendix D
for the coding frame). These responses can only be allocated to social class
74
categories very approximately.
Pupils were also asked “In choosing a job, what things about it do you think
are important”. This was a multiple choice response question.
3.3.7
Reliability of Rational Choice Questions
The questions on pupils’ expectations and aspirations, both educational and
occupational, would be invalid if pupils failed to put down what they truly
expect and aspire to. For instance, false modesty might prevent those who
expect to get mostly A grades at GCSE from saying so, or bravado might
prevent those who expect to fail from saying so. Similarly, pupils might
not be willing to admit that their aspirations are as high or as low as they
really are. However, given the confidentiality of the questionnaire, there is no
reason for pupils to present false expectations and aspirations. (Of course,
the fact that there is no good reason for distorting one’s answers will not stop
some people. For example, one girl said to me that she thought it would be
“tempting fate” to say she expected to get Bs and Cs.)
The questions about pupils’ attitudes towards education may be subject to
the problem of pupils giving a response that they perceive to be desirable
rather than a truthful response. Again, the confidentiality of the questionnaire should allay this fear to some extent. I have avoided the problem of
“yea saying” by using a mixture of positive and negative statements. These
questions were used to construct two scales for pupils’ evaluation of the intrinsic value of education, and of the instrumental value of education in the
job market.
The reliability for the whole 16-item scale is α = 0.80. The reliability for
the 7-item sub-scale for attitudes to education as an intrinsic good is α =
0.73. The reliability for the 5-item sub-scale for attitudes to education as an
instrumental good is less high, at α = 0.43. Nevertheless, I think that use
of this sub-scale is justified in that it meets my theoretical purpose. Also,
one must bear in mind that reliability analysis is based on an assumption
that the responses for each item are normally distributed. This assumption
is strongly violated in the case of this educational attitude scale.
3.3.8
75
Parents’ Social Class and Qualifications
Parents’ social class was determined from pupils’ responses to the following
questions regarding each parent.
• Please tell me about your parents’ jobs. If they are not working at the
moment, please tell me about their most recent jobs.
• What is the name of the job?
• What kind of work do they do?
• What sort of organisation do they work for?
• Do they have their own business?
The responses were coded using the Goldthorpe class schema and the Standard Occupational Classification (Office of Population Censuses and Surveys,
1991) class schema. The main reason for using the Goldthorpe schema was
comparability with other quantitative work in the field of sociology of education, and this is the schema that I have used in most of the analysis. I
collapsed the schema into six categories in order to avoid small cells, due to
my small sample size.1 I took mother’s or father’s class, whichever was the
higher, as determined by a simplified version of Erikson’s (Erikson, 1984)
dominance schema2 . I tried using a composite measure such as that suggested by Heath and Britten (Heath and Britten, 1984), and found that the
model fit was generally extremely similar to that gained using the dominance
approach.3 The social class breakdown of the sample is given in Table 3.6.
1
• I = 1 (service class, higher)
• II = 2 (service class, lower)
• IIIa + IIIb = 3 (non-manual)
• IVa+IVb + IVc = 4 (small proprietors)
• V + VI = 5 (skilled manual)
• VII = 6 (unskilled manual)
2
1 dominates 2, 2 dominates 4, 4 dominates 3, 3 dominates 5, 5 dominates 6.
However, the results were harder to interpret because, given my small sample, It was
not desirable to increase the number of class categories. Having used a four category
schema to derive a ten category composite schema, and then having collapsed these cat3
76
Parents’ qualifications were determined as follows. Pupils were asked “What
is the highest qualification that each of your parents has?” A set of multiple
choice boxes followed for both “mother” and “father”. The options were:
• No qualifications
• CSE
• Olevel
• A level/Scottish higher
• Vocational qualification (please specify)
• Degree
• Postgraduate qualification (masters, Ph.D. or D. Phil.)
Vocational qualifications that were equivalent to a degree or postgraduate
qualification were recoded as such. Mother’s or father’s qualifications were
selected according to which was higher. The resulting schema was then
collapsed into a four-category schema. 1. No qualifications or CSE, 2. O
level, 3. Intermediate (A level or vocational), 4. Degree (undergraduate or
postgraduate).
3.4
Sample
I surveyed 465 pupils. Clearly, from the point of view of minimising standard
error, the larger the sample the better. However, I was clearly limited by time
and other resources. There are two main issues here. Firstly, getting access
to schools is difficult and takes time. Secondly, the data must be coded and
inputted. I would have had difficulty both in getting a much larger sample,
and in dealing with it once I had it.
I have carried out my survey at two Oxfordshire comprehensives, and two
London comprehensives. Both cultural reproduction theory and rational
choice theory are concerned with general processes, which are not contingent on any particular school context. Therefore, these theories could be
egories further to derive a six class schema, the meaning of the resulting classes was far
from intuitive.
77
tested using any type of school. While a representative sample of the national year 11 population would be ideal, it would clearly be impossible for
me to collect a sufficiently large sample to achieve this. What is really crucial
for me is to have a good spread of pupils in terms of ability, social class, etc.,
rather than to have a sample that is representative. The schools I have chosen will inevitably be unrepresentative in some respects. However, it should
be borne in mind that I am not attempting to make population estimates,
but rather to examine processes that the theories under examination suggest
should operate right across the educational system
The reason for only using comprehensive schools is that, while I want my
sample to contain a good spread in terms of characteristics such as social
class and ability, I do not wish to examine school type effects, and therefore prefer to keep the variable “school type” constant as far as possible. Of
course, the problem with this is that one cannot get a representative sample while excluding certain types of school. Pupils attending private schools
in particular represent a significant section of the student population, and
may be expected to differ from state school pupils in many relevant respects
(e.g. the importance of cultural vs. economic capital in their academic attainment). However, given the necessarily small size of my sample, I felt
that school type effects really did have to remain outside the scope of my
thesis as far as possible. Furthermore, the comprehensive sector is extremely
important in its own right, as a large majority of British secondary school
pupils attend comprehensive schools.
School 1 is a voluntary aided boys’ comprehensive (126 boys in year 11).
School 2 is school 1’s “sister school”. It is a girls’ comprehensive which
maintains a joint sixth form with school 2. School 3 is the pilot school.
School 4 is a large comprehensive in Oxfordshire (213 pupils in year 11)
that takes both boys and girls. Clearly, it is important for my sample to be
reasonably mixed between boys and girls, given that some of my hypotheses
concern gender. I have achieved this balance by using one boys’ and one
girls’ school, as well as two co-ed schools. It might be argued that, ideally, I
should have used only co-ed schools in order to avoid the complication of the
possible “school effects” of single sex schools. However, the issue of possible
differences in attainment between pupils at mixed and single-sex schools is
highly contentious. Even if there were large differences in attainment, this
would not imply that the relationships between the variables I am considering
would vary between single sex and mixed schools. In addition, given the
difficulties involved in getting access to schools, I think it would have been
foolhardy to reject the option of using single-sex schools.
78
The social class breakdown of the schools is given in Table 3.6. School 1 and
School 2 have a large proportion of service class families (44.8% and 42.0%
respectively), compared to schools 3 and 4 where 14.7% and 30.8% per cent
of families respectively are categorised as belonging to the service class. The
proportion of families categorised as belonging to the skilled or unskilled
manual classed is higher in schools School 3 and School 4(34.6% and 28.6%
respectively) than in School 1 and School 2(10.5% and 11.0% respectively).
School
Parent’s class
Missing
Service class
— higher
Service class
— lower
Routine
non-manual
Petty
bourgeois
Skilled
manual
Manual
1
2
3
4
f
17
26
%
16.2
24.8
f
20
27
%
20.0
27.0
f
8
5
%
10.7
6.7
f
12
20
%
6.5
10.8
Total
f
%
57 12.3
78 16.8
21
20.0
15
15.0
6
8.0
37
20.0
79
17.0
13
12.4
14
14.0
26
34.7
44
23.8
97
20.9
17
16.2
13
13.0
4
5.3
19
10.3
53
11.4
5
4.8
4
4.0
13
17.3
28
15.1
50
10.8
6
5.7
7
7.0
13
17.3
25
13.5
51
11.0
Table 3.6: Social class breakdown for each school
Another factor in choosing which schools to approach was location, since I
do not have a car.
Getting access to schools was difficult. Although I approached several schools
by writing to the head, only School 4 was successfully approached in this way.
The other schools were approached through personal contacts.
3.4.1
Administration of questionnaire
I administered a self-completion questionnaire to whole class groups of year
11 pupils. I decided to do this rather than interview pupils individually,
since this approach allows one to get a larger sample. The questionnaire
took pupils about 45 minutes to complete.
79
In all schools, the questionnaire was administered during a “PSE” (Personal
and Social Education) period. Pupils were not allowed to confer while completing the questionnaire, but were able to ask either a teacher or me if they
did not understand any of the questions. Pupils and schools were, of course,
assured of the confidentiality of their responses.
I introduced the questionnaire was there to answer questions while the students were completing the questionnaire. I introduced the questionnaire
something along the lines of the following:
My name is Alice Sullivan and I am a research student from Oxford
University. The questionnaire I would like you to fill in asks questions about things like your views about education, your plans for
the future, and what you do in your spare time. There are also some
questions about your family. The reason for these questions is that
the circumstances in which people live, and the people they live with
can have an important influence on how they do at school and on
the decisions they make about what to do after leaving school. Everything in the questionnaire will be treated as confidential. I won’t
pass any information about you on to the school or your parents. The
only reason that you have to be identified on the questionnaire at all
is so that I can follow up on what GCSEs you get and what you do
after your GCSEs. Most of the questions are answered by ticking a
box. If you have any problems with any of the questions, just put
your hand up and ask.
The fact that I had to identify students in order to follow up on their GCSE
results and on what they do after GCSEs posed a problem, as it meant that
the questionnaires could not be anonymous. There are two alternative ways
of dealing with this. One is simply to ask pupils to put their names on the
questionnaire. The disadvantage of this is that pupils may be concerned
that the questionnaire could be seen by people other than the researcher,
who would then easily be able to identify individuals. The other alternative
is to number the questionnaires and have a list attaching each pupil’s name
to a number. This is the system which I agreed to use at School 4. At first
I intended to hand out the questionnaires according to a pre-decided list of
names and numbers. However, this did not work out because I could not get
a list of which pupils would be there on any given day out of the school. On
my first trip to the school, I ended up asking students to write their names on
80
the questionnaires. Thereafter, I numbered the questionnaires and asked the
teachers present to make a note of who had been given each number as they
handed the questionnaires out. However, even this was rather more timeconsuming than I had expected. I decided that the disadvantage of using
named questionnaires was outweighed by the practical problems associated
with numbering. Therefore, I was relieved to find that the heads of the other
schools had no objection to pupils being asked to provide names.
3.5
Response Rates
In three out of the four schools, the entire year group was surveyed. In
the remaining school, for time-tabling reasons, five out of seven forms were
surveyed. Out of a potential sample of 557 pupils, 465 questionnaires were
adequately completed, giving a response rate of 80%. The bulk of the nonresponse was due to absenteeism (17%), with the remaining 3% due to noncompletion or inadequate completion of the questionnaire.
The level of missing data on social class is 12% (57 cases)4 . This is mainly
because many students did not respond to the question on their parents’
occupations in sufficient detail for the responses to be categorised. In the
case of parents’ qualifications, this problem is still more severe (122 missing
cases). This is almost certainly due to pupils’ simply not knowing what
qualifications their parents had. This level of missing data certainly reflects
a problem with using children as a proxy source of information on their
parents. Many children have no clearer idea of what their parents do for a
living than “works in an office” for instance, and many children may never
have talked to their parents about what qualifications they have. This lack of
information may be informative in itself, perhaps revealing a lack of “social
capital” within the home in Coleman’s sense (see for instance, (Coleman,
1990)). Therefore, I have included these missing cases within my analyses as
separate categories.
4
This is a normal level of non-response in national surveys where people are asked
about their parents’ occupations, such as the British Election Survey.
3.6
81
Ethics
The confidentiality of participating individuals and schools has been and will
be preserved. No data relating to individuals has been or will be passed on
to third parties.
Pupils were not asked whether they wanted to complete the questionnaire.
However, if any of them had objected to answering specific questions, they
would not have had to do so. No one objected verbally, but some pupils
did not fill in sections of the questionnaire. One pupil refused to fill in the
questionnaire at all, and a few gave false names or no name. I told pupils that
I would be looking at their GCSE results, and there were no objections to
this. This is perhaps unsurprising given that GCSE results are not generally
kept private within schools.
3.7
Conclusions
Responses on the cultural participation of parents and pupils and the linguistic abilities and cultural knowledge of pupils will allow me to assess:
• Whether cultural capital is passed down from parents to children
• Whether middle class people have higher levels of cultural participation
than working class people, and whether male or female respondents
have higher levels of cultural participation.
• Whether cultural participation affects educational attainment
• Which elements of cultural participation are most important in determining attainment
• Whether the effect of cultural participation on educational attainment
is direct, or mediated by cultural knowledge and linguistic ability
Responses on the attitudes and aspirations of pupils will allow me to assess
the impact of class and gender on decision making in educational careers.
• By comparing pupils’ self-predicted GCSE grades and view of their own
academic abilities to actual GCSE results I will assess whether working
82
class and female pupils underestimate their abilities as compared to
middle class and male pupils.
• By surveying pupils on their assessment of the intrinsic and instrumental value of education, I will be able to assess whether working class
and middle class, and male and female pupils differ in their assessments
of the worth of education both in the job-market and more generally.
• I will be able to assess whether occupational aspirations vary by class
and gender once GCSE performance has been controlled for
• I will be able to assess whether educational plans vary by social class
and gender once GCSE results are controlled for, and to examine possible mechanisms behind this, such as gender and class differences in
subjectively perceived ability, attitudes to education, and occupational
aspirations.
Chapter 4
Cultural Capital and
Educational Attainment at
GCSE
4.1
Introduction
This chapter will assess the merits of the cultural reproduction approach to
the examination of class and gender differentials in educational attainment.
According to Bourdieu’s theory of cultural reproduction, children from middle class families are advantaged in gaining educational credentials due to
their possession of cultural capital. In order to assess this theory, I have
developed a broad operationalisation of the concept of cultural capital, and
have surveyed pupils on both their own and their parents’ cultural capital.
In chapter 1 I argued that for Bourdieu’s theory to be backed empirically, he
would need to show that:
1. parental cultural capital is inherited by children.
2. children’s cultural capital is converted into educational credentials.
3. educational credentials are a major mechanism of social reproduction
in advanced capitalist societies.
In this chapter, I will address the following questions:
83
CHAPTER 4. EDUCATIONAL ATTAINMENT AT GCSE
84
• How cultural capital is distributed according to social class and educational level.
• The extent to which cultural capital is passed down from parents to
children.
• Whether male and female pupils possess different levels of cultural capital.
• What effect cultural capital has on GCSE attainment at age 16.
4.2
Analysis
I have used linear regression to examine the associations between pupils’ and
parents’ cultural capital, and between cultural capital and GCSE attainment.
I have used nested models in order to show the pattern of effects in greater
detail, so that both direct and indirect effects of explanatory variables on
the outcome variable are apparent. The order in which variables have been
inserted into these nested models is to some degree arbitrary, but has been
chosen in accordance with the theoretical framework I am using, and the
questions I am trying to answer. For instance, social class is always included
in the first model because I am interested in seeing whether the effect of social
class can be explained by other variables, which are inserted in subsequent
models.
4.2.1
Parental Cultural Capital
The first step in assessing the theory of cultural reproduction is to look at
the distribution of cultural participation by social class. (The parental cultural participation variable has mean 4.78 and standard deviation 3.89). In
figure 4.1 we can see a clear positive relationship between social class and
parental cultural participation. Similarly, taking mother’s or father’s educational level, whichever is higher, there is a positive relationship between
parents’ qualifications and their cultural participation — see figure 4.2. Parents with degrees have particularly high levels of cultural participation.
85
Parents’ cultural participation
10
8
6
...........
....
...
...
............................................
....
...
....
....
....
....
....
...
....
...
....
....
...........
....
.....
....
....
...
...
.............................................
....
....
....
............................................
....
....
...
....
..
...
................................................
............
....
.
.....
....
....
....
....
....
....
...
...
..
....
.
....
...........................................
....
....
....
...
....
....
....
... ............................................
..
.
.
.
.
.
..........
.................................................
.............................................
.
.
.
.
.....
.....
.....
.............................................
...
....
....
................................................
....
.....
....
...
..
.
....
....
....
...
.............................................
....
...
....
...
.... ............................................
....
...
....
...
....
....
....
....
..
.
.
.
.
...........
................................................
...........................................
.
.....
...
....
....
....
....
...
....
.....................................................................................................................................................................................................................................................................................................................................................................................................................................
r
r
r
r
r
Skilled manual
(50)
Manual (51)
Routine
non-manual (97)
Petty bourgeois
(53)
Service class —
lower (79)
Missing (57)
2
r
r
Service class —
higher (78)
4
Figure 4.1: Parents’ cultural participation by parents’ class (95% confidence
intervals)
4.2.2
Pupils’ Cultural Capital
Having established an association between parental social class and cultural
participation, we can move on to the question of whether cultural capital is
transmitted within the home. To what extent is parental cultural participation associated with pupils’ cultural participation, controlling for background
variables?
Pupils’ Cultural Activities
I used linear regression to analyse the determinants of the activities component of pupils’ cultural capital. For each parameter, the B statistic is
Intercept
Parents’
qualifications
Missing
Degree
Intermediate
O-level
CSE or no
qualifications
Parents’ class
Missing
Service class higher
Service class lower
Routine non
manual
Petty bourgeois
Skilled manual
Unskilled
manual
Sex
Male
Female
School
School 1
School 2
School 3
School 4
Parents’ cultural
participation
N
η 2 for the model
Model 1
B(s.e)
4.16(0.42)***
***
η
0.177
0.047
0.13(0.35)
1.71(0.42)***
0.43(0.41)
0.53(0.36)
0.000
0.036
0.002
0.005
*
0.14(0.45)
1.35(0.46)**
2
Model 2
B(s.e)
3.33(0.38)***
86
η2
0.148
0.007
−0.12(0.30)
0.41(0.38)
0.24(0.36)
0.21(0.32)
0.000
0.003
0.000
0.001
0.035
0.000
0.019
0.16(0.39)
0.29(0.41)
0.013
0.000
0.001
0.48(0.43)
0.003
−0.88(0.38)
0.000
−0.18(0.40)
0.000
−0.37(0.35)
0.002
0.28(0.46)
0.28(0.46)
0.001
0.001
−0.17(0.40)
0.23(0.40)
0.000
0.001
−0.46(0.28)
0.006
0.006
−0.34(0.25)
0.004
0.004
*
0.45(0.33)
0.41(0.34)
−0.66(0.32)*
0.023
0.004
0.003
0.009
0.29(0.29)
−0.05(0.30)
−0.24(0.28)
0.005
0.002
0.000
0.002
0.35(0.03)***
0.233
464
0.218
Table 4.1: Pupils’ cultural participation
464
0.400
87
Parents’ cultural participation
10
...........
....
...
...
....
.....................................................................
....
...
....
....
....
....
...
....
....
...
...........
...
.....
.....................................................................
....
...
....
....
....
....
...
..
............
.....
....
...
.....................................................................
....
...
....
....
....
....
.....................................................................
....
...
....
...
....
..
..
.
.
....
.
.......... .....................................................................
.
.
..
.
.
.....................................................................
.....
.....
.....
.
...
....
.
.
.
....
.
...
....................................................................
..................................................................
....
...
....
.... .....................................................................
....
....
...
....
...
....
....
.
.
....................................................................
.............
.
.....
...
....
....
....
....
...
....
.....................................................................................................................................................................................................................................................................................................................................................................................................................................
r
8
6
4
r
r
r
Degree (105)
Intermediate
(61)
O-level (109)
Missing (122)
2
None/CSE (68)
r
Figure 4.2: Parents’ cultural participation by parents’ educational level (95%
confidence intervals)
given, followed by the standard error. In addition, the η 2 statistic is shown.
This describes the proportion of total variability in the dependent variable
attributable to the variation in the independent variable. It is the ratio of
the between groups sum of squares to the total sum of squares. Significance
levels are denoted by * if p ≤ 0.05, ** if p ≤ 0.01, and *** if p ≤ 0.001.
Table 4.1 shows two models. Model 1 shows the effects of parents’ qualifications, parents’ class, pupils’ gender and school attended on pupils’ cultural
activities. All of these variables except gender have significant effects (at the
0.05 level) in this model. Having a parent who is a graduate and having a
parent who is in the higher service class are significantly positively associated
with pupils’ cultural activities. Parents’ cultural participation is introduced
in Model 2. This shows that parents’ cultural participation (with an η 2 of
0.233) is by far the most important factor in accounting for the variation in
88
pupils’ cultural activities. (The η 2 statistic describes the proportion of total
variability in the dependent variable attributable to the variation in the independent variable. It is the ratio of the between groups sum of squares to
the total sum of squares).
Neither social class nor educational credentials are significant once parental
cultural participation has been included. This shows that the effect of these
background variables on pupils’ cultural activities is entirely mediated by
parents’ cultural participation. The effect of school attended is quite insignificant once parental cultural capital is taken into account. The absence
of a school effect is important, as a crucial claim about cultural capital is that
it is not transmitted by the school. (However, bear in mind the small number
of schools in my sample, and that these are all comprehensive schools.)
The Pearson correlation between parents’ cultural capital and pupils’ cultural
activities is 0.617 (p ≤ 0.001). The strength of this relationship provides
support for Bourdieu’s view that cultural resources are strongly transmitted
from parents to children.
4.2.3
Language and Knowledge
Next, I modelled the pupils’ tested vocabulary and cultural knowledge scores.
Parental cultural participation mediates the background variables to some
extent, but not to the same extent as in the case of the activities component
of cultural capital. This is unsurprising, as the parental cultural participation
score is composed of similar items to the activities component of pupils’
cultural capital, whereas I have no direct measure of parental vocabulary or
cultural knowledge.
I modelled the pupils’ vocabulary score in stages (see table 4.2), first of all
just including the background variables — parents’ qualifications, parents’
class, gender and school. Model 1 shows that gender and school are insignificant. These variables have overall significance values of 0.680 and 0.357
respectively. (By overall significance values I mean the significance value for
the variable as a whole rather than for each parameter of the variable.)
In Model 2, parental cultural capital is added to the model. This shows that
the effects of parents’ social class and qualifications are partially mediated
by parental cultural capital. The effect of parents’ qualifications is reduced
from an η 2 of 0.041 in model 1 to 0.024 in model 2. The effect of social class
Intercept
Parents’
qualifications
Missing
Degree
Intermediate
O-level
CSE or no
qualifications
Parents’ class
Missing
Service class lower
Routine non
manual
Petty bourgeois
Skilled manual
Unskilled
manual
Sex
Male
Female
0.97(1.55)
−0.58(1.54)
−0.28(0.96)
0.055
0.003
0.025
0.006
0.000
0.002
0.000
0.000
0.000
***
−1.78(1.53)
5.34(1.58)**
2.37(1.49)
0.38(1.37)
1.53(1.56)
−0.52(1.56)
−0.42(0.98)
0.000
0.000
0.001
0.000
0.000
0.003
0.035
0.003
0.014
0.001
0.013
0.016
0.008
η
0.178
0.023
2
−0.01(0.94)
1.11(1.52)
−0.76(1.51)
0.43(1.33)
1.74(1.45)
*
−1.89(1.48)
3.80(1.56)*
1.05(1.15)
3.24(1.44)*
3.71(1.37)**
2.13(1.21)
Model 3
B(s.e)
11.67(1.54)***
Table 4.2: Language score (continued overleaf. . . )
0.14(1.35)
1.67(1.48)
*
−1.76(1.51)
4.03(1.60)*
0.95(1.17)
3.57(1.47)*
3.73(1.39)**
2.30(1.23)
0.002
0.029
0.020
0.010
1.26(1.19)
5.18(1.42)***
4.24(1.41)**
2.70(1.24)*
Model 2
B(s.e)
14.32(1.45)***
*
η
0.200
0.038
2
Model 1
B(s.e)
15.35(1.45)***
**
0.000
0.000
0.001
0.001
0.000
0.003
0.035
0.004
0.013
0.002
0.011
0.016
0.007
η
0.113
0.021
2
−0.61(0.97)
0.68(1.56)
−0.91(1.51)
0.05(1.33)
0.99(1.46)
−2.26(1.50)
2.90(1.59)
0.86(1.16)
3.06(1.46)*
3.60(1.37)**
2.04(1.21)
Model 4
B(s.e)
12.37(1.57)***
0.001
0.001
0.000
0.001
0.000
0.001
0.028
0.005
0.008
0.001
0.010
0.016
0.007
η2
0.127
0.021
School
School 1
School 2
School 3
School 4
Parents’ cultural
participation
Pupils’ cultural
participation
Reading
Formal cultural
particpation
Music
TV
N
η 2 for the model
η
0.007
0.006
0.001
0.000
2
464
0.189
0.44(0.12)***
−2.05(1.10)
−1.51(1.14)
0.43(1.08)
Model 2
B(s.e)
0.030
η
0.013
0.008
0.004
0.000
2
0.043
0.80(0.18)***
464
0.224
0.003
η
0.016
0.010
0.004
0.001
2
0.16(0.13)
−2.28(1.08)*
−1.47(1.12)
0.62(1.05)
Model 3
B(s.e)
Table 4.2: Language score (. . . continued from overleaf)
464
0.164
−1.86(1.12)
−0.94(1.15)
−0.08(1.08)
Model 1
B(s.e)
0.019
0.000
0.000
0.040
−0.07(0.55)
2.24(0.53)***
448
0.239
0.008
η2
0.020
0.011
0.005
0.002
0.71(0.24)**
−0.01(0.49)
0.26(0.14)
Model 4
B(s.e)
*
−2.36(1.10)*
−1.65(1.13)
0.88(1.07)
91
is reduced from an η 2 of 0.065 to 0.043. I am referring here to the η 2 statistic
for the variables as a whole, i.e. the proportion of total variability in pupils’
vocabulary scores attributable to the variation in parents’ qualifications and
social class. Model 3 shows that the effect of parental cultural capital is in
turn mediated by the activities component of pupils’ cultural capital. This
leaves parental social class and pupils’ cultural activities accounting for very
similar proportions of the variation in pupils’ language score.
The next step is to break the measure of pupils’ cultural activities down
into its constituent parts in order to determine which cultural activities are
associated with pupils’ vocabulary score. (The overall score is removed from
the model before inserting the component scores). On Crook’s (Crook, 1997)
view that public cultural participation serves to communicate status, whereas
reading helps to develop abilities, reading should be positively associated with
vocabulary and formal culture should not. And indeed, this is the case. But
reading is not the only form of cultural participation that is positively and
significantly associated with pupils’ vocabulary. In fact TV viewing habits
account for a greater proportion of the variation in pupils’ vocabulary than
does reading. The “music” variable however, (whether a pupil listens to
classical music and/or plays an instrument) is not significant.
Using the same procedure for pupils’ cultural knowledge (see table 4.3),
Model 1 shows the background variables. In this model, gender is insignificant, but parents’ qualifications, parents’ social class and school attended are
all highly significant. Higher service class backgrounds are significantly associated with cultural knowledge. Graduate parents are particularly strongly
associated with cultural knowledge, but intermediate and O level qualifications are also significant.
Model 2 shows that again, parental cultural participation partially mediates
the background variables. The effect of a higher service class background is
rendered insignificant in this model, and the overall effect of parents’ qualifications is reduced from η 2 = 0.073 to η 2 = 0.031. Model 3 shows that
the effect of parental cultural participation is itself partially mediated by
the activities component of pupils’ cultural capital. However, the direct effect of parents’ cultural participation is still highly significant in this case.
Gender becomes significant, with a small advantage in favour of boys, once
pupils’ cultural activities are included in the model. This means that all the
variables in Model 3 have significant effects.
Again, I broke down pupils’ cultural activities to see which elements of this
measure are actually doing the work. I found the same pattern as for pupils’
Intercept
Parents’
qualifications
Missing
Degree
Intermediate
O-level
CSE or no
qualifications
Parents’ class
Missing
Service class lower
Routine non
manual
Petty bourgeois
Skilled manual
Unskilled
manual
0.008
0.067
0.014
0.015
0.065
0.012
0.015
0.006
0.004
0.003
0.000
1.15(0.62)
4.25(0.75)***
1.86(0.74)*
1.71(0.65)**
***
−1.91(0.80)*
2.20(0.83)**
1.27(0.78)
0.94(0.72)
1.01(0.82)
0.04(0.82)
0.56(0.79)
−0.01(0.79)
0.75(0.69)
0.70(0.76)
**
−1.90(0.77)*
1.14(0.82)
0.90(0.60)
2.94(0.75)***
1.45(0.71)*
1.38(0.63)*
Model 2
B(s.e)
6.27(0.74)***
**
0.001
0.000
0.003
0.002
0.043
0.013
0.004
0.005
0.033
0.009
0.011
η
0.137
0.034
2
0.62(0.78)
−0.10(0.77)
0.89(0.68)
0.73(0.74)
**
−1.96(0.76)**
1.02(0.80)
0.95(0.59)
2.79(0.74)***
1.44(0.70)*
1.30(0.62)*
Model 3
B(s.e)
5.00(0.79)***
**
0.001
0.000
0.004
0.002
0.048
0.015
0.004
0.006
0.031
0.009
0.010
2
η
0.082
0.031
Table 4.3: Cultural knowledge scores (continued overleaf. . . )
η
0.164
0.072
2
Model 1
B(s.e)
7.11(0.76)***
***
0.52(0.80)
−0.10(0.78)
0.81(0.69)
0.62(0.75)
**
−2.06(0.77)**
0.98(0.82)
0.98(0.60)
2.58(0.75)**
1.43(0.70)*
1.28(0.62)*
Model 4
B(s.e)
5.08(0.81)***
*
0.001
0.000
0.003
0.002
0.047
0.016
0.003
0.006
0.027
0.010
0.010
η2
0.085
0.027
Sex
Male
Female
School
School 1
School 2
School 3
School 4
Parents’ cultural
participation
Pupils’ cultural
participation
Reading
Formal cultural
particpation
Music
TV
N
η 2 for the model
0.045
0.000
0.015
0.020
η
0.004
0.004
2
0.074
0.36(0.06)***
464
0.338
0.025
0.000
0.008
0.012
η
0.006
0.006
2
**
0.01(0.56)
1.09(0.58)
−1.29(0.55)*
0.80(0.49)
Model 2
B(s.e)
0.038
0.38(0.09)***
464
0.363
0.024
0.024
0.000
0.008
0.011
η
0.008
0.008
2
0.22(0.07)**
*
−0.10(0.56)
1.11(0.57)
−1.19(0.54)*
0.93(0.49)
Model 3
B(s.e)
Table 4.3: Cultural knowledge scores (. . . continued from overleaf)
464
0.286
***
0.17(0.59)
1.56(0.60)**
−1.70(0.57)**
0.68(0.51)
Model 1
B(s.e)
0.035
0.000
0.001
0.022
−0.15(0.28)
0.85(0.27)**
448
0.369
0.029
0.022
0.000
0.006
0.012
η2
0.005
0.005
0.49(0.12)***
−0.10(0.25)
0.25(0.07)***
*
−0.26(0.57)
0.92(0.58)
−1.25(0.55)*
0.70(0.50)
Model 4
B(s.e)
94
vocabulary. Reading has a significant association with pupils’ cultural knowledge. Participation in formal culture does not. The “music” variable is
insignificant, whereas television viewing habits are significant.
These findings support the view that participation in formal or public culture does not foster the intellectual resources that may give an advantage
at school, and that reading does foster these resources. However, reading is
not the only cultural activity that is associated with linguistic ability and
cultural knowledge. Watching relatively sophisticated programmes on TV is
also associated with these skills. Of course, these associations cannot tell us
whether reading and watching sophisticated TV programmes foster knowledge or whether pupils’ reading and TV viewing habits simply reflect their
level of intellectual ability. It seems highly likely that both of these processes
occur. Ideally, one would control for measured ability at a given age (say 11
or younger) and then examine whether cultural participation has an effect
on later performance in tests of ability and examinations controlling for the
earlier ability score.
4.2.4
Gender
Gender does not account for a significant proportion of the variance in pupils’
activities or pupils’ vocabulary score, and only has a significant effect on
pupils’ cultural knowledge once participation in cultural activities is controlled for. However, there are small differences in the average level of cultural capital of girls and boys. These differences generally favour girls as is
shown in table 4.4.
Male
Female
Total
µ
σ
µ
σ
µ
σ
Max
Reading
3.35
1.73
3.46
1.70
3.40
1.71
8
Other Activities
1.40
1.28
1.72
1.30
1.55
1.30
9
Knowledge
10.22
4.67
10.05
5.07
10.14
4.86
25
Language
18.10
8.56
18.86
8.87
18.46
8.71
40
Table 4.4: Gender differences in cultural capital (µ=mean, σ=standard deviation)
95
Girls have slightly more cultural capital than boys in terms of both reading
and other activities, and score more highly on the language test. Boys,
however, slightly outperform girls on the test of cultural knowledge. Of
these differences, only the difference in cultural activities other than reading
is actually significant at the 0.05 level.
4.2.5
GCSE attainment
Finally, what impact does cultural capital have on grades achieved in the
GCSE examinations? I have modelled GCSE results using a point score for
the total of GCSEs gained — giving 1 point for a G grade, 2 for an F etc..
This point score is approximately normally distributed.
The effects of the background variables on pupils’ GCSE scores are shown by
Model 1 (see table 4.5). Compared to unskilled manual backgrounds, all nonmanual backgrounds are associated with increased GCSE performance, with
higher service class backgrounds providing the strongest advantage. Parents’
qualifications in the intermediate category (A level or vocational) and at
degree level were significantly associated with GCSE scores. Family structure and the number of siblings a pupil has are also included in this model.
(These variables have been excluded from previous analyses simply due to
a desire not to overload the analyses with variables.) Being in an “intact”
family (i.e. two original parents as opposed to a step- or single parent family) is quite significantly positively associated with GCSE performance. The
number of siblings has a significantly negative effect (i.e. the greater the number of siblings, the lower the average GCSE performance). Model 2 shows
that the effects of the background variables are mediated to an extent by
parental cultural participation. For instance, the overall effect of parents’
class on pupils’ GCSE attainment is reduced from an η 2 of 0.094 in Model
1, to an η 2 of 0.067 in Model 2. The effect of having a graduate parent is
rendered insignificant in this model, although intermediate qualifications are
still significantly positive. Model 3 shows that the effect of parental cultural
participaiton on pupils’ GCSE scores is partially mediated by the activities
component of pupils’ cultural capital. In this model, all the variables are
significant.
Breaking down pupils’ cultural activities into formal, reading, music and TV,
as before, we can see that the effect of reading is significant, and the effect
of participation in formal culture is insignificant. TV viewing habits are also
significant (though just barely at the 0.05 level), and music is not significant.
Intercept
Parents’
qualifications
Missing
Degree
Intermediate
O-level
CSE or no
qualifications
Parents’ class
Missing
Service class lower
Routine non
manual
Petty bourgeois
Skilled manual
Unskilled
manual
Sex
Male
Female
η
0.145
0.057
0.006
0.012
0.016
0.002
0.094
0.000
0.060
0.036
0.023
0.023
0.003
0.017
0.017
−3.76(2.36)
6.40(2.85)*
7.48(2.81)**
2.22(2.47)
***
0.18(3.08)
16.65(3.17)***
11.99(2.99)***
8.84(2.76)**
10.16(3.19)**
3.67(3.14)
**
−5.41(1.95)**
2
Model 1
B(s.e)
27.42(3.21)***
***
0.016
0.016
0.018
0.003
0.021
0.028
0.067
0.000
0.039
0.009
0.002
0.011
0.001
η
0.122
0.046
2
*
−4.50(1.87)*
9.26(3.06)**
3.31(3.00)
8.76(2.64)**
10.54(2.88)***
***
−0.16(2.94)
13.03(3.10)***
−4.49(2.26)*
1.68(2.84)
5.85(2.70)*
1.10(2.37)
Model 3
B(s.e)
19.53(3.32)***
**
0.013
0.013
0.021
0.003
0.025
0.030
0.073
0.000
0.039
0.009
0.001
0.011
0.001
η
0.075
0.044
2
Table 4.5: GCSE results (continued overleaf. . . )
**
−5.07(1.90)**
8.77(3.13)**
3.50(3.06)
8.15(2.69)**
10.29(2.94)**
***
−0.05(3.01)
13.33(3.16)***
−4.59(2.31)*
2.48(2.89)
6.04(2.76)*
1.37(2.42)
Model 2
B(s.e)
24.62(3.18)***
***
**
−5.26(1.94)**
8.77(3.19)**
3.29(3.05)
8.21(2.69)**
9.68(2.94)**
***
−0.06(3.02)
11.65(3.18)***
−4.60(2.31)*
1.92(2.90)
5.78(2.74)*
1.16(2.40)
Model 4
B(s.e)
20.58(3.40)***
**
0.018
0.018
0.018
0.003
0.022
0.026
0.060
0.000
0.031
0.010
0.001
0.011
0.001
η
0.081
0.045
2
***
−6.08(1.56)***
8.12(2.55)**
4.57(2.49)
7.97(2.20)***
9.18(2.39)***
**
3.22(2.46)
9.94(2.59)***
−6.01(1.88)**
−2.90(2.38)
2.37(2.26)
−1.33(1.97)
Model 5
B(s.e)
3.95(2.97)
***
0.034
0.034
0.023
0.008
0.030
0.033
0.051
0.004
0.033
0.023
0.003
0.003
0.001
η2
0.004
0.049
School
School 1
School 2
School 3
School 4
Family structure
Family intact
Step or single
parent family
Sibling number
Parents’ cultural
participation
Pupils’ cultural
participation
Reading
Formal cultural
particpation
Music
TV
Cultural
knowledge
Language
N
η 2 for the model
0.043
0.043
0.015
***
6.58(1.50)***
−1.08(0.43)*
449
0.364
η
0.050
0.008
0.000
0.030
2
Model 1
B(s.e)
***
4.11(2.24)
−0.75(2.29)
−7.93(2.15)***
0.009
0.052
0.041
0.041
η
0.038
0.006
0.002
0.022
2
0.044
1.56(0.35)***
449
0.423
0.011
0.011
0.042
0.042
η
0.034
0.005
0.002
0.021
2
−0.92(0.41)*
0.58(0.26)*
***
6.19(1.43)***
Model 3
B(s.e)
**
3.07(2.14)
−2.24(2.20)
−6.24(2.07)**
0.000
0.013
0.27(1.08)
2.48(1.06)*
434
0.417
0.032
0.001
0.012
0.019
0.037
0.037
η
0.033
0.003
0.004
0.022
2
1.79(0.49)***
−0.46(0.97)
−0.95(0.42)*
0.79(0.28)**
***
5.79(1.46)***
Model 4
B(s.e)
**
2.37(2.21)
−2.95(2.26)
−6.51(2.12)**
Table 4.5: GCSE results (. . . continued from overleaf)
449
0.397
−0.82(0.42)
1.14(0.23)***
***
6.25(1.46)***
Model 2
B(s.e)
**
3.59(2.18)
−2.26(2.25)
−6.63(2.11)**
0.103
0.116
0.66(0.09)***
449
0.604
0.007
0.000
0.005
0.051
0.051
η2
0.054
0.019
0.006
0.018
1.20(0.17)***
0.51(0.30)
0.04(0.35)
0.31(0.22)
***
5.70(1.19)***
Model 5
B(s.e)
***
5.22(1.79)**
−3.04(1.85)
−4.82(1.74)**
98
This follows the pattern that was seen in modelling pupils’ linguistic ability
and cultural knowledge.
Previously, I stated that, if participation in cultural activities is linked to
examination success, this may be due to the development of knowledge or a
set of competencies. Including scores for vocabulary and cultural knowledge
in the model, we can see that the effects of parents’ and pupils’ cultural particiption on GCSE attainment are indeed mediated in this way. In Model 3,
parents’ cultural participation and pupils’ cultural participation both have
highly significant effects. Once pupils’ vocabulary and cultural knowledge
scores are included, in Model 4, both parents’ and pupils’ cultural activities
become insignificant. So, Model 4 shows very strong effects for both vocabulary and cultural knowledge, leaving no significant direct effects for parents’
cultural participation or pupils’ culturalparticipation. That the effects of
these variables are entirely mediated by cultural knowledge and language
ability is striking given that this is not the case for parental social class,
which remains highly significant after the knowledge and language variables
are added to the model. This suggests that the mechanism through which
cultural participation improves educational attainment is in fact the possession of knowledge or a set of competencies, whereas the effect of social class
cannot be explained in this way. The effect of parents’ qualifications in the
intermediate category is rendered insignificant by the inclusion of knowledge
and language scores in the model. However, the negative effect of the missing category becomes highly significant in this model. It is possible that this
reflects a lack of communication about and interest in education in homes
where the children do not know what their parents’ educational level is as
compared to those homes where the children know that their parents have
no qualifications. (Although CSE and no qualifications are grouped together
in this model, the same results emerged using only “no qualifications” as the
comparison category.)
The difference in GCSE scores in favour of girls cannot be explained by gender
differences in cultural capital. The proportion of the variance explained by
gender decreases by only a tiny amount (from η 2 0.020 to 0.018) when the
activities component of pupils’ cultural capital is added to the model, and
the gender effect actually increases once the knowledge and language scores
are included.
99
Standard Occupational Classification class schema
So far, I have used the Goldthorpe class schema in my analyses. One disadvantage of this schema from the point of view of assessing Bourdieu’s theory
of cultural reproduction is that it does not have a separate professional class
category. It may be that professionals, as opposed to owners and managers,
have a distinctive dependence on cultural mechanisms for reproducing privilege. Using the SOC (Standard Occupational Classification) class schema,
which has a separate professional class category, will allow me to assess this
possibility.
I have collapsed the SOC classes into 5 categories as follows.
1. Managers and employers: 1.1, 2.1, 1.2, 2.2, 13
2. Professionals: 3, 4
3. Non-manual and own account: 5, 6, 12, 14
4. Skilled manual: 8, 9
5. Semi- and Unskilled manual: 10, 11, 7, 15, 16
As before, I have determined parents’ social class according to whichever is
“higher” — mother’s or father’s class. The order of dominance I have used
is 1 dominates 2, 2 dominates 3, etc. (I appreciate that I could equally have
categorised professionals as being “higher” than employers and managers.
However, I chose to rank employers and managers above professionals as the
former group are in positions of control over others.) Professional workers
are defined as “persons engaged in work normally requiring qualifications
of university degree standard”. Professional parents had a higher level of
cultural participation (mean score 9.2, standard error 3.9) than parents who
were employers or managers (mean score 5.4, standard error 4.0).
In table 4.6 I model GCSE performance as before, but using SOC classes instead of Goldthorpe classes. Model 1 shows that that pupils’ whose parental
class is defined as professional have a stronger advantage in GCSE performance than pupils’ whose parental class is defined as “manager or employer”.
This is despite the fact that parental qualifications are controlled for in this
model, and the fact that professionals are defined as being in work that generally requires a degree. Of course, many graduates are not in professional
employment, and it may be that the classification of occupations generally
Intercept
Parents’
qualifications
Missing
Degree
Intermediate
O-level
CSE or no
qualifications
Parents’ SOC
class
Missing
Employers and
managers
Professional
Non-manual and
own account
Skilled manual
Semi and
unskilled
manual
Sex
Male
Female
0.006
0.010
0.021
0.003
0.091
0.001
0.008
0.054
0.028
0.000
0.020
0.020
−3.90(2.38)
6.11(2.88)*
8.44(2.80)**
2.94(2.46)
***
−2.29(2.87)
5.58(3.08)
16.94(3.41)***
8.06(2.28)***
0.66(3.00)
**
−5.80(1.95)**
**
−5.36(1.90)**
0.45(2.92)
13.62(3.39)***
6.84(2.23)**
−2.35(2.80)
4.47(3.00)
***
−4.70(2.32)*
1.99(2.93)
6.76(2.74)*
2.00(2.40)
Model 2
B(s.e)
27.12(2.97)***
***
0.018
0.018
0.000
0.036
0.021
0.002
0.005
0.067
0.009
0.001
0.014
0.002
η
0.161
0.052
2
**
−4.82(1.87)**
0.55(2.86)
12.70(3.33)***
7.11(2.19)**
−2.56(2.74)
4.48(2.95)
***
−4.61(2.28)*
1.42(2.88)
6.58(2.69)*
1.77(2.35)
Model 3
B(s.e)
22.38(3.13)***
***
0.015
0.015
0.000
0.033
0.024
0.002
0.005
0.068
0.009
0.001
0.014
0.001
η
0.106
0.051
2
Table 4.6: GCSE results using SOC class schema (continued overleaf. . . )
η
0.190
0.065
2
Model 1
B(s.e)
30.15(2.99)***
***
***
−6.24(1.57)***
2.21(2.39)
8.03(2.80)**
5.72(1.83)**
0.98(2.31)
6.79(2.47)**
**
−6.00(1.90)**
−2.88(2.42)
2.76(2.27)
−0.77(1.97)
Model 4
B(s.e)
6.14(2.86)*
***
0.035
0.035
0.002
0.019
0.022
0.000
0.017
0.036
0.023
0.003
0.003
0.000
η2
0.011
0.051
School
School 1
School 2
School 3
School 4
Family structure
Family intact
Step or single
parent family
Sibling number
Parents’ cultural
participation
Pupils’ cultural
participation
Cultural
knowledge
Language
N
η 2 for the model
0.044
0.044
0.016
***
6.75(1.50)***
−1.12(0.43)**
449
0.396
−0.85(0.42)*
1.15(0.23)***
***
6.41(1.46)***
Model 2
B(s.e)
***
3.78(2.18)
−2.87(2.25)
−6.85(2.11)**
0.009
0.054
0.042
0.042
η
0.042
0.007
0.004
0.024
2
0.039
1.48(0.35)***
449
0.420
0.012
0.014
0.043
0.043
η
0.037
0.005
0.004
0.022
2
−0.93(0.41)*
0.64(0.26)*
***
6.36(1.44)***
Model 3
B(s.e)
**
3.24(2.15)
−2.82(2.21)
−6.48(2.07)**
0.005
0.105
0.112
1.23(0.17)***
0.65(0.09)***
449
0.598
0.000
0.006
0.050
0.050
η2
0.056
0.021
0.006
0.018
0.45(0.30)
0.03(0.35)
0.35(0.22)
***
5.73(1.20)***
Model 4
B(s.e)
***
5.47(1.81)**
−3.12(1.86)
−4.86(1.75)**
Table 4.6: GCSE results using SOC class schema (. . . continued from overleaf)
449
0.362
η
0.055
0.009
0.001
0.033
2
Model 1
B(s.e)
***
4.49(2.24)*
−1.37(2.29)
−8.24(2.14)***
102
demanding a degree used by SOC is out-dated. Today, many positions in
management require a degree. In addition, many graduates must now take
up clerical positions that would previously have been taken up by schoolleavers.
Parental cultural participation is included in model 2. This somewhat reduces the effect of social class on GCSE performance, but social class in
general, and the professional class in particular, remains highly significant.
Model 3 shows the partial mediation of parents’ cultural participation by
pupils’ cultural participation (the inclusion of pupils’ cultural participation
has little impact on the social class effect).
Pupils’ knowledge and language scores are included in model 4. This has the
effect of reducing the advantage associated with the professional classes, but
actually increasing the advantage of the employers and managers, so that, in
this model, the proportion of the variance in GCSE results explained by each
of these class categories is very similar. So, passing on cultural knowledge
and linguistic ability seems to be an important part of the explanation of
the academic performance of the children of professionals, but the advantage
enjoyed by the children of employers and managers appears to be at its
strongest once we control for these factors. This could suggest that the
educational advantage enjoyed by the employer/manager group is largely
due to material factors, whereas cultural factors play an important role for
the professional classes.
Arts and Sciences
Bourdieu never suggests that cultural capital is a less significant advantage
in science subjects than it is in arts subjects. However, it might be suspected
that cultural capital, especially the linguistic element of this, would in fact
be a less valuable resource in science subjects. To examine this possibility, I
will compare pupils’ grades in maths and English. The reason for choosing
these subjects is that, unlike most GCSE subjects, they are not optional in
any of the schools in my survey.
The models for English GCSE (table 4.7) and maths GCSE (table 4.8) follow
the same pattern as the models for total GCSE attainment. As in the case
of pupils’ total GCSE score, the effects of both parents’ cultural activities
and pupils’ cultural activities are mediated by pupils’ cultural knowledge and
linguistic ability.
Intercept
Parents’
qualifications
Missing
Degree
Intermediate
O-level
CSE or no
qualifications
Parents’ class
Missing
Service class lower
Routine non
manual
Petty bourgeois
Skilled manual
Unskilled
manual
Sex
Male
Female
η
0.172
0.029
0.000
0.010
0.015
0.008
0.076
0.002
0.038
0.017
0.009
0.008
0.000
0.043
0.043
−0.02(0.26)
0.65(0.31)*
0.80(0.31)*
0.52(0.27)
***
−0.32(0.34)
1.46(0.35)***
0.92(0.33)**
0.59(0.30)
0.68(0.35)
0.02(0.35)
***
−0.95(0.22)***
2
Model 1
B(s.e)
3.35(0.35)***
*
0.042
0.042
0.006
0.000
0.007
0.013
0.055
0.002
0.025
0.000
0.002
0.011
0.006
η
0.150
0.023
2
***
−0.86(0.21)***
0.62(0.34)
−0.01(0.33)
0.61(0.29)*
0.80(0.32)*
***
−0.35(0.33)
1.14(0.34)**
−0.07(0.25)
0.23(0.32)
0.66(0.30)*
0.42(0.26)
Model 3
B(s.e)
2.51(0.37)***
*
0.038
0.038
0.007
0.000
0.010
0.014
0.060
0.003
0.025
0.000
0.001
0.011
0.006
η
0.097
0.022
2
***
−0.93(0.21)***
0.64(0.35)
−0.03(0.34)
0.51(0.29)
0.66(0.32)*
**
−0.39(0.33)
0.96(0.35)**
−0.12(0.25)
0.26(0.32)
0.64(0.30)*
0.40(0.27)
Model 4
B(s.e)
2.64(0.37)***
*
Table 4.7: English GCSE results (continued overleaf. . . )
***
−0.93(0.21)***
0.57(0.35)
0.02(0.34)
0.54(0.30)
0.78(0.33)*
***
−0.33(0.33)
1.19(0.35)**
−0.09(0.26)
0.32(0.32)
0.68(0.31)*
0.45(0.27)
Model 2
B(s.e)
3.09(0.35)***
*
0.043
0.043
0.008
0.000
0.007
0.010
0.049
0.003
0.018
0.001
0.002
0.011
0.005
η
0.107
0.023
2
***
−1.02(0.18)***
0.47(0.29)
0.09(0.29)
0.49(0.25)
0.64(0.28)*
*
−0.03(0.28)
0.80(0.30)**
−0.24(0.22)
−0.26(0.28)
0.29(0.26)
0.15(0.23)
Model 5
B(s.e)
0.94(0.34)**
0.068
0.068
0.006
0.000
0.009
0.012
0.031
0.000
0.016
0.003
0.002
0.003
0.001
η2
0.017
0.020
School
School 1
School 2
School 3
School 4
Family structure
Family intact
Step or single
parent family
Sibling number
Parents’ cultural
participation
Pupils’ cultural
participation
Reading
Formal cultural
particpation
Music
TV
Cultural
knowledge
Language
N
η 2 for the model
0.008
−0.09(0.05)
453
0.301
−0.06(0.05)
0.10(0.03)***
**
0.44(0.16)**
Model 2
B(s.e)
**
0.90(0.24)***
−0.07(0.25)
−0.09(0.24)
0.004
0.030
0.017
0.017
η
0.036
0.031
0.000
0.000
2
0.047
0.18(0.04)***
453
0.334
0.006
0.003
0.017
0.017
η
0.031
0.028
0.000
0.000
2
−0.08(0.05)
0.03(0.03)
**
0.44(0.16)**
Model 3
B(s.e)
**
0.84(0.24)***
−0.07(0.25)
−0.05(0.23)
0.000
0.011
0.01(0.12)
0.25(0.12)*
438
0.329
0.039
0.000
0.006
0.007
0.013
0.013
η
0.026
0.022
0.000
0.000
2
0.22(0.05)***
−0.03(0.11)
−0.08(0.05)
0.05(0.03)
*
0.38(0.16)*
Model 4
B(s.e)
*
0.75(0.24)**
−0.10(0.25)
−0.06(0.23)
Table 4.7: English GCSE results (. . . continued from overleaf)
0.018
0.018
**
0.47(0.17)**
453
0.280
η
0.040
0.032
0.000
0.002
2
Model 1
B(s.e)
***
0.94(0.25)***
0.05(0.25)
−0.20(0.24)
0.087
0.088
0.07(0.01)***
453
0.508
0.010
0.001
0.000
0.018
0.018
η2
0.062
0.056
0.001
0.001
0.13(0.02)***
0.07(0.04)*
0.03(0.04)
0.00(0.03)
**
0.39(0.14)**
Model 5
B(s.e)
***
1.04(0.21)***
−0.16(0.22)
0.11(0.20)
Intercept
Parents’
qualifications
Missing
Degree
Intermediate
O-level
CSE or no
qualifications
Parents’ class
Missing
Service class lower
Routine non
manual
Petty bourgeois
Skilled manual
Unskilled
manual
Sex
Male
Female
1.31(0.37)***
0.52(0.36)
0.00(0.22)
0.091
0.000
0.063
0.019
0.022
0.032
0.005
0.000
0.000
***
0.12(0.35)
1.97(0.36)***
1.01(0.34)**
0.98(0.32)**
1.40(0.37)***
0.52(0.36)
−0.02(0.22)
0.000
0.000
0.029
0.005
0.020
0.015
0.072
0.000
0.048
0.008
0.001
0.016
0.000
η
0.061
0.050
2
0.05(0.22)
1.34(0.36)***
0.50(0.35)
0.98(0.31)**
0.91(0.34)**
***
0.09(0.35)
1.71(0.37)***
−0.51(0.27)
0.20(0.34)
0.84(0.32)**
−0.02(0.28)
Model 3
B(s.e)
1.56(0.39)***
***
0.000
0.000
0.031
0.005
0.022
0.016
0.075
0.000
0.048
0.008
0.001
0.015
0.000
η
0.035
0.049
2
−0.01(0.23)
1.25(0.38)**
0.52(0.36)
0.97(0.32)**
0.85(0.35)*
***
0.11(0.36)
1.62(0.38)***
−0.51(0.27)
0.22(0.34)
0.85(0.33)**
−0.04(0.29)
Model 4
B(s.e)
1.63(0.40)***
***
Table 4.8: Maths GCSE results (continued overleaf. . . )
0.94(0.31)**
0.89(0.34)**
***
0.11(0.35)
1.74(0.37)***
−0.52(0.27)
0.26(0.34)
0.85(0.32)**
−0.00(0.28)
0.007
0.006
0.019
0.000
−0.47(0.27)
0.54(0.33)
0.95(0.32)**
0.06(0.28)
Model 2
B(s.e)
1.96(0.37)***
***
η
0.075
0.057
2
Model 1
B(s.e)
2.17(0.37)***
***
0.000
0.000
0.025
0.005
0.022
0.014
0.065
0.000
0.043
0.008
0.001
0.016
0.000
η
0.038
0.050
2
−0.10(0.20)
1.20(0.32)***
0.60(0.31)
0.87(0.27)**
0.75(0.30)*
**
0.41(0.31)
1.36(0.32)***
−0.67(0.24)**
−0.27(0.30)
0.46(0.28)
−0.29(0.25)
Model 5
B(s.e)
−0.01(0.37)
***
0.001
0.001
0.032
0.009
0.023
0.014
0.053
0.004
0.039
0.018
0.002
0.006
0.003
η2
0.000
0.050
School
School 1
School 2
School 3
School 4
Family structure
Family intact
Step or single
parent family
Sibling number
Parents’ cultural
participation
Pupils’ cultural
participation
Reading
Formal cultural
particpation
Music
TV
Cultural
knowledge
Language
N
η 2 for the model
453
0.303
−0.13(0.05)*
0.08(0.03)**
**
0.46(0.17)**
0.57(0.25)*
0.05(0.26)
0.31(0.25)
Model 2
B(s.e)
0.015
0.020
0.017
0.017
η
0.013
0.011
0.000
0.004
2
0.020
0.12(0.04)**
453
0.317
0.017
0.004
0.016
0.016
η
0.012
0.010
0.000
0.004
2
−0.13(0.05)**
0.04(0.03)
**
0.46(0.17)**
0.53(0.25)*
0.06(0.26)
0.34(0.25)
Model 3
B(s.e)
Model 4
B(s.e)
0.000
0.007
0.00(0.13)
0.21(0.13)
438
0.310
0.015
0.000
0.017
0.006
0.013
0.013
η
0.010
0.007
0.000
0.004
2
0.14(0.06)*
−0.04(0.12)
−0.13(0.05)**
0.05(0.03)
*
0.41(0.17)*
0.46(0.26)
−0.02(0.27)
0.31(0.25)
Table 4.8: Maths GCSE results (. . . continued from overleaf)
0.019
−0.14(0.05)**
453
0.289
0.018
0.018
η
0.013
0.012
0.001
0.002
2
**
0.48(0.17)**
0.60(0.26)*
0.16(0.26)
0.21(0.25)
Model 1
B(s.e)
0.066
0.086
0.07(0.01)***
453
0.475
0.000
0.001
0.000
0.017
0.017
η2
0.031
0.025
0.000
0.011
0.12(0.02)***
0.01(0.04)
−0.03(0.04)
0.01(0.03)
**
0.41(0.15)**
Model 5
B(s.e)
**
0.75(0.22)**
−0.02(0.23)
0.48(0.22)*
107
One cannot use η 2 to directly compare effects in regressions with different
independent variables, since the size of η 2 is affected by the total variance in
the independent variable. (However, note that the variances for maths and
English GCSE scores are not very different, being 4.18 and 3.72 respectively.)
Therefore, the parameter estimates should be used to compare the role of
cultural resources in the two models. The effects of parents’ cultural capital
in Model 2 and of pupils’ cultural activities in Model 3 are larger in the
case of English GCSE than in the case of maths GCSE. Since reading is
the key component of cultural activities, it is unsurprising that this variable
has a greater effect on performance in English than in maths. However, the
effects of both reading and TV watched are significant in the case of English
GCSE, whereas only reading is significant in the case of maths GCSE. The
vocabulary and cultural knowledge scores have similar effects on performance
in English and maths. So, the effect of cultural capital on examination
attainment does not seem to be limited to arts subjects, although its impact
may be strongest in arts subjects.
4.3
Conclusions
The concept of cultural capital has often been assimilated to the data available to researchers. By using data specifically designed to measure pupils’
and parents’ cultural capital, I have been able to provide a better test of
Bourdieu’s theory.
The first element of Bourdieu’s theory that I set out to test is the claim that
cultural capital is transmitted by higher-class parents to their children. I
broke this down into two questions, firstly, what is the social distribution of
cultural capital, and secondly, to what extent is cultural capital transmitted from parents to their children. I found that parental cultural capital
is strongly associated with parental social class and with parental qualifications. These associations back Bourdieu’s view that cultural capital is
unequally distributed according to social class and education.
The view that cultural capital is transmitted from parents to their children is
strongly supported in the case of pupils’ cultural activities. This component
of pupils’ cultural capital varies by social class, but this variation is entirely
mediated by parental cultural participation. Further evidence to back the
view that cultural capital is transmitted in the home is the lack of a school
effect in determining this component of pupils’ cultural capital. The link
108
between parental cultural capital and pupils’ knowledge and language scores
is weaker, but this is unsurprising given that my measure of parental cultural
capital is a measure of activities. There is no school effect on the test of
linguistic ability, and there is only a small school effect on cultural knowledge.
This contrasts with a strong school effect on GCSE attainment, and suggests
that linguistic ability and cultural knowledge are more strongly transmitted
within the home than in the school. However, it must be borne in mind
that my sample only contains four schools. Ideally one would collect a larger
sample including different types of schools, as it is possible that school type
might affect pupils’ cultural capital. For instance, it is possible that private
schools may instil cultural capital in pupils.
Pupils’ reading and the type of TV watched by pupils account for a significant proportion of the variance in linguistic ability and cultural knowledge,
whereas participation in formal culture does not. This backs the view that
reading develops the intellectual abilities of pupils, whereas participation in
formal culture does not. This could be interpreted as supporting the views
of Crook (Crook, 1997) and de Graaf et. al. (De Graaf et al., 2000) that
public cultural participation serves to communicate status, whereas private
cultural consumption is a means of intellectual self-development. Television
watching is not an indicator of cultural capital that has been used by previous authors, but TV, in common with books, transmits information and may
introduce an individual to new vocabulary and styles of expression. Note,
however, that listening to classical music and playing an instrument are not
associated with linguistic ability or cultural knowledge. Perhaps, then, the
important distinction is not that of “public” or “formal” vs. “private” or
“informal” cultural participation, but rather that of verbal or literary forms
which use words to transmit information or content, vs. visual or musical
forms which are not based on words and are therefore less likely to develop
the skills that are rewarded within the school.
Gender does not account for an important proportion of the variance in any
component of pupils’ cultural capital. Although there are slight variations
in cultural capital according to gender, these differences do not account for
girls’ superior performance at GCSE level.
I went on to examine whether cultural capital affects pupils’ attainment at
GCSE level. The activities component of pupils’ cultural capital is a significant determinant of pupils’ GCSE score, as is parents’ cultural capital.
Again, reading and watching TV are the only significant elements of pupils’
cultural participation. Of these, reading has by far the greater effect. These
109
effects are entirely mediated by pupils’ vocabulary and cultural knowledge.
This firmly backs the view that the reason for the effect of cultural participation on academic attainment is that cultural participation is associated
with intellectual resources which help pupils’ at school. This research gives
no support to the view that teacher’s are prejudiced against working-class
pupils because of their lack of cultural capital. (Note that Hurrel (Hurrel,
1995) has provided strong empirical evidence against the view that teachers are prejudiced against working-class pupils.) Furthermore, in the British
context of an enormous decline in the status of the teaching profession, it
increasingly seems odd to portray teachers as an élite (cultural or otherwise)
who are prejudiced against non-élite pupils.
It may be argued that the association between cultural knowledge and GCSE
attainment must be due to a bias towards high culture in the curriculum.
However, it may be that pupils are rewarded highly in examinations and
assessed coursework for demonstrating precisely that knowledge which they
are unlikely to have gained within the school. This would be consistent with
Bourdieu’s claim that the school fails to give explicitly to everyone that which
it implicitly demands of everyone. In this case, pupils from backgrounds
poor in cultural capital may suffer most from a curriculum that is designed
to avoid content and styles that are associated with the dominant culture.
Furthermore, it seems unlikely that it would be possible, and certain that
it would be undesirable, to introduce a form of assessment that would not
reward linguistic ability or cultural knowledge, broadly defined.
Parents’ social class retains a large and significant direct effect on GCSE
attainment, controlling for the cultural capital variables. Therefore, it seems
that cultural capital is one mechanism through which higher-class families
ensure educational advantage for their children, but it leaves most of the
social class differential in attainment unexplained. Other mechanisms, such
as class differentials in material resources and educational aspirations must
account for the remaining differential in educational attainment. However,
using the SOC (Standard Occupational Classification) class schema, I have
shown that cultural resources do seem to be an important part of the explanation for the educational advantage enjoyed by the children of professional
parents, whereas this is not the case for the children of employers and managers.
In conclusion, I have tried to give a fair test of Bourdieu’s theory of cultural
reproduction, and have found that, although it provides some useful insights,
and helps to explain class differentials in educational attainment, it does not
110
provide a complete account of these differentials. Cultural capital is associated with social class, and is transmitted from parents to children. The
possession of cultural capital does have a significant effect on GCSE attainment. However, this gives us only a partial explanation of class differentials
in GCSE attainment. I stated previously that one cannot say which cultural
activities should be seen as “capital” without an analysis into which cultural
activities are associated with educational success. Reading and TV viewing habits are associated with GCSE attainment and with cultural knowledge and linguistic ability (which in turn are associated with GCSE success).
This is evidence that it is reasonable to see these activities as cultural capital.
There is no evidence here, on the other hand, that musical habits (listening
and playing) or participation in formal culture constitute capital.
In sum, this work vindicates the usefulness of “cultural capital” as an explanatory concept, but does not support the grand theory of “cultural reproduction”.
Chapter 5
Beliefs and Desires
5.1
Introduction
It is well documented that, even controlling for initial educational attainment or ability, working-class pupils have often been less likely to remain
in post-compulsory education and to pursue prestigious educational options
than middle-class pupils. (See for instance, Shavit and Blossfeld (Shavit and
Blossfeld, 1993), and Micklewright (Micklewright, 1989)). Rational choice
theorists have focused on these “secondary effects of stratification” rather
than on initial inequalities in educational attainment.
According to rational choice theory, actors maximise the fulfilment of their
desires on the basis of their beliefs about the situation (or at least, in general
they behave as if this was what they were doing). This means that rational
choice theorists assume that if any agent x wants d and believes that action
a is the best means to attain d, then x does a (all other things being equal).
Or, as Elster puts it:
“The action should be the best way of satisfying the agent’s desires
given his beliefs.” (Elster, 1990, p. 19)
So, actions follow rationally from the beliefs and desires of the agent. (Or, in
the language of economics, the expectations and preferences of the agent.)
But, any action can be interpreted as rational in the light of some set of
beliefs and desires. For example, if I clap my hands together 100 times every
111
CHAPTER 5. BELIEFS AND DESIRES
112
day at sunset, this may be explained by my belief that in doing so I am
summoning evil spirits, and by my desire to destroy the universe. So, the
assumption of rationality alone does not rule out any possible action. It is
clear that we cannot generate any hypotheses about action simply on the
basis that it is rational without making any claims about the beliefs and
desires held by the actor.
One response to this problem on the part of rational choice theorists has
been the attempt to give a more substantive definition of rationality, one
that specifies what counts as a rational belief or a rational desire. This is
inherently problematic, as in general we do not choose our beliefs and desires, or at least we do not choose them in the same sense that we choose our
actions, so the rationality of beliefs and desires cannot be defined instrumentally, as the rationality of actions is. Attempts to define “rational” desires
have been unconvincing, failing to overcome the basic point that one cannot
argue with taste. It may seem more reasonable to attempt to categorise beliefs as rational or irrational. For instance, Elster (Elster, 1987) states that
for a belief to be rational, it must be based on the available evidence, caused
by the evidence, and caused by the evidence in the right way. This condition
is violated if an agent’s beliefs are affected by their interests or desires, or
if cognitive errors (i.e. illogical thinking) play a part in the formation of the
belief. This seems perfectly sensible as a description of how we ought to form
beliefs. Unfortunately, strong evidence has been provided by social psychologists that this is not the way that most of us actually form beliefs, and that
cognitive errors are in fact widespread and systematic (Kahneman et al.,
1982). For Elster, this is perhaps unimportant, as he sees rational choice
primarily as a normative theory. However, if rational choice is to work as an
explanatory theory, then it should not be founded on a false view of belief
formation. Boudon’s (Boudon, 1994) response to evidence of the importance
of cognitive biases in belief formation has been to adapt his concept of rationality to allow for these errors. This leaves us nearly back where we started,
with a version of rational choice theory that rules very little out.
Since what we mean by “rationality” when talking about beliefs, desires, and
actions are three entirely different things, why should a sociologist who makes
rational choice assumptions about the relation between an agents beliefs and
desires and their actions assume that the agents’ beliefs and desires are also
“rational”? The point that an agent’s desires may not be best explained by
any kind of rationality assumption, even if the step from desires to actions
is, is made by Brennan (Brennan, 1990). I see no reason why this principle
cannot also be extended to apply to the beliefs of the agent. So, rather than
113
theorising about what might count as a rational motivation for an action, it
may be more useful to analyse agents’ motivations empirically.
Rational choice theorists analysing class differentials in educational participation have made different assumptions about students’ beliefs and desires.
To date, these claims have not been assessed empirically. Yet it is crucial to
examine the motivations of student decision-makers directly, as their actions
alone cannot tell us what their motivations are.
5.2
Rational Choice Theories
As I stated in chapter 2, several rational choice theorists have addressed the
issue of social class differences in educational participation, and the claims
they make regarding students’ beliefs and desires vary considerably. Murphy
(Murphy, 1990) takes issue with the view that class differentials in educational attainment have anything to do with inequality of opportunity. He
states that the equation of class differentials in educational attainment with
class inequality is due to a failure to take class differences in educational
aspiration seriously. In Murphy’s view, working class youth simply demand
lower levels of education than do middle class youth. This difference in the
levels of demand for education is not due to inequality of opportunity, but
is simply a matter of taste. Murphy treats preferences as given, and not in
need of explanation. His analysis is problematic in that the supposed social class difference in tastes is inferred from the differential in educational
participation. Yet no direct evidence is given to show that the difference
in educational participation is actually due simply to a relative distaste for
education on the part of the working classes, as opposed to, for instance,
class differences in resources or in students’ perceptions of their own ability.
Gambetta (Gambetta, 1987) also explains the class differential in educational
participation in terms of preferences, but introduces an additional mechanism
in the form of social norms.
“. . . it could either be that relatively more subjects in the middle
class feel a greater normative pressure to resist the temptation to
abandon school after a failure or, on the other hand, that relatively
more subjects in the working class do not attach as high a value
to education.” (Gambetta, 1987, p. 173)
114
Surprisingly, there does not seem to be a huge gulf between these rational
choice accounts of class differentials in educational participation, and culturalist accounts such as Willis’ (Willis, 1977) explanation of the behaviour of
his ‘lads’ in terms of a working-class anti-school culture. (Note though, that
only some of the working-class pupils Willis spoke to were actually part of
this ‘lads” culture. Many of the others were keen to do well at school.) The
common feature of these accounts is the view that working class people have
relatively negative attitudes to education or schooling in comparison with
middle class people.
However, other rational choice theorists reject the view that values, norms
and beliefs about education vary according to social class. Boudon (Boudon,
1974) states that class differentials in educational attainment are explained
by the difference in the costs and benefits that are associated with different
educational options for students from different social classes.
For Boudon, the benefits associated with each educational option vary with
social class because ambition is relative to the social starting point of an
individual. So, a working class child who wants to be a lawyer must be more
ambitious than a middle class child who wants to be a lawyer. Therefore,
prestigious educational options may be essential in avoiding social demotion for middle class pupils, whereas working class pupils can avoid social
demotion without pursuing such options. This leads to middle class pupils
being more likely to pursue prestigious educational options than working
class pupils at any given level of ability.
This view is developed by Breen and Goldthorpe (Breen and Goldthorpe,
1997). They make explicit Boudon’s implicit assumptions, 1) that people’s
priority is to avoid social demotion, rather than to pursue social mobility,
and 2) that failure in a high prestige option is believed to be more likely to
lead to social demotion than not attempting to pursue such an option. Breen
and Goldthorpe explicitly state that values, norms and beliefs regarding education do not vary by social class, and that classes differ in terms of two
factors only, average ability and resources.
In sum, rational choice theorists have made various assumptions about the
attitudes and beliefs of individuals regarding education. The question of
whether preferences vary by social class is the subject of much disagreement.
But no attempt has been made to ground these claims about students’ beliefs
and desires in empirical evidence.
5.3
115
Research Questions
I have stated that some rational choice theorists strongly reject the idea that
attitudes to education vary by social class, while others accept it. It certainly
seems possible that attitudes towards education may vary by social class.
Education may be valued as a good in itself and/or as a means to success in
the labour market. There could be a social class difference in the evaluation
of the worth of education on the labour market, for instance if working class
pupils saw educational credentials as less essential to occupational success
than did middle class pupils. Another possibility is that middle class students
may place a higher intrinsic value on education than working class pupils,
perhaps being more likely to see education as enjoyable, or as a tool for
self-development.
A further possibility, perhaps applying especially strongly to the intrinsic
value of education, is that the subjective evaluation of education may be
especially positive among those families with high levels of “cultural capital”.
In households rich in cultural capital, familiarity with the dominant culture
is important for leisure activities, and this may lead to an increase in the
value placed on education as an intrinsic good.
There may be a gender difference in attitudes to education. Girls are often
seen as having a more positive or “mature” attitude towards education than
do boys, and therefore being better behaved and harder working in school.
This could be a possible explanation for girls’ superior performance at GCSE.
If attitudes to education do vary by social class and gender, then this may
help to explain differentials in educational participation. If, on the other
hand, attitudes to education do not vary by social class, then the idea that
the class differential in educational participation simply reveals a difference
in preferences for education between the social classes must be abandoned.
Beliefs about ability may not be entirely determined by actual ability, and
it is possible that an individual’s social background and gender may affect
these beliefs. Breen and Goldthorpe’s formulation of rational choice theory
does not assume that beliefs are necessarily formed entirely rationally, on the
basis of the relevant evidence. What it does assume is that there is no systematic distortion of beliefs by social category. So, if it is the case that people
in general overestimate their abilities, this would not contradict Breen and
Goldthorpe’s view. However, if, for example, men overestimate their abilities in comparison with women, this would violate Breen and Goldthorpe’s
116
assumption.
This assumption is in need of empirical test. It seems quite possible that
pupils in different social categories may form systematically distorted views
of their abilities, due perhaps to different evaluations of their abilities from
parents and teachers. If it is the case that working class pupils underestimate
their abilities in comparison to middle class pupils, this might help to provide
an explanation for the comparatively low rate of participation of working class
pupils in post-compulsory education, and particularly in more prestigious
educational courses.
So, in this chapter, I will assess the posssibility that beliefs and attitudes
regarding education may be socially structured in the ways outlined above.
5.4
5.4.1
Analysis
Attitudes to Education
The pupils in my sample generally responded positively to the questions on
their attitudes towards education. For instance, 81 per cent either agreed
or strongly agreed that ‘The more qualifications you get, the better the job
you are likely to get’, and 69 per cent either agreed or strongly agreed that
‘Studying is worthwhile for its own sake’. See appendix C for the educational
attitudes scale and a reliability analysis of this scale.
Table 5.1 shows pupils’ mean scores in the educational attitude scale, according to gender, social class and parental educational level. The maximum
possible score (for a pupil who responded ‘agree strongly’ to all the positively
phrased questions, and ‘disagree strongly’ to all the negatively phrased ones)
would have been 80. Female pupils, pupils from the service class and pupils
who’s parents are graduates all have relatively high mean scores on the scale
of educational attitudes, but only slightly so. An analysis of variance will
show whether the differences in educational attitudes between these social
categories are significant.
It is often claimed that there is a difference between the social classes in
their attitudes to education. As I stated previously, Murphy puts social class
differentials in educational attainment and participation entirely down to the
relative distaste for education of the working classes, and several theorists
Female
Service class
Non service
class
Degree
No Degree
Mean
Std. Deviation
117
Male
56.7
7.62
58.3
6.64
58.8
6.90
56.9
6.93
58.5
6.43
57.6
7.29
Table 5.1: Pupils’ attitudes to education — mean scores
state that there are social class differences in norms and attitudes regarding
education. Boudon and Breen and Goldthorpe on the other hand, explicitly
exclude the possibility of social class differences in attitudes or tastes for
education. And it is this latter view that seems to be borne out by table
5.2, which shows no significant association between social class and attitudes
to education for the pupils’ in my sample, even before GCSE performance
is controlled for. This is true both of attitudes to the intrinsic value of
education and attitudes to the value of education in the job market. (There
is, however, a significant association between social class and those items
that fell into neither of these categories.) This lack of association between
social class and attitudes to education is all the more striking given that
pupils from working class backgrounds tend to receive markedly worse GCSE
grades than pupils’ from middle class backgrounds, and GCSE performance
is significantly associated with attitudes to education, as table 5.2 shows.
However, it may be argued that the lack of a significant association is due
to the use of a 6-category measure of social class. Perhaps if a bivariate
measure is used, simply comparing the service class to the non service-class,
a significant association may be revealed. Table 5.3 shows the results of this
analysis. It shows a small but significant association between service class
status and educational attitudes. However, breaking down the scale into
its constituent parts shows that there is no significant association between
service class status and attitudes to education either as an intrinsic good
or as an instrumental good, but only with those items that fit into neither
of these categories. This is difficult to interpret, but one possibility is that
the items that fit into neither the ‘instrumental’ nor the ‘intrinsic’ category
reflect behaviour rather than attitudes, (the item ‘I muck about in lessons’
would be an example of this).
Table 5.4 shows the results controlling for GCSE attainment, in order to
Pupils’ Cultural
Capital
GCSE score
Neither
Parents’
Cultural Capital
Intrinsic
Qualifications
Instrumental
Class
Educational
Attitudes
118
Sex
η2
0.012
0.025
0.026
0.048
0.057
0.052
p
η2
p
η2
p
η2
p
0.018
0.000
0.865
0.014
0.010
0.012
0.014
0.068
0.014
0.353
0.020
0.148
0.029
0.038
0.104
0.024
0.139
0.021
0.200
0.021
0.194
0.000
0.015
0.008
0.032
0.000
0.049
0.000
0.000
0.007
0.071
0.048
0.000
0.063
0.000
0.000
0.012
0.018
0.013
0.016
0.132
0.000
Instrumental
Intrinsic
Neither
Qualifications
Educational
Attitudes
Class
Table 5.2: Pupils’ attitudes to education — analysis of variance
η2
0.017
0.012
p
η2
p
η2
p
η2
p
0.018
0.011
0.081
0.013
0.054
0.017
0.018
0.061
0.011
0.077
0.007
0.194
0.013
0.050
Table 5.3: Pupils’ attitudes to education using bivariate parental class and
education
119
Pupils’ Cultural
Capital
Neither
Parents’
Cultural Capital
Intrinsic
Qualifications
Instrumental
Class
Educational
Attitudes
Sex
show the variation in the attitudes of pupils at the same level of educational
attainment. The reason for controlling for GCSE attainmment is that only
those differences in pupils’ attitudes which remain apparent once GCSE performance has been controlled for will be useful in explaining any differences
in the tendency to remain in further education of pupils from different social
groups with similar levels of attainment at GCSE. In order to be charitable
to the social class variable, I have used the bivariate version of this variable.
Controlling for GCSE results, there is a clear lack of association between
social class and attitudes to education. This is important, as, if the lower
staying on rates of working class pupils compared to middle class pupils at
the same level of initial examination performance were to be explained by
social class differences in attitudes to education, then these differences would
have to remain after controlling for GCSE performance.
η2
0.009
0.003
0.001
0.019
0.028
p
η2
p
η2
p
η2
p
0.045
0.000
0.966
0.012
0.018
0.008
0.052
0.533
0.005
0.311
0.007
0.181
0.001
0.836
0.795
0.006
0.264
0.003
0.481
0.001
0.726
0.003
0.008
0.050
0.022
0.001
0.005
0.122
0.000
0.002
0.285
0.038
0.000
0.015
0.008
Table 5.4: Pupils’ attitudes to education controlling for GCSE results
One might guess that, even if there is no relationship between social class and
attitudes to education, there might be a relationship between parents’ qualifications and pupils’ attitudes to education. After all, it is quite plausible that
parents who have benefited from education themselves will pass on a positive
view of education to their children. In fact, table 5.2 shows no significant association between parents’ qualifications and pupils’ attitudes to education.
Table 5.3 shows that this lack of association holds when a bivariate categorisation of parental qualifications (degree or no degree) is used. This lack of
120
association holds both for attitudes to education as an intrinsic good and for
attitudes to education as providing an advantage in the labour market. (The
association with those items that fell into neither category is just significant
at the 0.05 level.) Again, it is striking that, despite the strong association
between parents’ educational level and pupils’ educational performance, the
lack of association between parents’ qualifications and pupils’ attitudes to
education is apparent even before controlling for pupils’ GCSE attainment.
Table 5.4 shows that the insignificance of association becomes even clearer
once GCSE results are controlled for. (Again, a bivariate categorisation of
parental qualifications is used.)
Given the absence of a clear association between attitudes to education and
social class, it is striking that there is a highly significant association between
parents’ cultural participation and pupils’ attitudes to education, which remains after controlling for performance at GCSE. This is most significant
in the case of pupils’ view of the intrinsic worth of education, (η 2 = 0.022,
p ≤ 0.001). The association between parents’ cultural participation and
pupils’ attitudes to education as an instrumental good is just barely significant at the 0.05 level. A slightly stronger association holds between pupils’
own cultural participation and attitudes to education than between parents’
cultural participation and pupils’ attitudes to education. Again, this association is at its strongest in the case of pupils’ view of the intrinsic worth
of education, (η 2 = 0.38, p ≤ 0.000). There is no significant association
between pupils’ cultural participation and the extent to which pupils value
education as a labour market good. The high value placed on the intrinsic
worth of education by families rich in cultural capital may be due to the fact
that the leisure activities valued by these families, such as reading, may demand and/or develop intellectual abilities which also provide an educational
advantage. Since these abilities are important to social life, (where social life
includes cultural participation), as well as to work, education is more likely
to be seen as crucial to self-development in families that have high levels of
participation in the dominant culture.
Girls have significantly more positive attitudes to education than do boys, but
this association is limited to attitudes to the intrinsic value of education, and
is not very large. The association between gender and the evaluation of education as an intrinsic good, controlling for GCSE attainment, is η 2 = 0.012,
p ≤ 0.018. There is no significant difference in boys’ and girls’ evaluations
of the value of education in the job market, suggesting that this aspect of
educational attainment is equally important to both boys and girls.
121
So, attitudes to education vary only very slightly by social class, and this
association is mediated by pupils’ GCSE performance. However, cultural
participation and gender are significantly associated with attitudes to education as an intrinsic good.
5.4.2
Beliefs About Ability
I have stated that it is possible that individuals’ beliefs about their own
abilities may not be entirely determined by their actual abilities, and that
the truth or falsity of this claim may be determined by empirical test. I asked
pupils in their final GCSE year what grades they thought they were likely
to get in their GCSEs. I also asked them to rate their academic abilities in
comparison to other pupils at their school. I then compared these responses
to the actual grades achieved by the pupils.
Firstly, how accurate are the pupils’ beliefs about their own abilities in general? It seems that, overall, pupils’ overestimate themselves. Table 5.5 shows
pupils’ responses to the question ‘In general, how do you rate your academic
abilities as compared to other pupils at your school?’. Strikingly, only 4 per
cent of pupils rated themselves as either below average or poor. Pupils did
not show a strong tendency to rate themselves as ‘excellent’, (only 5.4% did
so). The great majority of pupils, (90.9%) rated themselves as either average
or above average. Table 5.6 shows pupils responses to the question ‘What
grades do you think you are most likely to get in your GCSEs?’. No pupil
placed themselves in the lowest category, ‘mostly Gs and ungraded’, although
in fact 16 pupils in my sample failed to get any GCSEs at all.
Excellent
Above average
Average
Below average
Poor
f
25
164
256
15
2
%
5.4
35.5
55.4
3.2
0.4
Table 5.5: Pupils’ self-assessed ability — frequencies
I gave pupils a score for their actual GCSE results, giving one point for a G
grade, 2 for an F etc., up to 8 points for an A*. I then divided this score
by the number of GCSEs each pupil said they were studying for at the time
Mostly
Mostly
Mostly
Mostly
Mostly
Mostly
Mostly
Mostly
As
As and Bs
Bs and Cs
Cs and Ds
Ds and Es
Es and Fs
Fs and Gs
Gs and ungraded
122
f
13
69
160
162
45
12
1
0
%
2.8
14.9
34.6
35.1
9.7
2.6
0.2
0.0
Table 5.6: Pupils’ self-predicted grades — frequencies
of the survey, to give a mean score. Table 5.7 shows a comparison of pupils’
self-predicted grades to the actual mean scores pupils’ achieved. This table
shows a general tendency for pupils to be overly optimistic in predicting their
results. For instance, I gave 4 points for a D grade and 5 for a C. Therefore,
it seems fair to say that a pupil who expects to get ‘mostly Cs and Ds’ is
expecting to achieve an average score of around 4.5. Table 5.7 shows that,
in fact, pupils’ who expected to achieve mostly Cs and Ds gained on average
a mean score of 3.4. This means that these pupils’ mean grade scores were
on average approximately one grade lower than they predicted.
Mostly
Mostly
Mostly
Mostly
Mostly
Mostly
Mostly
Mostly
As
As and Bs
Bs and Cs
Cs and Ds
Ds and Es
Es and Fs
Fs and Gs
Gs and ungraded
Average
score
should
be
7.0
6.5
5.5
4.5
3.5
2.5
1.5
0.5
Actual
mean
score
6.3
5.9
4.7
3.4
2.8
1.8
2.3
−
Table 5.7: Pupils’ self-predicted grades and actual grades
However, this data only contradicts the strong view that rational actors must
form accurate beliefs. It has no bearing on Breen and Goldthorpe’s view that
123
beliefs are not systematically distorted by social category. If the view that
there is no systematic social belief distortion is to be supported, there should
be no relationship between social class, gender, etc. and pupils’ estimation of
their own abilities once actual academic performance is taken into account.
The claim that there is no systematic distortion of beliefs about ability is
perhaps at its least plausible when applied to gender, as it has often been
claimed that girls underestimate themselves in comparison to boys. And
indeed, this view is borne out by this study.
Table 5.8 shows pupils’ estimation of their own academic abilities broken
down by gender. We can see that girls have a lower estimation of their
abilities than do boys. In particular, boys are considerably more likely to
rate themselves as being above average than girls are (42.1% compared to
28.2%). Girls are much more likely to rate themselves as average than boys
are, (63.2% compared to 48.3%). Table 5.9 shows pupils’ self-predicted grades
broken down by gender. Girls are more likely than boys to have placed themselves in the bottom half of the table (mostly Ds and Es or less). 15.7% of
girls are in the bottom four categories, as compared to 10% of boys. This
is in contrast to the fact that the girls achieved on average superior GCSE
results to the boys, and in particular, to the fact that fewer girls than boys
failed to achieve 5 Cs or more. 53.8% of girls failed to achieve this, as compared to 63.9% of boys. So, the unrealistic optimism of some less able pupils
seems to be more extreme in the case of boys, although the low proportion of
pupils of either sex who place themselves in the bottom four categories may
be seen as more striking than the gender difference.
Excellent
Above average
Average
Below average
Poor
Male
f
%
15
6.2
102
42.1
117
48.3
6
2.5
2
0.8
Female
f
%
10
4.5
62
28.2
139
63.2
9
4.1
0
0
Table 5.8: Self-assessed ability by gender
Table 5.10 shows that boys predicted significantly higher grades for themselves than did girls, controlling for actual GCSE grades (η 2 = 0.022, p ≤
0.001). Boys also rated their own academic abilities significantly more highly
than girls rated their own abilities, controlling for both actual GCSE grades
Mostly
Mostly
Mostly
Mostly
Mostly
Mostly
Mostly
Mostly
124
Male
f
%
6
2.5
35
14.6
88
36.8
87
36.4
18
7.5
5
2.1
0
0.0
0
0.0
As
As and Bs
Bs and Cs
Cs and Ds
Ds and Es
Es and Fs
Fs and Gs
Gs and ungraded
Female
f
%
7
3.1
34
15.2
72
32.3
75
33.6
27
12.1
7
3.1
1
0.4
0
0.0
Table 5.9: Self-predicted grades by gender
Qualifications
Parents’
Cultural Capital
Pupils’ Cultural
Capital
Ability
η2
p
η2
p
Class
Grades
Sex
and school attended (η 2 = 0.019, p ≤ 0.003).
0.022
0.001
0.019
0.003
0.041
0.004
0.036
0.011
0.062
0.000
0.025
0.129
0.037
0.000
0.015
0.010
0.052
0.000
0.025
0.001
Table 5.10: Pupils’ self-predicted grades and estimation of own abilities
If pupils’ perceptions of their own abilities are affected by social class, then
this is a serious problem for Breen and Goldthorpe. On their model, pupils’
subjective probabilities of success affect their decisions regarding post-compulsory education, but these subjective probabilities are not distorted by
social class, and the class difference in educational participation rates is
explained entirely by differences in average ability and resources. Table
5.10 shows that there is a highly significant association between social class
and pupils’ estimation of their own abilities, controlling for actual academic
achievement. (This is using the six-category class categorisation.) The proportion of the variance in pupils’ self-predicted grades that is explained by
social class, controlling for pupils’ actual GCSE grades, is 0.041 (p ≤ 0.004).
125
In addition, pupils from the higher social class categories rated their academic abilities significantly more highly than other pupils, controlling for
actual GCSE performance and school attended (η 2 = 0.036, p ≤ 0.011).
Another factor that may influence pupils’ perceptions of their own abilities
is the educational level of their parents. Table 5.10 shows that parents’ qualifications have a significant effect on pupils’ self-predicted grades, controlling
for actual grades, with the children of more highly qualified parents overestimating their grades in comparison with other pupils (η 2 = 0.062, p ≤ 0.000).
This is using 7 categories for parental qualifications. The effect of parents’
qualifications on pupils’ estimation of their academic abilities compared to
others in their school is not significant, however.
The level of cultural participation of pupils and their parents might also be
expected to have an influence on pupils’ estimations of their own academic
ability. Table 5.10 shows that parents’ cultural participation has a significant positive association with both pupils’ self-predicted grades, controlling
for actual grades, and with pupils’ evaluation of their general academic abilities, controlling for GCSE grades and school attended. In both cases, the
associations with pupils’ own cultural participation are stronger still. Pupils’
cultural capital seems to have a particularly strong association with pupils’
self predicted grades (η 2 = 0.052, p ≤ 0.000). Perhaps cultural participation affects pupils’ self-image, giving them exaggerated confidence in their
own abilities. (Recall that pupils’ cultural participation is strongly associated with parents’ cultural participation — Pearson correlation = 0.617,
p ≤ 0.000.)
5.5
Conclusions
I have stated that no determinate predictions can be derived from the decision rule of rational choice alone, i.e. we can say that people will behave
rationally in pursuit of their desires, on the basis of their beliefs about the
situation, but this tells us nothing if we do not know anything about those
desires and beliefs. I have described some of the assumptions made by rational choice theorists about the beliefs and desires of individuals regarding
education. Rational choice theorists have generally not used subjective data
on individuals, preferring to focus on actual behaviour. Perhaps data on
outcomes is seen as more reliable than data on attitudes and beliefs. However, I would contend that, in this case, actions do not speak louder than
126
words, since we cannot make confident inferences about either the beliefs and
desires that form the basis for a decision or the rationality of the decision
making process from the action alone. Therefore, an understanding of social
class and gender differences in educational participation demands an empirical analysis of the beliefs and desires relevant to the decision of whether to
stay on in education or training.
The view that attitudes to education vary according to social class gains little support from this research. In order to reveal any significant association
between social class and attitudes to education it was necessary to use a bivariate class classification. The association between social class and attitudes
to education is small and is limited to those items within the educational attitudes scale that could not be categorised as applying to either the value
of education as a positional good, of value in the labour market, or to its
worth as an intrinsic good. Furthermore, this small effect disappeared once
GCSE performance was controlled for. This is consistent with the view that
any social class difference in attitudes to education is actually caused by the
comparatively low levels of educational success of working class pupils, rather
than vice versa.
It could be that previous claims that working class youth disliked or were
hostile to education were false. Alternatively, my findings could reflect a
real social change, as working class pupils recognise the need for educational
credentials in the current labour market. Whereas Willis’ lads could be
confident of finding unskilled manual work, young people today have no
certainty of getting a job, and certainly cannot expect job security. One might
also point to the process of “credential inflation” whereby, since absolute
levels of educational attainment are increasing, it is all the more crucial
for each individual to gain at least some qualifications in order to keep up.
Schools may also play a role, as they are under increasing pressure to improve
examination performance, and to reduce the number of pupils who leave
school with no qualifications. This could lead teachers to attempt to instil a
belief in the value of education in all their pupils, rather than just those who
seem likely to do well.
Perhaps more surprisingly, there is no relationship between parents’ qualifications and their children’s attitudes to education. In contrast, both parents’
and pupils’ cultural capital have a significant effect on pupils’ attitudes to
the intrinsic value of education. This ties in with the idea of the cultured
family promoting the value of education in the development of an individual,
rather than just its value in the labour market.
127
Girls’ are significantly more positive in their attitudes to education than are
boys. However, this effect is not large and is limited to the evaluation of
education as an intrinsic good. The greater tendency of girls, as compared
to boys, to see education as having some intrinsic worth, may improve girls’
levels of motivation in school, as it is not always easy to work hard in pursuit
of a distant goal. Boys and girls do not seem to differ in the value they place
on education as a good in the labour market.
The view put forward by Breen and Goldthorpe (1997) that beliefs are not
systematically distorted by social category is not supported by this research.
Boys significantly overestimate themselves compared to girls, both in predicting their GCSE results and in evaluating their general academic abilities
compared to others at their school. Pupils from higher social class categories
significantly overestimate their general academic abilities and their GCSE
grades as compared to pupils from lower social class categories. Pupils whose
parents are relatively well qualified overestimate their GCSE performance significantly compared to pupils whose parents are less well qualified. Pupils
whose parents have relatively high levels of cultural participation significantly
overestimate both their GCSE grades and their general academic abilities.
The association with pupils own cultural participation is even stronger, suggesting that participation in ‘cultured’ activities gives pupils a high estimation of their own abilities.
In sum, the view that the social class differences in educational participation that remain when taking account of initial examination performance
can be explained by social class differences in attitudes towards education
cannot be supported, since there is no significant association between social
class and pupils’ attitudes to education once GCSE performance is controlled
for. However, neither can the view that attitudes to education do not vary
by any social category be supported, since both gender, and parents’ and
pupils’ cultural capital have some impact on pupils’ attitudes to education.
Systematic belief distortion may help to explain differentials in educational
participation, since it seems that gender and social background do systematically distort pupils’ beliefs about their own abilities. A possible explanation
for this is that parents and/or teachers evaluate children differently according
to their gender and social background, and this affects pupils’ perceptions
of their own abilities. Parents and teachers also have a role in encouraging
or discouraging pupils from pursuing further and higher education. Therefore, their beliefs about pupils’ abilities are likely to have a direct impact on
pupils’ educational decisions. Of course, the fact that systematic distortion
by social class, gender, etc. of pupils’ beliefs about their own academic abil-
128
ities occurs does not in itself show that this is even part of the explanation
for class and gender differentials in educational participation rates. However,
it does suggest that analysis of this possibility is worthwhile. This analysis
will be presented in the next chapter.
Chapter 6
Pupils’ Choices
6.1
Introduction
It has been noted that educational participation is associated with social
class, and that this association remains once academic ability is controlled
for (Shavit and Blossfeld, 1993). Micklewright (Micklewright, 1989) provides
evidence that this has also been the case in Britain. Note however that
Micklewright’s analysis is based on the NCDS, and the study children reached
the minimum school-leaving age of 16 in 1974.
This chapter will address the question of social class and gender differences
in rates of educational participation post-16. There are three important
decisions that pupils make at this stage.
1. Whether or not to stay on in further education,
2. What curriculum to follow (e.g. the academic option of A levels, or the
vocational option of a GNVQ),
3. What subjects to take.
Previous chapters have outlined the positions taken by various rational choice
theorists on the question of educational decision-making. Boudon (Boudon,
1974) distinguishes between the ‘primary’ effects of stratification, which are
inequalities in educational attainment occurring before any important educational branching point (such as the decision to stay on in post-compulsory
129
CHAPTER 6. PUPILS’ CHOICES
130
education) has been reached, and the secondary effects of stratification, which
are differences in educational participation which remain once the primary
effects have been controlled for. In the British context, class differences in
GCSE results can be seen as the primary effects of stratification, whereas
class differences in rates of participation in further education that remain
once GCSE performance has been controlled for can be seen as the secondary
effects of stratification.
Working within Boudon’s framework, Breen and Goldthorpe (Breen and
Goldthorpe, 1997) claim that class differentials in educational survival rates
are due to the greater value placed on obtaining service class positions by
pupils from service class origins as compared to pupils from non-service class
origins, due to the priority attached by all pupils to maintaining their current class status, rather than pursuing upward mobility. (This view is termed
relative risk aversion by Breen and Goldthorpe, but, as I have argued, it is
better understood as loss aversion). This view implies that service class
pupils will attach a higher value to the qualifications that are potentially a
gateway to service class positions than do non-service class pupils. Other
possible explanations of social class differences in educational survival rates
have been suggested, such as the fact that working class pupils evaluate their
own abilities less favourably than do service class pupils at the same level
of academic performance (see 5). This could mean that working class pupils
underestimate their chances of success in further education (by which I mean
16-18 education or training) as compared to similarly able service class pupils.
Girls have higher rates of staying on in further education than do boys, although girls and boys opt for the prestigious option of A levels at similar
rates. One possible explanation for girls’ relatively high rate of staying on
would be if there were greater labour market incentives for girls to stay on
than for boys. This would be the case if there were greater job opportunities for male early school leavers than for female early school leavers, for
instance. Another possibility is that girls’ higher estimation of the intrinsic
value of education may play a role in making them want to stay on in further
education, or that the jobs girls aspire to are more likely to require further
educational qualifications than are the jobs boys aspire to.
Subject choices may also vary by social class and gender. Since some subjects
are perceived as more difficult than others (e.g. maths and science are often
perceived as particularly hard) perceptions of ability may play a role here.
Occupational aspirations can be expected to play a role in determining subject choice, as qualifications in different subject areas are useful in entering
131
different occupations. So, social class and gender differences in occupational
aspirations may help to explain social class and gender differences in subject
choice.
Since pupils’ occupational aspirations are likely to affect their educational
decisions, it is important to examine how these aspirations differ by class
and gender, and how these differences may be explained. In chapter 2 I discussed Breen and Goldthorpe’s view that an individual’s current class status
is crucial in determining their aspirations, and that individuals are concerned
primarily to preserve their current class status rather than to achieve upward
mobility. So, how strong is the link between current class status and aspirations? To what extent does the link remain once GCSE performance is
taken into account? And is the link between current class and occupational
aspirations due entirely to what Breen and Goldthorpe term “relative risk
aversion” or could other factors play a role? Also, since in the case of gender differences in aspirations, current class status cannot be the explanation,
what other possible explanations are there? For instance, it could be that
there are social class and gender differences in values regarding work, effectively leading to different subjective hierarchies of occupations. Working
class and female pupils’ lower estimation of their own abilities compared to
middle class and male pupils could be another factor in explaining social class
and gender differences in aspirations. Social networks are also a potential explanatory factor behind social class differences in occupational aspirations.
Having adult “role models” who do a particular type of job may bring that
type of job into a pupil’s field of possibilities, and may increase the pupils’
knowledge about how to get into this type of employment. Since parents are
likely to have friends who are predominantly of their own class, middle class
pupils will know a larger number of adults doing middle class jobs, and will
therefore be likely to have a greater awareness of a larger number of middle
class jobs than will working class pupils.
Cultural participation may play a part in many of the causal mechanisms
suggested above, as it is associated with pupils’ self-perceived ability and
with their view of the intrinsic value of education. It is also possible that
parents’ cultural participation is associated with the family’s social networks.
6.2
Research Questions
1. How are occupational aspirations distributed by social class, gender,
parental education and cultural participation? Do pupils’ values about
132
what is important in a job vary according to social category?
2. What causal mechanisms underlie the differences in occupational aspirations that remain once GCSE performance is controlled for? To what
extent are:
(a) perceived ability,
(b) attitudes to desirable factors in a job, and
(c) social networks
useful in explaining social class and gender differences in occupational
aspirations. Is there an effect of cultural participation, and is it mediated by the above factors?
3. How do choices regarding participation in further education break down
by social class and gender?
4. What causal mechanisms underlie the differences in educational participation rates that remain once GCSE performance is controlled for?
To what extent are:
(a) attitudes to education,
(b) perceived ability, and
(c) occupational aspirations
useful in explaining these differences. Does cultural participation have
an effect, and if so, to what extent is it mediated by the above three
factors?
5. What explains gender differences in subject choice?
6.3
Occupational Aspirations
Pupils were asked what job they would like to be doing in 10 years time and
what job they thought they would actually be doing in 10 years time. Table
6.1 gives their responses to what they would like to be doing in 10 years time
according to gender.
Boys are more likely to aspire to professional positions, and girls to “associate
professional” positions such as teaching, nursing, and social work. No girls
owner or
manager
professional
associate
professional
skilled manual
unskilled
manual
armed forces
routine
non-manual
protective
services
travel and
tourism
hair and beauty
shop work or
waiting
catering
computing
animals
children/carer
arts and media
sport
fantasy
other/don?t
know
total
133
Male
f
%
13
5.5
Female
f
%
11
5.1
Total
f
%
24
5.4
41
5
17
2.2
28
22
13.0
10.2
69
27
15.4
6.0
27
5
11.6
2.2
0
0
0.0
0.0
27
5
6.0
1.1
8
5
3.4
2.2
1
18
0.5
8.4
9
23
2.0
5.1
9
3.9
2
0.9
11
2.5
1
0.4
10
4.7
11
2.5
3
0
1.3
0.0
9
1
4.2
0.5
12
1
2.7
0.2
6
12
0
0
40
25
2
30
2.6
5.2
0.0
0.0
17.2
10.8
0.9
12.9
0
4
10
27
39
3
1
29
0.0
1.9
4.7
12.6
18.1
1.4
0.5
13.5
6
16
10
27
79
28
3
59
1.3
3.6
2.2
6.0
17.7
6.3
0.7
13.2
232
100.0
215
100.0
447
100.0
Table 6.1: Occupational aspirations by gender
134
aspire to manual positions, and girls are more likely than boys to aspire to
routine non-manual positions. A large proportion of girls aspire to work with
children or animals, whereas no boys aspire to this. About 10 per cent of
boys aspire to be involved in sport in some way, whereas this is an unusual
choice for girls.
Table 6.2 shows the same information broken down by social class.
Unsurprisingly, service class pupils are more likely to aspire to professional
occupations and occupations in the arts and the media than are non-service
class pupils. More surprisingly, non-service class pupils are more likely to
aspire to ownership or management positions than are service class pupils.
The data for ‘what job do you think you will actually be doing in 10 years
time’ (not shown in detail here) shows much the same class and gender variations as the data for what pupils would like to be doing, but unsurprisingly,
pupils in general are less likely to give the more ambitious or desirable options
in response to this question.
6.4
Attitudes to Jobs
One factor that could affect pupils’ ambitions is their view of what are the
characteristics of a good job. Pupils were asked “In choosing a job, what
things about it do you think are important? ”. They were given a list of
options (see questionnaire in appendix B), and asked to tick as many of
them as they liked, but then to choose which was most important, second
most important and third most important.
Table 6.3 shows pupils’ choices of the most important characteristic of a job
broken down by sex.
By far the most common choice for boys is “It should be well paid”, whereas
girls are most likely to choose “It should give me the opportunity of helping
others”. Surprisingly, boys are almost twice as likely as girls to say that not
getting in the way of their family responsibilities is the most important thing
in choosing a job. Perhaps boys see their family responsibilities as being to
provide for the family financially, and therefore would see only badly paid or
insecure jobs as incompatible with this. Alternatively, perhaps girls see their
future family responsibilities as more demanding and therefore accept that
a job will necessarily interfere with these, whereas boys, seeing their family
owner or
manager
professional
associate
professional
skilled manual
unskilled
manual
armed forces
routine
non-manual
protective
services
travel and
tourism
hair and beauty
shop work or
waiting
catering
computing
animals
children/carer
arts and media
sport
fantasy
other/don’t
know
total
135
Service class
f
%
6
3.9
Non service class
f
%
16
6.6
37
15
24.2
9.8
29
10
12.0
4.1
4
2
2.6
1.3
18
3
7.4
1.2
2
1
1.3
0.7
7
15
2.9
6.2
1
0.7
9
3.7
2
1.3
8
3.3
2
0
1.3
0.0
9
0
3.7
0.0
0
2
2
5
43
11
1
17
0.0
1.3
1.3
3.3
28.1
7.2
0.7
11.1
6
14
7
13
30
13
2
33
2.5
5.8
2.9
5.4
12.4
5.4
0.8
13.6
153
100.0
242
100.0
Table 6.2: Occupational aspirations by parental class
136
Male
hands
respect
outdoor
thought
excel
family
control
variety
pay
creative
convenient
people
clean
promotion
own boss
secure
not responsible
helping
f
7
9
6
8
7
29
1
7
93
15
1
5
3
6
1
21
1
19
Female
%
2.9
3.8
2.5
3.3
2.9
12.1
0.4
2.9
38.9
6.3
0.4
2.1
1.3
2.5
0.4
8.8
0.4
7.9
f
4
6
1
12
14
15
2
10
32
22
3
18
0
6
2
28
0
45
%
1.8
2.7
0.5
5.5
6.4
6.8
0.9
4.5
14.5
10.0
1.4
8.2
0.0
2.7
0.9
12.7
0.0
20.5
Table 6.3: Occupational values by gender
responsibilities as smaller, may think that a job that interfered with these
would be overly demanding. The above table includes many small cells, and
it is easier to see patterns in the responses if the items are grouped together
into categories. Below is a categorisation, designed to group together items
reflecting similar values.
Control and respect
It should make others respect me
It should give me control over others
The job should let me be my own boss
Cronbach’s α for this subscale = 0.40
137
Intellectual stimulation
It should involve using my head and need thought and
concentration
It should allow me to excel
It should allow me to be creative
It should involve variety
α = 0.45
Material Factors: Money, security, prospects
It should be well paid
It should be a secure job
It should offer chances of promotion
α = 0.50
Manual
The job should involve working with my hands
It should be an outdoor job
α = 0.38
People
It should give me the chance of being with other people
It should give me the opportunity of helping others
α = 0.36
Not too arduous
It should not get in the way of my family
responsibilities
It should have convenient hours and conditions
It should be a clean job
It should not have too much responsibility
α = 0.46
A factor analysis is shown in table 6.4. This shows that component 1 loads
highly on those items intended to reflect material values, but also on some
of those items designed to reflect “control and respect” and a job that is
“not too arduous”. Nonetheless, I think that I am justified in separating
these three categories of values. For example, although a Cronbach’s α of
0.21 between “It should be well paid” and “It should have convenient hours
and conditions” shows a fair likelihood that an individual who selected one of
these items also selected the other, in practice these values may be in conflict
in actually choosing a job. Therefore, in examining which items respondents
Hands
Respect
Outdoor
Thought
Excel
Family
Control
Variety
Pay
Creative
Convenient
Being with
people
Clean
Promotion
Own boss
Secure
Not
responsibility
Helping others
138
1
0.18
0.56
0.06
0.28
0.33
0.39
0.43
0.25
0.46
0.26
0.44
0.28
2
0.50
0.21
0.58
−0.08
0.06
−0.31
0.31
−0.14
−0.06
0.33
−0.21
−0.03
Component
3
4
−0.22
0.33
−0.03
0.04
−0.17
0.18
0.45
0.34
0.64 −0.12
−0.31
0.07
−0.08
0.07
0.50
0.08
−0.07 −0.33
0.36
0.18
−0.34 −0.25
−0.05
0.36
0.48
0.55
0.23
0.46
0.39
−0.07
−0.29
0.49
−0.32
0.25
−0.08
0.18
0.05
−0.15
−0.21
−0.26
−0.04
−0.30
0.15
−0.31
−0.17
0.03
0.19
0.27
0.09
0.26
−0.26
0.17
−0.24
0.14
0.23
−0.16
−0.26
0.69
0.00
0.07
5
0.39
−0.33
0.15
0.06
−0.01
0.39
−0.42
0.31
−0.03
0.13
0.09
−0.55
6
−0.01
−0.12
−0.37
−0.25
−0.14
0.15
−0.02
0.35
−0.42
0.39
0.25
0.27
Table 6.4: Factor analysis of occupational values
selected as most important (rather than those simply selected as important)
it seems fair to put these items into different categories. Component 2 loads
most highly on the two items reflecting manual work. Component 3 scores
most highly on the four items reflecting a desire for ‘intellectual stimulation’.
Component 4 loads most highly on the two items reflecting being with and
helping others. Components 5 and 6 seem to just reflect statistical noise.
Table 6.5 shows that the material aspects of a job (money, security and
prospects) are valued as the most important factor in choosing a job by
nearly 50% of the boys, as compared to nearly 30% of the girls. The three
categories that were most popular with the girls were material factors, “people” (being with others or helping others) and intellectual stimulation. Boys
were considerably less likely than girls to choose people or intellectual stimulation as the most important factor in choosing a job. Boys were more
Control
Intellectual
Material
Manual
People
Not arduous
Total
Male
f
%
11
4.5
37
15.3
120
49.6
13
5.4
24
9.9
34
14.0
242 100.0
139
Female
f
%
10
4.5
58
26.0
66
29.6
5
2.2
63
28.3
18
8.1
223 100.0
Table 6.5: Occupational value categories by gender
likely than girls to choose one of the items in the “not too arduous” category
than were girls, and, unsurprisingly, boys were more likely to say that being outside or working with their hands was the most important factor. The
proportion of pupils valuing control and respect most highly was surprisingly
small, and equally so for boys and girls.
Control
Intellectual
Material
Manual
People
Not arduous
Total
Service class
f
%
5
4.8
40
29.9
105
37.6
2
5.2
28
17.1
16
13.1
157
100.0
Non service class
f
%
12
3.2
47
15.9
59
41.1
13
1.3
43
17.8
33
10.2
251
100.0
Table 6.6: Occupational values by parental class
Table 6.6 shows that both service class and non-service class pupils were
most likely to choose money, security or prospects as the most important
factor in choosing a job. The main difference between the service class and
non-service class is in the likelihood of choosing intellectual stimulation as
the most important factor in choosing a job. Service class pupils were nearly
twice as likely to give an item in this category as their first choice compared to
other pupils. (This social class difference may well be expected to disappear
once educational attainment at GCSE is controlled for). Unsurprisingly,
service class pupils are less likely than others to choose working outdoors or
140
with their hands as the most important factor in choosing a job. Overall
though, pupils from the service class and non-service class seem quite similar
in their choices of the most important factor in choosing a job.
6.5
Occupational Aspirations and Values
Breen and Goldthorpe put forward a model of social class differences in occupational aspirations which they describe as “relative risk aversion”, although,
as I have argued, it is better understood as ‘loss aversion’. On this model,
individuals aim primarily to avoid social demotion. The idea that people are
keener to avoid losses than to make gains has been found useful in explaining
the behaviour of subjects in certain experiments in social psychology (Kahneman et al., 1982). However, it is very difficult to see how the view that “loss
aversion” is the mechanism behind social class differences in occupational
aspirations could be tested directly.
Perhaps the most we can say is that if “loss aversion” is the mechanism
behind the lower aspirations of working class pupils, there should be a direct
association between social class and aspirations. The association should be
unmediated, especially by cultural factors.
One alternative to the “loss aversion” mechanism is the possibility that cultural factors, such as attitudes to what constitutes a desirable attribute in an
occupation, may mediate the effect of social class. Another possibility is that
knowledge of jobs at the higher end of the social spectrum may be associated
with social class, and this may limit the aspirations of lower class pupils.
Both attitudes towards and knowledge of employment may be associated
with cultural participation.
It seems likely that aspirations are adjusted in response to perceived possibilities. One would expect academic performance, and perhaps also subjective
perceptions of ability to affect individuals’ occupational aspirations. If no
association between social class and aspirations remains once these factors
are controlled for, then there will be no need for an additional explanation
in terms of “loss aversion”.
As discussed in chapter 2, the idea of “loss aversion” does not apply readily to
gender differences in educational and occupational aspirations and outcomes.
An analysis in terms of the alternative mechanisms suggested above may shed
more light on gender differences.
6.5.1
141
Occupational values
Gender and social class differences in responses to the question of what is
the most important factor in choosing a job have been described. I will now
model these responses in order to show whether these gender and social class
effects are mediated by other factors.
Table 6.7 shows the model for whether pupils’ chose an item reflecting intellectual stimulation as most important.
Model 1 shows that social class is significant, as pupils from the upper service
class are more likely to have chosen intellectual stimulation as the most important factor in choosing a job. Parents’ education is insignificant, and the
only significant parameter for this variable is the missing category. Therefore, this variable is removed from the analysis (bear in mind though that
parents’ education may have been significant if there had been less missing
data). Boys are significantly less likely than girls to have chosen intellectual
stimulation as the most important factor in choosing a job.
Model 2 shows the effects of parental cultural participation, and parents’
social contacts. Parents’ social contacts are insignificant. Parents’ cultural
participation is significantly positively associated with pupils’ valuing intellectual stimulation most highly. This mediates the social class effect, rendering it insignificant.
Model 3 shows that the effect of parents’ cultural participation is mediated by
pupils’ own cultural participation. Model 4 shows that the effect of pupils’
cultural participation is itself mediated by pupils’ GCSE grades. Model 5
shows that the effect of GCSE grades becomes insignificant once pupils’ cultural knowledge is included in the model.
So, pupils with high levels of cultural knowledge are more likely to say they
rate intellectual stimulation most highly in choosing a job. Boys are less
likely than girls are to value this most highly.
Table 6.8 shows the model for whether pupils selected an item reflecting
material values as the most important in choosing a job.
Model 1 shows that neither social class nor parental education is significant
as a whole. Boys are significantly more likely to value material factors most
highly in choosing a job, and school attended is also significant. Parents’
cultural participation is negatively associated with valuing material factors
0.09(0.60)
0.90(0.54)
0.36(0.54)
0.48(0.52)
0.37(0.58)
0.14(0.63)
−0.93(0.37)*
0.49(0.40)
−0.07(0.35)
−0.31(0.44)
0.04(0.60)
1.13(0.55)*
0.42(0.55)
0.52(0.53)
0.51(0.57)
0.28(0.63)
−0.65(0.43)
0.20(0.45)
0.15(0.45)
−0.35(0.43)
−0.97(0.37)**
0.58(0.40)
0.09(0.35)
−0.42(0.44)
0.44(0.40)
−0.05(0.36)
−0.34(0.44)
−0.87(0.37)*
0.40(0.58)
0.10(0.63)
0.49(0.52)
0.50(0.41)
0.08(0.36)
−0.14(0.45)
−0.81(0.38)*
0.41(0.62)
0.24(0.66)
0.50(0.56)
0.23(0.59)
0.19(0.63)
0.79(0.60)
−0.01(0.60)
0.84(0.55)
0.31(0.54)
Model 4
B(s.e)
−3.30(1.06)**
Model 3
B(s.e)
−2.37(0.53)***
Table 6.7: Models of desire for intellectual stimulation (continued overleaf. . . )
Constant
Parents’ class
Missing
Service class lower
Routine non
manual
Petty bourgeois
Skilled manual
Parents’
qualifications
Missing
Degree
Intermediate
O-level
Male
School
School 1
School 2
School 3
Model 2
B(s.e)
−2.12(0.52)***
Model 1
B(s.e)
−1.38(0.53)**
0.51(0.42)
−0.14(0.37)
−0.14(0.45)
−1.04(0.39)**
0.45(0.62)
0.31(0.66)
0.52(0.57)
0.28(0.59)
0.34(0.64)
0.75(0.61)
Model 5
B(s.e)
−3.37(0.64)***
40.8
465
43.8
465
0.07(0.06)
Model 2
B(s.e)
0.09(0.04)*
47.5
465
0.13(0.06)*
Model 3
B(s.e)
0.05(0.04)
52.1
458
0.01(0.01)
0.02(0.01)*
0.10(0.23)
0.09(0.04)*
0.01(0.02)
61.6
461
0.09(0.06)
Model 5
B(s.e)
0.01(0.05)
0.11(0.06)
Model 4
B(s.e)
0.03(0.04)
Table 6.7: Models of desire for intellectual stimulation (. . . continued from overleaf)
Parents’ cultural
participation
Parents’ social
contacts
Pupils’ cultural
participation
GCSEs
Self-assessed
ability
Cultural knowledge
Language
χ2
N
Model 1
B(s.e)
Class
Missing
Service class lower
Routine non
manual
Petty bougeois
Skilled manual
Parents’
qualifications
Missing
Degree
A-level
Vocational
O-level
CSE
Sex (M)
School
School 1
School 2
School 3
Parents’ cultural
participation
0.81(0.27) ∗ ∗
∗∗
0.61(0.28)∗
0.56(0.32)
0.98(0.30) ∗ ∗∗
−0.05(0.03)
Model 2
B(s.e)
0.78(0.27) ∗ ∗
∗∗
0.64(0.29)∗
0.56(0.33)
1.02(0.29) ∗ ∗∗
−0.02(0.03)
Model 3
B(s.e)
Table 6.8: Material values (continued overleaf. . . )
−0.25(0.36)
0.36(0.48)
−0.38(0.49)
−0.15(0.35)
−2.15(1.08)∗
−0.40(0.42)
0.87(0.28) ∗ ∗
∗∗
0.50(0.30)
0.60(0.34)
1.02(0.30) ∗ ∗∗
−0.08(0.03)∗
0.43(0.39)
0.47(0.47)
0.34(0.44)
0.51(0.43)
0.07(0.43)
0.23(0.46)
Model 1
B(s.e)
0.83(0.28) ∗ ∗
∗∗
0.56(0.29)
0.55(0.33)
1.14(0.30) ∗ ∗∗
−0.02(0.04)
Model 4
B(s.e)
Parents’ social
contacts
Pupils’ cultural
participation
GCSEs
Cultural
Knowledge
Vocabulary
χ2
N
43.31 ∗ ∗∗
465
52.99 ∗ ∗∗
465
47.29 ∗ ∗∗
465
−0.12(0.05)∗
Model 3
B(s.e)
Table 6.8: Material values (. . . continued from overleaf)
Model 2
B(s.e)
−0.07(0.04)
Model 1
B(s.e)
−0.03(0.02)
50.80 ∗ ∗∗
461
0.01(0.01)
0.00(0.03)
−0.12(0.04)∗
Model 4
B(s.e)
146
most highly.
Model 2 shows that parents’ social contacts are not significantly associated
with material values, but the inclusion of this variable renders parental cultural participation insignificant. There seems to be a problem of collinearity
with these two variables, in that each is significant in the absence of the
other, but neither is significant in the presence of the other. It is unsurprising that these two items should be highly correlated, as people are likely to
choose friends who share their interests.
Model 3 shows that pupils’ cultural participation is negatively associated
with material values, and mediates the effect of parents’ cultural participation. Model 4 shows that GCSE performance, cultural knowledge scores, and
vocabulary scores have no significant association with material values.
In sum, boys are more likely than girls to say they value material factors
most highly in choosing a job. It could be that boys are still disposed to
see themselves as potential breadwinners for a family. Cultural participation
is negatively associated with the choice of material factors as being most
important in choosing a job. It could be that this is because those engaged
in cultural participation value intellectual stimulation more highly, or that
the choice of material factors as most important conflicts with a cultivated
self-image.
Table 6.9 shows the model for valuing control and respect most highly in
choosing a job. Model 1 shows that neither social class, parental qualifications, sex, school, parental cultural participation, nor parental social contacts
are significant. Model 2 shows that pupils’ cultural participation is negatively
associated with choosing an item reflecting respect or control as most important in choosing a job. In model 3, this effect is mediated by GCSE grades.
Model 4 shows that neither cultural knowledge nor vocabulary scores are
significant.
The only variable associated with choosing being with or helping others as
the most important factor in choosing a job is gender. In the full model, the
coefficient for males as compared to females is −0.28 (0.36), significance level
0.0003. No variable except school attended is associated significantly with
the choice of items reflecting a non-arduous job as most important, so it is
unnecessary to show a table for this item.
Working outside or with one’s hands was chosen as the most important factor
in choosing a job by only 18 people, so providing a model for this variable
0.19(0.73)
0.55(0.72)
−0.28(0.71)
7.44(15.91)
6.99(15.94)
8.73(15.91)
7.92(15.90)
7.95(15.94)
7.96(15.91)
0.26(0.62)
−0.33(0.70)
−1.53(1.15)
−0.06(0.80)
−0.33(0.95)
−0.48(0.84)
−1.69(1.17)
Model 2
B(s.e)
Model 3
B(s.e)
Table 6.9: Control and respect (continued overleaf. . . )
Class
Missing
Service-class higher
Service-class lower
Routine non
manual
Petty Bourgeois
Skilled manual
Parents’
qualifications
Missing
Degree
A-level
Vocational
O-level
CSE
Sex (M)
School
School 1
School 2
School 3
Model 1
B(s.e)
Model 4
B(s.e)
20.469
465
0.02(0.12)
6.218∗
465
−0.25(0.11)∗
Model 2
B(s.e)
11.578 ∗ ∗
461
−0.03(0.02)
−0.02(0.07)
−0.03(0.01)∗
0.01(0.07)
11.723∗
461
−0.17(0.12)
Model 4
B(s.e)
−0.18(0.11)
Model 3
B(s.e)
Table 6.9: Control and respect (. . . continued from overleaf)
Parents’ cultural
participation
Parents’ social
contacts
Pupils’ cultural
participation
GCSEs
Cultural
Knowledge
Vocabulary
χ2
N
Model 1
B(s.e)
0.03(0.08)
149
would be rather pointless, although there is a significant negative association
between GCSE grades and choosing this item.
So, there are no direct social class effects on pupils’ responses as to what
the most important factor in choosing a job is. Pupils’ from service class
backgrounds are more likely to choose intellectual stimulation as being most
important, but this effect is mediated by cultural participation, which in turn
is mediated by GCSE grades and cultural knowledge. Cultural participation
is negatively associated with valuing material factors and control and respect
most highly. Boys are less likely than are girls to value intellectual stimulation
or being with or helping others most highly, and are more likely to value
material factors as most important in choosing a job.
6.5.2
Occupational Aspirations
In order to model pupils’ occupational aspirations as a bivariate logistic regression, pupils’ aspirations have to be categorised simply as “high” or “low”.
I have categorised positions of ownership and control, professional and associate professional positions, and positions in the arts and media as “high”,
and all others as “low”. I think the positions I have categorised as “high” approximately correspond to service class positions, but I realise the imprecise
and crude nature of this measure. Ultimately, one cannot quite get round
the fact that most people do not respond to questions about their aspirations
with answers such as “manager in a company with 50+ employees”.
Table 6.10 shows the model for pupils’ aspirations. Model 1 shows that pupils
from higher service class backgrounds had significantly increased aspirations.
Boys have significantly lower aspirations than do girls on this model, but bear
in mind that the more detailed breakdown of aspirations given earlier showed
girls to be more likely than boys to aspire to associate professional positions,
but less likely to aspire to professional positions. Model 2 shows that there
is a positive association between parents’ cultural participation and pupils’
aspirations, but parents’ social contacts are not significant. This is perhaps
surprising, as one might have thought that parents’ social contacts would
affect the pupils’ knowledge of different types of occupations, and therefore
affect their aspirations. However, it could be that cultural participation has a
greater effect than social contacts on people’s knowledge of the occupational
structure, and, in particular, of the prestigious positions within it.
Model 3 shows that pupils’ cultural participation is highly significantly as-
Constant
Parents’ class
Missing
Service class lower
Routine non
manual
Petty bourgeois
Skilled manual
Parents’
qualifications
Missing
Degree
Intermediate
O-level
Male
School
School 1
School 2
School 3
Parents’ cultural
participation
0.00(0.39)
0.42(0.44)
−0.65(0.49)
−0.05(0.40)
0.46(0.43)
−0.63(0.49)
−0.76(0.31)*
**
1.03(0.34)**
−0.42(0.36)
−0.46(0.36)
−0.04(0.04)
0.61(0.47)
−0.69(0.51)
−0.02(0.41)
0.54(0.43)
Model 3
B(s.e)
−1.72(0.47)***
**
−0.64(0.49)
1.18(0.47)*
−0.77(0.33)*
**
1.10(0.36)**
−0.37(0.37)
−0.17(0.38)
−0.06(0.04)
0.43(0.50)
−0.69(0.54)
−0.26(0.44)
0.20(0.47)
Model 4
B(s.e)
−2.01(0.97)*
*
−0.55(0.52)
0.80(0.50)
Table 6.10: Job aspirations (continued overleaf. . . )
−0.83(0.29)**
**
1.07(0.32)**
−0.38(0.33)
−0.49(0.35)
0.08(0.03)*
0.57(0.41)
0.49(0.42)
−0.08(0.35)
0.69(0.41)
0.32(0.40)
0.30(0.37)
−0.90(0.29)**
***
1.08(0.32)**
−0.37(0.33)
−0.59(0.34)
Model 2
B(s.e)
−0.64(0.39)
***
−0.56(0.47)
1.28(0.45)**
Model 1
B(s.e)
−0.38(0.41)
**
−0.58(0.46)
1.28(0.46)**
−0.90(0.34)**
**
1.21(0.37)**
−0.50(0.38)
−0.12(0.39)
−0.09(0.04)*
0.50(0.50)
−0.67(0.54)
−0.18(0.45)
0.26(0.47)
Model 5
B(s.e)
−3.17(0.57)***
*
−0.41(0.52)
0.89(0.51)
Parents’ social
contacts
Pupils’ cultural
participation
Occupational
values
Control and
respect
Intellectual
stimulation
Material
Manual
Non arduous
GCSEs
Self-assessed
ability
Cultural knowledge
Language
χ2
N
92.0
447
90.7
447
0.61(0.37)
0.32(0.33)
−2.00(1.18)
0.48(0.44)
0.03(0.01)***
0.19(0.22)
0.78(0.36)*
0.38(0.32)
−1.98(1.13)
0.36(0.42)
154.2
435
0.00(0.65)
−0.24(0.62)
131.9
441
0.25(0.06)***
Model 4
B(s.e)
0.29(0.06)***
Model 3
B(s.e)
Table 6.10: Job aspirations (. . . continued from overleaf)
Model 2
B(s.e)
0.02(0.05)
Model 1
B(s.e)
0.09(0.04)*
0.01(0.02)
163.3
437
0.34(0.34)
−1.90(1.20)
0.51(0.45)
0.03(0.01)**
0.51(0.38)
0.09(0.65)
0.24(0.06)***
Model 5
B(s.e)
152
sociated with pupils’ aspirations, and this mediates the effect of parents’
cultural participation. Model 4 shows that, overall, the effect of occupational values is insignificant. Only the “intellectual stimulation” value is
significantly associated with pupils’ aspirations (the comparison value here
is “non-arduous”).
Model 5 shows that the effect of GCSE performance is significant, but selfassessed ability has no effect controlling for this. The inclusion of GCSE
performance renders the social class effect insignificant. This would seem to
show that the social class effect is due to the revision of aspirations in line
with chances of success. Crucially, this leaves no need for an explanation in
terms of “relative risk aversion”.
Model 6 shows that cultural knowledge has a significant effect on pupils’
aspirations, but vocabulary scores do not. The effect of cultural knowledge
partially mediates the effect of GCSE performance. Cultural knowledge may
be associated with knowledge of prestigious occupational positions — both of
what these positions entail, and how people get into these positions. It may
also be that pupils with high levels of cultural knowledge perceive themselves
(quite possibly correctly so) as being more likely than others to achieve these
positions.
6.6
Educational Choices
This section will describe the breakdown of pupils’ intentions regarding further education, before giving an analysis in the form of logistic regressions.
6.6.1
Choosing to stay on in different courses
Table 6.11 shows that A levels were the most common option, with nearly
half the pupils in my sample intending to pursue this option. Boys and girls
were about equally likely to say they intended to study for A levels. However,
girls seemed to be considerably more likely to intend to take GNVQs than
boys were, and correspondingly less likely to say they intend not to pursue
any further education at all.
Table 6.12 shows pupils’ stated intentions according to their social class status. Service class students were very likely to intend to do A levels, with
A levels
GNVQ
GCSE retakes
Leave
Total
Male
f
%
113
47.7
70
29.5
6
2.5
48
20.3
237 100.0
153
Female
f
%
105
47.5
89
40.3
4
1.8
23
10.4
221 100.0
Total
f
%
218
47.6
159
34.7
10
2.2
71
15.5
458
10.0
Table 6.11: Educational intentions and gender
A levels
GNVQ
GCSE retakes
Leave
Total
Service class
f
%
119
75.8
23
14.6
3
1.9
12
7.6
157 100.0
Non service class
f
%
82
33.6
110
45.1
4
1.6
48
19.7
244 100.0
Missing
f
%
17
29.8
26
45.6
3
5.3
11
19.3
57 100.0
Table 6.12: Educational intentions and social class
around three quarters of service class pupils giving this response, as compared with only about a third of non service class pupils. Non service class
pupils were considerably more likely both to opt for GNVQs and to reject
further education than were service class pupils. Since the amount of missing
data on parents’ jobs is not small, this category has been included in the table. It seems clear that those pupils who did not give an occupation (current
or previous) for either parent approximate the non service class rather than
the mean. This makes sense, as pupils who’s parents have high status occupations may be both more willing to respond to this item on the questionnaire,
and less likely to give answers that are too unclear to be coded.
In sum, pupils from lower social classes were more likely to intend to leave
education altogether after their GCSEs, and more likely to opt for GNVQs
rather than A levels if they intended to stay on. Male and female pupils were
similarly likely to opt for A levels, but female pupils were considerably more
likely to intend to stay on and take a GNVQ, rather than leave.
154
Reliability of Intentions
Unfortunately, two of the schools in my sample were unable to give me adequate information on the actual post-GCSE destinations of their pupils, and
therefore, I am using the pupils stated intentions as the outcome variable in
the analyses to follow. However, the reliability of the pupils’ stated intentions can be tested using data from the two schools who were able to provide
this information.
Stated Intention
Not A
levels
A levels
Total
Outcome
Not A levels
A levels
f
%
f
%
58
71.6
39
30.5
23
81
28.4
100.0
89
128
69.5
100.0
Table 6.13: Reliability of A level choices
Table 6.13 shows that about 72% of those who stated that they would not do
A levels did not, and about 70% of those pupils who stated that they would
do A levels actually embarked on an A level course.
Stated Intention
Not FE
FE
Total
Outcome
Not further education Further education
f
%
f
%
22
40.0
13
8.4
33
60.0
141
91.6
55 100.0
154 100.0
Table 6.14: Reliability of further education choices
Table 6.14 shows that about 92% of those who said that they intended to
pursue further education actually did so. However, only 40% of those who
did not intend to continue in education or training actually left. This low
level of reliability may be accounted for by the very low number (only 55)
who are confirmed as not having pursued further education.
−0.33(0.62)
−0.76(0.52)
−0.07(0.49)
0.00(0.59)
−0.44(0.53)
−0.39(0.63)
−0.73(0.33)*
**
1.13(0.41)**
1.26(0.56)*
0.42(0.36)
−0.23(0.41)
0.51(0.61)
0.20(0.57)
−0.02(0.46)
−0.73(0.34)*
*
1.10(0.43)**
1.38(0.58)*
0.34(0.36)
−0.70(0.33)*
**
1.16(0.41)**
1.32(0.56)*
0.44(0.36)
Model 3
B(s.e)
−1.08(0.49)*
0.04(0.01)***
−0.81(0.35)*
*
1.12(0.42)**
1.10(0.56)
0.38(0.37)
Model 4
B(s.e)
−1.61(1.64)
0.04(0.01)***
Table 6.15: Pupils’ intention to puruse further education (continued overleaf. . . )
Constant
GCSEs
Parents’
qualifications
Missing
Degree
Intermediate
O-level
Male
School
School 1
School 2
School 3
Parents’ class
Missing
Service class lower
Routine non
manual
Petty bourgeois
Skilled manual
Model 2
B(s.e)
−0.68(0.45)
0.04(0.01)***
Model 1
B(s.e)
−0.17(0.62)
0.04(0.01)***
−0.76(0.35)*
*
0.96(0.44)*
1.19(0.56)*
0.53(0.37)
Model 5
B(s.e)
−0.52(0.52)
0.03(0.01)**
91.4
443
87.1
443
−0.01(0.07)
0.08(0.05)
Model 2
B(s.e)
0.86(0.28)**
89.4
443
0.15(0.07)*
Model 3
B(s.e)
0.91(0.29)**
95.0
440
0.04(0.02)
0.41(0.27)
0.12(0.07)
Model 4
B(s.e)
0.82(0.29)**
*
0.07(0.08)
Model 5
B(s.e)
0.70(0.30)*
2.49(1.04)*
98.1
425
−0.69(0.65)
1.84(0.76)*
0.69(0.80)
Table 6.15: Pupils’ intention to pursue further education (. . . continued from overleaf)
Family intact
Sibling number
Parents’ cultural
participation
Parents’ social
contacts
Pupils’ cultural
participation
Self-assessed
ability
Educational
attitudes
Occupational
aspirations
Owner or manager
Professional
Associate
professional
Arts and media
χ2
N
Model 1
B(s.e)
0.92(0.29)**
0.02(0.08)
6.6.2
157
Further Qualifications
I will model pupils’ decision as to whether to stay on in further education
(table 6.15). However, only 81 of the pupils in my sample intended not to
pursue any form of further education, and, perhaps as a result of this, the
overall model fit is not good.
Model 1 shows that, unsurprisingly, GCSE results have a strong positive
association with the decision to stay on rather than leave. Controlling for
this, there is no significant social class effect, and no significant effect of parents’ qualifications. The effect of gender is (just) significant at the 0.05 level,
with boys being less likely to decide to stay on as compared to girls. School
attended was also significant. The number of siblings has no significant effect. Pupils from households with an “intact” family structure (i.e. living
with both mother and father) are significantly more likely than those from
“non-intact” families to stay on in further education. One can break down
this effect to examine the effect of stepfamilies and single parent families.
The effect of being in a step-parent family (B = −1.16(0.36), p > 0.002)
seems to be more significantly negative than the effect of being in a single
parent family (B = 0.84(0.36), p > 0.02). This is surprising in the sense
that stepfamilies are generally not economically disadvantaged to the extent
that single parent families are. Interestingly, Kiernan (Kiernan, 1992) finds,
in her analysis of NCDS data, that young people from stepfamilies formed
after death or divorce were most likely to leave home early, and for reasons
of friction. It could be that stepparents are reluctant to support young people financially once compulsory education has come to an end. (However, in
the case of the NCDS cohort, remarriage did not appear to have a negative
effect on educational outcomes for children (Elliott and Richards, 1991). It
may well be that there is a difference in outcomes between children whose
stepparents are in a stable relationship with their natural parents, and children whose stepparents are not in a stable relationship with their natural
parents. Some of the latter may have to put up with a series of “boyfriends”
or “girlfriends” of their natural parent).
Model 2 shows that neither parents’ cultural participation nor parents’ social
contacts have any significant effect.
Model 3 shows that pupils’ cultural participation is significantly positively
associated with deciding to stay on in further education. Model 4 shows that
pupils’ self-assessed ability and attitudes to education are not significant.
Model 5 shows the effect of pupils’ occupational aspirations. Those who
158
aspire to professional and arts and media positions are significantly more
likely to decide to stay on in further education, controlling for GCSE results,
etc.
So, GCSE performance is an important factor in determining intentions to
pursue further qualifications. The overall model fit is relatively poor, and it
may be that the small number of pupils who did not intend to go into further
education accounts for this, and (possibly) for the lack of either a social class
effect or a parental education effect. Given the lack of these effects, the effect
of family structure is striking.
The gender effect on staying on is not mediated by any of the other variables,
and remains significant in the final model. Since this model does not seem
to explain the gender effect, one must look for alternative explanations. One
possibility would be that 16 year old boys could simply be better able than
girls to find relatively attractive employment, perhaps with some element of
on the job training.
6.6.3
A Levels
Model 1 (table 6.16) shows that, unsurprisingly, the effect of GCSE attainment on the decision to do A levels is highly significant. There is no significant
sex effect. There is a significant school effect, which could be due to a peer
group effect, a school policy effect, or an effect of uncontrolled for parental
inputs. Having a parent who is a graduate is highly positively associated
with the decision to do A levels. Since A levels are the traditional route
to university, it may be that the children of graduates are simply brought
up assuming that they will do A levels. (Note that the “missing” category
seems highly similar to the “no degree” category.) It is probably safe to assume that respondents who were unable or unwilling to state their parents’
educational level were rarely the children of graduates. Social class has no
effect once parents’ qualifications are included in the model. (If we remove
parents’ qualifications from the model, a significant advantage for the upper
service class is revealed.)
The effects of parents’ social contacts and cultural participation are insignificant (model 2). Model 3 shows that pupils’ own cultural participation has
a significant positive association with the decision to do A levels. Model
4 shows that the effect of pupils’ cultural participation is mediated by the
effects of pupils’ self-assessed ability and pupils’ attitudes to education. Cul-
Constant
GCSEs
Parents’ class
Missing
Service class lower
Routine non
manual
Petty bourgeois
Skilled manual
Parents’
qualifications
Missing
Degree
Intermediate
O-level
Family intact
Sibling number
Male
***
−0.15(0.42)
1.82(0.52)**
0.17(0.46)
−0.22(0.42)
−0.18(0.36)
−0.19(0.42)
1.79(0.54)**
0.10(0.46)
−0.23(0.41)
−0.18(0.36)
Model 3
B(s.e.)
−4.37(0.60)***
0.09(0.01)***
**
Model 2
B(s.e.)
−3.97(0.58)***
0.09(0.01)***
−0.18(0.36)
−0.15(0.42)
1.82(0.52)**
0.17(0.46)
−0.22(0.42)
***
Model 4
B(s.e.)
−4.37(0.60)***
0.09(0.01)***
−0.41(0.39)
−0.16(0.44)
1.93(0.53)***
−0.01(0.49)
−0.23(0.44)
***
Model 5
B(s.e.)
−7.01(1.75)***
0.08(0.01)***
Table 6.16: Pupils’ intention to pursue A-levels (continued overleaf. . . )
0.03(0.43)
1.84(0.56)**
0.26(0.48)
−0.17(0.43)
0.04(0.29)
0.05(0.08)
−0.09(0.37)
−0.61(0.58)
−0.78(0.61)
**
0.05(0.50)
0.29(0.56)
−0.48(0.58)
0.43(0.63)
Model 1
B(s.e.)
−4.05(0.69)***
0.09(0.01)***
−0.67(0.43)
−0.12(0.47)
2.01(0.57)***
0.25(0.53)
−0.20(0.48)
***
Model 6
B(s.e.)
−6.81(1.85)***
0.07(0.01)***
School
School 1
School 2
School 3
Parents’ cultural
participation
Parents’ social
contacts
Pupils’ cultural
participation
Self-assessed
ability
Educational
attitudes
Occupational
aspirations
Owner or manager
Professional
Associate
professional
Arts and media
χ2
N
272.8
454
−0.02(0.06)
Model 2
B(s.e)
**
1.27(0.39)**
0.06(0.41)
−0.82(0.45)
0.08(0.05)
275.8
454
0.15(0.06)*
Model 3
B(s.e)
**
1.25(0.40)**
0.06(0.40)
−0.81(0.44)
275.8
454
0.15(0.06)*
Model 4
B(s.e)
**
1.25(0.40)**
0.06(0.40)
−0.81(0.44)
0.09(0.02)***
0.08(0.02)***
297.5
451
0.64(0.28)*
0.58(0.26)*
1.03(0.42)*
310.6
433
−1.57(0.93)
1.83(0.51)***
0.23(0.57)
***
0.07(0.08)
Model 6
B(s.e)
**
1.25(0.46)**
−0.53(0.47)
−1.34(0.53)*
0.11(0.07)
Model 5
B(s.e)
**
1.29(0.42)**
−0.32(0.43)
−1.08(0.48)*
Table 6.16: Pupils’ intention to pursue A-levels (. . . continued from overleaf)
272.3
443
Model 1
B(s.e)
***
1.40(0.41)**
0.34(0.42)
−0.82(0.45)
161
tural participation is associated with confidence in one’s own abilities and a
positive attitude towards education, and this seems to explain the effect of
pupils’ cultural participation.
Model 5 shows the effect of pupils’ occupational aspirations. Aspirations
to work in professional or arts and media positions are strongly positively
associated with the decision to do A levels.
In sum, GCSE performance is extremely important in determining pupils’
decisions to do A levels. This is unsurprising, as schools and colleges generally
require a minimum of 4 or 5 C grades at GCSE before allowing a student to
enrol for an A level course. In addition, students’ perceptions of their chances
of success in an A level course are likely to be largely determined by their
knowledge of their own abilities. (Note that students’ subjective perceptions
of their own abilities have a significant impact, controlling for actual GCSE
performance as measured by GCSE success.)
Parents’ education has a significant impact on A level plans, mediating the
effect of social class. This could suggest that it is important to parents with
degrees, and to their children, that they are not downwardly mobile educationally. This may be more important than the fear of downward occupational mobility, contrary to the model put forward by Breen and Goldthorpe
(1997).
6.6.4
Subject Choices
This section will outline the distribution of subject choices according to social
class and gender.
Arts
Science
Humanities
Languages
Social science
Male
f
%
59
52.2
71
62.8
40
35.4
8
7.1
10
8.8
Female
f
%
61
55.0
45
40.5
51
45.9
17
15.3
37
33.3
Table 6.17: A level subject choice by gender
Table 6.17 shows the proportion of the 113 male and 111 female students
162
intending to do A levels who stated that they intended to take at least one
subject from each of the following subject groups: arts, sciences, languages,
humanities, and social sciences. (The reason for the greater number of pupils
shown as having chosen A level subjects than pupils shown as having chosen
to do A levels at all is that some pupils ticked more than one box in the
question of what they intended to do next, in which case only their first
response is used to calculate the number of pupils taking a particular option,
but all responses on subject choice are used.)
Table 6.17presents an unsurprising pattern. Boys are considerably more
likely than girls to opt for at least one science subject. Girls are somewhat
more likely than boys to opt for at least one humanity, and considerably more
likely to opt for a social science or a language. Male and female students are
about equally likely to choose at least one arts subject.
Arts
Science
Humanities
Languages
Social science
Service
f
70
60
60
17
21
class
%
58.3
50.0
50.0
14.2
25.8
Non Service class
f
%
40
46.5
48
55.8
26
30.2
6
7.0
12
14
Table 6.18: A level subject choice by class
Table 6.18 presents the same information broken down by social class.
Non service class pupils are slightly more likely to take at least one science,
but less likely to take at least one of any of the other subject categories.
In the case of GNVQs, it is more difficult to categorise the subjects according
to type. Table 6.19 shows the five most popular GNVQ choices broken down
by gender.
Since I am dealing with individual subjects rather than groups of subjects
here, the numbers are quite small, and one must be especially cautious in
drawing conclusions from them. The clearest gender difference is girls’ greater
likelihood of choosing health and social care. Boys seem to be more likely
to choose art and design and information technology, whereas for the girls
in this sample, business studies and leisure and tourism are more popular
options than they are for the boys. There is no point in breaking this data
Art and Design
Business
Health Social
IT
Leisure Tourism
Male
f
%
13
22.4
9
15.5
2
3.4
11
19.0
5
8.6
163
Female
f
%
9
12.2
16
21.6
28
37.8
7
9.5
8
10.8
Table 6.19: GNVQ subject choice by gender
down by social class as the numbers of service class pupils choosing each
GNVQ subject are very small indeed.
6.6.5
Gender and Subject Choice
Girls are less likely than boys to choose science subjects at A level and at
degree level. This is the subject of concern because subject choice has been
shown to have consequences in the labour market. For instance, Cheung
(Cheung, 1997) finds that a substantial proportion of the labour market
disadvantage suffered by female graduate respondents to the NCDS survey
was accounted for by the concentration of women graduates in the humanities
and education.
I will model the subject choices of the students who opted to do A levels. I will
examine the decision to take at least one “hard science” (maths, physics or
chemistry), since biology is often seen as the “feminine” science, and indeed,
male and female respondents to my survey had extremely similar likelihoods
of choosing biology. I will then go on to model the number of hard sciences
taken as a proportion of the total number of A levels. The results are shown
in table 6.20.
Model 1 shows that boys are significantly more likely than girls to have opted
for “hard science” A levels, controlling for both overall GCSE results and
maths results. Maths GCSE results are significantly positively associated
with the decision to do hard science A levels. For the pupils in my survey,
there is a slight difference in favour of boys in maths GCSE results (although
this is no longer true nationally). The average score for boys is 3.6, and for
girls 3.4. (Recall that 3 points are allocated for an ‘E’, 4 points represent a
‘D’, so this difference represents only a fraction of a grade). Overall GCSE
Constant
GCSEs
Maths GCSE
result
School
School 1
School 2
School 3
Male
Self-assessed
ability
Occupational
aspirations
Owner or
manager
Professional
Associate
professional
Arts and media
χ2
N
164
Model 1
B(s.e.)
−1.85(0.65)**
−0.01(0.02)
0.34(0.17)*
Model 2
B(s.e.)
0.42(1.03)
−0.01(0.02)
0.30(0.17)
Model 3
B(s.e.)
−0.41(1.17)
0.01(0.02)
0.12(0.18)
−0.96(0.43)*
−0.02(0.44)
−0.69(0.80)
1.03(0.50)*
−1.17(0.44)**
−0.12(0.45)
−0.81(0.79)
0.90(0.51)
0.69(0.26)**
−0.91(0.51)
0.00(0.54)
−0.52(0.90)
0.99(0.62)
0.63(0.29)*
***
0.25(0.98)
1.56(0.41)***
0.70(0.63)
19.6
222
26.4
220
−1.32(0.51)**
66.0
209
Table 6.20: Pupils’ choice of any hard science A levels
grades actually have a slightly negative, though insignificant, association with
the decision to take science A levels, once Maths grades have been controlled
for. This raises the interesting possibility that girls’ overall superiority at
GCSE could actually make them less likely to take science A levels, unless
their grades in maths and science keep up. For example, if a girl gets a B
grade in maths, but the rest of her grades are As, she is probably less likely
to take maths A level than a boy who gets a B in maths, and the rest of his
grades are Cs.
Model 2 shows the effect of self-assessed ability. A high subjective estimation
of a pupil’s own ability is associated with a greater likelihood of choosing
165
maths, physics or chemistry at A level. This mediates the sex effect, making
it insignificant. This suggests that self-confidence is an important factor
in choosing these subjects, since they are perceived as particularly difficult.
(Note that actual maths performance is rendered insignificant in this model,
suggesting that self-evaluated ability is more important than actual ability.)
Model 3 shows the effect of occupational aspirations. Aspiring to professional
positions is significantly positively associated with opting for hard sciences,
whereas aspiring to arts and media positions is negatively associated with
opting for hard sciences. We have seen that boys are more likely to aspire
to professional positions than are girls. In fact, the variable of “occupational
aspirations” is sufficient to mediate the sex effect, removing self-assessed
ability. Once again, self-confidence is the key to this effect. There is a
strongly significant association between self-assessed ability and professional
aspirations, controlling for GCSE results. (B = 0.63, S.E. = 0.24, p > .008.)
Table 6.21 shows models for the proportion of A levels taken that are in hard
science subjects. The statistic given is η 2 . Model 1 shows that gender is
significant, and accounts for a similar proportion of the variation explained
as maths GCSE results do. As in the previous regression, GCSE results have
a negative effect, though not significantly so, once maths GCSE grades have
been controlled for.
Model 2 shows that self-assessed ability has a significant effect. This has a
very slight mediating effect on the gender effect.
Model 3 shows the effect of occupational aspirations. Aspirations for a professional position are significantly positively associated with the proportion
of A level choices that are hard sciences, whereas aspirations towards arts
and media positions are negatively associated with the proportion of hard
sciences chosen. The effects of both maths GCSE grades and self-assessed
ability become insignificant in this model.
In conclusion, the fact that girls are less likely than boys to opt for hard
science A levels seems to have more to do with girls’ self-image than with their
actual abilities. It is perhaps surprising that girls continue to underestimate
themselves in comparison to boys given girls’ superior academic performance
at GCSE. Given the current data, it is not possible to determine why girls
should have lower self-perceptions than boys. One obvious possibility is that
parents and/or teachers may not evaluate girls as positively as boys of similar
ability.
Intercept
Sex
Male
Female
School
School 1
School 2
School 3
School 4
GCSEs
Maths GCSE
result
Self-assessed
ability
Occupational
aspirations
Owner or
manager
Professional
Associate
professional
Arts and media
Other
N
η 2 for the model
219
0.118
Table 6.21: Proportion of hard sciences
221
0.097
−0.00(0.00)
0.05(0.02)*
0.024
0.005
0.029
−0.13(0.05)*
−0.00(0.05)
−0.12(0.10)
0.040
0.034
0.000
0.009
0.07(0.03)*
0.031
0.026
0.000
0.007
η
0.024
0.023
0.023
0.008
0.023
*
−0.15(0.05)**
−0.01(0.06)
−0.13(0.10)
η
0.002
0.028
0.028
2
−0.00(0.00)
0.04(0.02)*
Model 2
B(s.e)
0.29(0.13)*
*
0.14(0.06)*
2
Model 1
B(s.e)
0.05(0.08)
*
0.16(0.06)*
0.187
0.002
0.088
0.006
0.026
***
−0.08(0.12)
0.19(0.04)***
0.08(0.07)
−0.11(0.05)*
208
0.279
0.018
0.001
0.008
0.025
0.020
0.000
0.005
η2
0.013
0.021
0.021
0.06(0.03)
−0.00(0.00)
0.02(0.02)
−0.10(0.05)*
0.00(0.05)
−0.10(0.09)
Model 3
B(s.e)
0.20(0.12)
*
0.12(0.06)*
Constant
GCSEs
Male
School = 1
School = 2
School = 3
School = 4
Self-assessed
ability
Occupational
aspirations
Owner or manager
Professional
Associate
professional
Arts and media
Other
Carer
Model 2
B(s.e)
2.22(1.58)
−0.04(0.02)∗
−3.26(0.85)∗∗∗
−6.64(27.21)
−1.41(0.61)∗
−1.12(0.64)
0
−0.22(0.43)
−9.50(37.52)
0
−1.01(1.17)
−7.05(67.59)
11.45(65.29)
Model 3
B(s.e)
2.59(0.97) ∗ ∗
−0.06(0.02)∗
−3.54(0.98)∗∗∗
−7.57(42.38)
−2.01(0.73) ∗ ∗
−1.83(0.80)∗
0
3.43(0.87) ∗ ∗∗
Model 4
B(s.e)
3.59(1.10) ∗ ∗
−0.02(0.02)
−2.33(0.86) ∗ ∗
−6.62(27.58)
−1.58(0.73)∗
−1.28(0.77)
0
Table 6.22: GNVQ subject choices (continued overleaf. . . )
Model 1
B(s.e)
1.52(0.76)∗
−0.04(0.02)∗
−3.20(0.84)∗∗∗
−6.57(27.27)
−1.35(0.59)∗
−1.09(0.64)
0
3.55(1.34) ∗ ∗∗
Model 5
B(s.e)
3.63(1.28) ∗ ∗
−0.01(0.03)
−2.66(0.94) ∗ ∗
−7.67(45.59)
−1.91(0.81)∗
−1.62(0.79)
0
Occupational
values
Control
Intellectual
Material
Hands
Easy
People
χ2
N
36.474 ∗ ∗∗
130
Model 2
B(s.e)
60.773 ∗ ∗∗
129
Model 3
B(s.e)
61.873 ∗ ∗∗
129
Model 4
B(s.e)
Table 6.22: GNVQ subject choice (. . . continued from overleaf)
36.214 ∗ ∗∗
130
Model 1
B(s.e)
0.50(1.34)
−1.33(1.31)
0.54(0.76)
2.20(1.34)
−8.36(40.89)
0
70.548 ∗ ∗∗
126
Model 5
B(s.e)
6.6.6
169
GNVQs
The most striking gender difference in GNVQ choices is the tendency of girls
to choose “health and social care”. Table 6.22 shows a logistic regression
examining this gender difference. Model 1 shows a highly significant gender
effect. GCSE performance is also significantly negatively associated with
choosing health and social care as a GNVQ subject. However, model 2 shows
that self assessed ability is not significant once GCSE performance has been
controlled for.
Model 3 shows no significant effect of occupational aspirations. However,
in model 4, I coded this variable differently, in order to look specifically at
whether the aspiration to work with children or as a carer affected the choice
of health and social care GNVQ. Unsurprisingly, this is highly significant.
The gender effect is partially mediated in this model, but still remains highly
significant.
Occupational values are included in the final model, as one might suspect that
the fact that girls are less likely to choose material factors as most important
and more likely to choose being with or helping people as most important
might help to explain the gender differential in choosing this GNVQ. However, occupational values in fact have no significant effect on this decision.
So, although A GNVQ in health and social care is perhaps unlikely to lead
to a well paid job, girls do not choose this option because of a relative lack
of concern for material circumstances. Perhaps girls simply see entering the
“caring” labour market as a good way to avoid unemployment. Boys may be
unwilling to enter this labour market due to their own gender stereotypes,
but, alternatively, they may suspect that this labour market is relatively
closed to males. In the case of working with children, they might well be
right in this view, as men who wish to work with young children are often
viewed with suspicion.
6.7
Conclusions
Social class appears to have little effect on what factors are deemed important in choosing a job. There are some gender differences, with girls valuing
intellectual stimulation and being with and helping others more than boys
do. However, these values were not significant in determining the level of
pupils’ occupational aspirations. Girls were more likely to aspire to associate
170
professional positions than were boys, but less likely to aspire to professional
occupations. There was no social class effect on aspirations once GCSE results were controlled for. This suggests that the lower aspirations of working
class pupils simply reflect an adjustment to their perceived chances of success.
The vast majority of the pupils in my sample intended to stay on in school,
and there was no social class effect on this decision. Boys were less likely to
choose to stay on than girls were, and there was a significant effect of family
structure, with pupils from step and single parent families considerably less
likely to stay on.
There was a social class effect on the decision to do A levels, controlling for
GCSE results. However, this effect was mediated by parents’ educational
qualifications. An explanation of the social class effect in terms of “relative
risk aversion” cannot be supported, as there is no direct social class effect.
Rather, it seems that pupils whose parents have degrees wish to pursue A
levels, which represent the traditional route to university. It could be that
these pupils do not wish to be educationally downwardly mobile. Alternatively, they may have a greater knowledge of the value of higher qualifications
on the labour market.
Girls are less likely than boys to choose hard sciences, and this seems to be
partly due to their lower estimation of their own abilities compared to boys,
and partly due to the lower proportion of girls who aspired to professional
positions. Girls choice of health and social care as a GNVQ subject is partly
explained by their greater tendency to aspire to “caring” occupations, such
as working with children. However, this tendency was not explained by the
value girls put on being with and helping others. It could be that boys and
girls are responding to the perceived openness of this labour market to each
sex.
Chapter 7
Conclusions
The aim of this thesis has been to explain social class and gender differentials
in educational outcomes, and, in particular, to evaluate the usefulness of
“cultural reproduction theory” and “rational choice theory” in examining
these questions. Very often in contemporary sociology, theory and empirical
work are separated, to the detriment of both. A basic aim of this thesis
has been to produce a genuinely theory-driven piece of empirical research,
avoiding the dangers of vacuous and irrefutable theory on the one hand and
tedious number crunching on the other
I have criticised Bourdieu for the vagueness of his theory. Such vagueness is a
shield against refutation, and the “heavily articulated sentences” (Bourdieu,
1990) in which the theory is expressed provide further protection. However,
the impossibility of giving an “authentic” operationalisation of Bourdieu’s
theory should not serve as an excuse for failing to provide anything better
than crude proxy measures. This thesis has examined the concept of cultural
capital empirically, and determined which elements of the dominant culture
are actually associated with educational success (and can therefore be termed
capital, in that they provide a return). Thus, I have shown how the originally
rather nebulous concept of cultural capital can be transformed into a concrete
and meaningful concept with genuine applications to empirical research.
I have critically analysed the rational choice approach, and have argued that
the principle of “revealed preference” relied upon (explicitly or implicitly)
by rational choice theorists makes their theories irrefutable. Rejecting the
doctrine of “revealed preference” I have examined the beliefs and attitudes
of students directly. This has allowed me to evaluate competing hypothe-
171
CHAPTER 7. CONCLUSIONS
172
ses as to the reasons for social class and gender differences in educational
participation post-16 empirically.
7.1
Cultural Reproduction
According to Bourdieu’s theory of cultural reproduction, participation in the
dominant culture is transmitted within the middle and upper class home,
and enables the children of these classes to succeed in gaining high levels
of educational credentials. It is unclear from Bourdieu’s formulation of his
theory whether the educational advantage bestowed by participation in the
dominant culture is due to the communication of status or to the acquisition
of useful skills, such as reading ability, or to some combination of the two.
I have used as broad as possible an operationalisation of cultural participation, including reading habits, TV viewing, music listened to, playing an
instrument, and attendance at art galleries and museums. I have also tested
pupils’ vocabularies (ability to use “educated” language being a crucial part
of Bourdieu’s concept of cultural capital) and pupils’ cultural knowledge.
I have found that, in line with Bourdieu’s theory, there is a strong association
between parents’ and pupils’ cultural participation, suggesting that cultural
capital is transmitted within the home.
Cultural participation is associated with GCSE attainment, but only partially explains the social class difference in attainment. Service class pupils
retain a large advantage over their non-service class peers once cultural participation and even vocabulary and cultural knowledge have been controlled
for. Analysis using the SOC class schema shows that vocabulary and cultural knowledge partially explain the advantage of pupils from professional
backgrounds, but not the advantage enjoyed by pupils from employer or
managerial backgrounds.
Not all elements of participation in the “dominant” culture are associated
with GCSE success (controlling for background variables). Participation in
“formal” culture (going to art galleries, museums, concerts) is not associated
with GCSE attainment, and neither is playing a musical instrument or listening to “high-brow” music (classical or jazz). Therefore, there are no grounds
for calling these activities cultural capital. Reading behaviour and watching
relatively “high-brow” programmes on TV, on the other hand, are associated
with GCSE success. This would seem to back the view, put forward by Crook
173
(Crook, 1997) and de Graaf et. al. (De Graaf et al., 2000), that cultural participation is associated with educational performance because certain forms
of cultural participation (such as reading) help to develop cognitive skills
which help at school. This is in contrast to the view that cultural reproduction works through the communication of status (to teachers for instance).
Further evidence to back the former rather than the latter view is given by
the fact that the effect of pupils’ cultural participation on GCSE attainment
is entirely mediated by their cultural knowledge and vocabulary test scores.
Both parents’ and pupils’ cultural participation had a significant effect on
pupils’ attitudes to the intrinsic value of education. This suggests that families rich in “cultural capital” promote the value of education in the development of an individual rather than just its value in the labour market.
Cultural participation is also strongly associated with self-assessed ability
and self-predicted grades, controlling for actual GCSE grades. This suggests
that participation in “cultured” activities gives pupils a high estimation of
their own abilities.
Cultural participation had a small effect on the decision to do A levels (controlling for GCSE results, social class, etc.), and this was mediated by the
effects of pupils’ attitudes to education and beliefs about their own abilities.
So, certain forms of cultural participation (reading and watching relatively
“high-brow” TV) are associated with cultural knowledge and linguistic ability. These abilities in turn are associated with educational attainment at
GCSE. In addition, cultural participation is associated with positive attitudes towards education and with self-confidence (albeit over-inflated selfconfidence) in one’s own abilities, and both of these characteristics are associated with the decision to pursue the ambitious educational option of A
levels.
7.2
Rational Choice
I have discussed various versions of rational choice theory, and the pitfalls
of both “strong” and “weak” assumptions regarding human motivations in
general, and motivations for educational choices in particular. I have argued
that there is no reason for a theory that assumes the link between beliefs and
desires on the one hand and actions on the other to be rational to assume
that people’s beliefs and desires are also “rational”. (The meaning of the
174
term rationality is at any rate quite different when applied to each of these
areas.) An agent’s motivations cannot be inferred from their actions, as any
given action may be consistent with an infinite number of combinations of
desires and beliefs about the situation. Therefore, it is crucial to gather
direct evidence on people’s beliefs and desires, attitudes and aspirations.
It is often assumed that attitudes to education vary according to social class.
However, I have found that the variation in pupils’ stated attitudes towards
education varied only slightly by social class, and this variation was entirely
mediated by success at school, as measured by GCSE results. Neither was
there any association between parents’ qualifications and pupils’ attitudes
to education. These results are surprising given the widespread assumption
that working class people have a more negative attitude to education than
middle class people do. As I stated in chapter 5, it may be that this view was
never justified, or it may be that a change has taken place, as qualifications
have become essential in the job-market, regardless of social class.
Pupils whose parents are in the higher occupational categories and have
higher levels of education overestimate themselves compared to others. Generally, those in dominant positions, whether in terms of gender, class, education, or culture, perhaps justify those positions by developing an exaggerated
belief in their own abilities.
Breen and Goldthorpe (Breen and Goldthorpe, 1997) suggest that working
class pupils’ attach a lower value to middle class occupations than do middle
class pupils due to “relative risk aversion”. On this view, the prime aim of
individuals from all classes is to avoid downward social mobility rather than
to achieve upward social mobility. As I pointed out in chapter 2, the term
“relative risk aversion” is misleading. The term “loss aversion” would more
accurately describe this process which is driven by fear of losing one’s current
class position.
I assessed the possibility that pupils’ occupations are driven by loss aversion by surveying pupils on their occupational expectations and aspirations,
and also on the factors they considered important in choosing a job, since
any difference in aspirations could be value driven rather than loss aversion
driven. I found that occupational values did not vary by social class, and, in
addition, no significant class difference in aspirations remained once GCSE
results had been controlled for, suggesting that pupils’ simply adjust their
expectations in line with their likely opportunities.
The great majority of pupils in my sample intended to say on in education
175
post-16, and there was no social class effect on this decision once GCSE performance had been controlled for. Pupils from single or step-parent families
were less likely to choose to stay on than pupils in “intact” families. This
effect is striking given the absence of a social class effect on staying on in
further education.
The social class effect on the decision to do A levels was mediated by the effect
of parents’ qualifications. This finding is crucial, as the lack of a direct social
class effect on this decision is inconsistent with Breen and Goldthorpe’s explanation in terms of “relative risk aversion”. Perhaps students’ decisions are
driven more by a fear of educational downward mobility than of occupational
downward mobility. Students’ self assessed abilities, attitudes to education,
and occupational aspirations have some effect on the intention to do A levels,
but these variables do not mediate the effect of parental education.
Overall, it seems that perhaps the stress that rational choice theorists put on
the “secondary effects” of stratification is misplaced, at least in the British
context. The social class effect on attainment at the compulsory level (GCSE)
is strong, and is only partially explained by other variables. It is striking that
this is not the case for educational participation beyond the compulsory level.
The crucial determinant of participation in further education is performance
at GCSE.
The reason for Breen and Goldthorpe’s assumption that there are important
social class effects on choices at 16+, controlling for ability, may be that this
was the case for the NCDS children (Micklewright, 1989). (Note that Micklewright controls for tested ability in maths and comprehension, but not for
examination performance. For evidence that social class affected the likelihood of getting a degree or higher vocational qualification, controlling for
both tested ability and examination performance at 16, see Sullivan and Cheung (Sullivan and Cheung, 2000)). However, the British educational system
has altered dramatically since these children, born in 1958, were at school.
The NCDS cohort grew up in an era of major educational reforms. They
entered secondary school aged 11 in 1969. The majority of the NCDS cohort
members went to comprehensive schools which were reorganised from the
three different types of schools under the so-called “tripartite system”. During the years of the comprehensive reorganisation, some NCDS cohort members still went to these grammar, technical and secondary modern schools
before particular schools were re-organised. Entry to these schools was primarily based on the results of the “11 plus” examination, which pupils sat at
age 11. Pupils who passed the 11 plus examination generally went to gram-
176
mar schools, whereas pupils who failed the examination went to either the
more vocationally oriented technical schools or to secondary modern schools.
(Private schools were largely untouched by government reforms). During this
period, there was a two-tier system of examinations. At age 16, most grammar school pupils took the General Certificate of Education at Ordinary level
(GCE O level), whereas the examination aimed at secondary modern pupils
was the Certificate of Secondary Education (CSE) which was set at a lower
level. Comprehensive reorganisation and the raising of the school leaving age
to 16 in 1974 gave a boost to the numbers taking the CSE exams, and the
majority of pupils subsequently left school with some kind of certificate.
Subsequently, comprehensivisation has continued, and only a small number
of grammar schools remain. The introduction of the GCSE (in 1988) means
that all pupils now follow a similar curriculum until age 16. It seems likely
that, by pushing the first key educational transition point back to age 16, this
reform has reduced the importance of the secondary effects of stratification.
Despite this, Breen and Goldthorpe have simply assumed that the secondary
effects of stratification not only continue to exist, but are crucial in explaining
social class differences in outcomes. In fact, I would suggest that, in the
British context at least, it is the primary effects of stratification, i.e. social
class effects on attainment at GCSE that are crucial in explaining social
class differences in educational outcomes. The reason for the neglect of these
primary effects by rational choice theorists is that they do not fit readily into
a rational choice framework.
7.3
Gender
Whereas social class differentials in educational attainment and participation
have remained fairly stable, gender differences have diminished rapidly in
most advanced industrialised societies, and in some instances, such as that
of attainment at GCSE, these differences have been reversed. Despite this,
strong differences in subject choice remain.
The girls in my sample were less likely to choose hard sciences at A level
than were boys. This effect seemed to be partly explained by girls’ lower
estimation of their own abilities, and partly by their lower tendency to aspire to professional occupations as compared to boys. Girls’ tendency to
reject science subjects is important in the light of evidence that women in
Britain still earn considerably less than similarly qualified men, even when
177
they are childless, and evidence that subject choice was an important factor
in explaining sex differences in occupational outcomes for the NCDS cohort
(Cheung, 1997).
Boys were less likely to choose to stay on in further education than girls, but
there was no significant gender difference in the tendency to choose A levels.
Despite girls’ relatively high levels of attainment at GCSE, girls still seem
to lack confidence in their abilities as compared to boys. Boys significantly
overestimated their academic abilities as compared to girls. It could be that
boys are still given more positive messages about their abilities (perhaps by
parents or teachers) than girls are given, even though the boys’ actual levels
of academic performance are lower.
Girls were slightly (but significantly) more positive than boys in their evaluation of education as an intrinsic good. There was no significant difference
between boys and girls in their evaluation of education as a labour market
good.
There were significant gender differences in occupational values, with girls being more likely to choose intellectual stimulation and being with and helping
people, as opposed to material factors, as being most important in choosing
a job. It could be that boys still see themselves as future breadwinners for
a family, and therefore feel the need to prioritise income more than girls do.
Girls were more likely to aspire to associate professional jobs and less likely
to aspire to professional jobs than were boys. This is perhaps unsurprising,
as associate professional jobs, such as teaching, nursing, and social work, are
female dominated. It could be that girls still feel the need to choose jobs
that are easily compatible with being the primary carer for children.
It is often claimed that girls’ cultural participation (notably reading) is higher
than that of boys. Therefore, it seemed plausible that the “cultural reproduction” approach might shed light on girls’ superiority at GCSE. In fact, for
this sample, gender differences in cultural participation proved to be slight,
and certainly could not provide an explanation for girls’ superior GCSE performance as compared to boys.
The gender difference in GCSE results is quite substantial for the pupils in
my sample (as it is nationally). The average GCSE points score (recall that
1 point is awarded for a G, 2 for and F, etc.) for a boy is 35 (equivalent to 5
Ds and 3 Cs) as compared to 39 for a girl (equivalent to 7 Cs and a D). This
is, of course, not as dramatic as the social class difference in GCSE results
178
- the average service class pupil in my sample gained 51 points (equivalent
to 5 Bs and 3 As) whereas the average non service class pupil gained 34
points (6 Ds and 2 Cs). (The mean GCSE score for those whose parental
occupation was recorded as “missing” was even lower at 28). However, the
gender gap is substantial, and it is unsurprising that it should be the focus of much interest. On the other hand, one should not be tempted into
drawing a parallel between the gender differential in GCSE performance and
the social class differential. When we concern ourselves with differentials in
educational achievement, rather than absolute levels of achievement, this is
because education is a positional good - i.e. a good whose value depends
on how much we have compared to others. Since men are still strongly advantaged in the labour market, their relatively poor performance at GCSE
should not be an object of great concern. However, that does not mean that
gender differences at GCSE should not be an object of great interest. Girls’
superior performance is interesting precisely because it breaks the rule that
socially disadvantaged groups tend to under-perform educationally too.
7.4
Family Size and Structure
Family size (i.e. the number of siblings a pupil has) and family structure
(intact two parent family, step-parent or single parent family) were primarily included in the analyses simply as “control variables”. However, both of
these variables had significant associations with GCSE attainment. The negative effect of having a higher number of siblings was mediated by the pupils’
cultural knowledge and language scores. This is consistent with previous research suggesting that relatively thinly-spread parental attention in larger
families affects children’s intellectual development (Nisbet, 1953). However,
the negative effect on GCSE attainment of being in a step- or single-parent
family is not mediated in this way. Furthermore, controlling for GCSE performance, pupils’ from step- and single parent families were less likely to plan
to stay on in post-compulsory education or training after their GCSEs. The
effect of being in a step-family actually seemed to be more negative than
the effect of being in a single-parent family, which is surprising in that one
would expect step-parent families to be considerably better off financially
than single-parent families. It seems likely, however, that the level of economic resources that are available to a child are affected by factors other
than total household income. For instance, a step-parent may be less willing
to support a child beyond compulsory education than a “real” parent would
be.
7.5
179
Suggestions for future research
I have argued for the superiority of the approach of collecting data which is
specifically designed to address the research questions at hand over the approach of using whatever data is to hand to address one’s research questions.
However, I cannot deny the enormous disadvantage of collecting one’s own
data with the limited resources available to a student. My data set is modest
in size, and unrepresentative.
The fact that my data-set does not contain the private sector is a particular
problem for the assessment of cultural reproduction theory. The mechanisms
of social reproduction may vary between the state and private sectors. Private schools may be more likely to encourage cultural activities and to promote knowledge of the dominant culture. Private schools may function as an
alternative route to educational credentials for families who have economic
capital but lack cultural capital.
A comparison between the state and private sectors would also be relevant
to rational choice theory. Staying on rates are thought to be higher in the
private sector than in the state sector once attainment at the compulsory
level has been controlled for. If this is true once family characteristics such
as social class have been controlled for, is this due to the efforts of such
schools to promote staying-on as worthwhile, or to a peer-group effect, or
perhaps to private school pupils having relatively high estimations of their
own abilities?
Claims about educational attainment are often made on the basis of outdated evidence, such as NCDS. The claim that social class affects rates of
educational participation in Britain, controlling for educational attainment at
the compulsory level, is often made in the present tense, although it relates
to information on children born in the 1950’s. There is clearly a need for
an up to date representative data-set dealing with educational attainment
and participation. Enormous changes in the educational system, and in the
wider society, have taken place since the NCDS children left school in the
1970’s, and we cannot simply assume that these have made no difference
to pupils’ experiences. Claims regarding the effect of family structure on
children’s educational experiences are also made on the basis of NCDS data,
despite the huge changes that have taken place in family life. My data
suggests that family structure is still very important, but much more detailed
information on family structure would be needed in order to analyse the
reasons (emotional, social, economic) behind this effect. Another limitation
180
of my data set is the absence of information on ethnicity. My data set is
too small too have provided reasonable sized sub-samples of minority ethnic
groups. A larger survey would be able to examine whether there are true
ethnic differences in educational attainment (controlling for social class) and
what the reasons are for the differing educational performance of different
minority ethnic groups.
A regrettable omission from my research in terms of theory is that I have not
drawn on Coleman’s concept of “social capital”. For Coleman, social capital
can exist within the home, in the form of the level of support given to pupils
by parents. Social capital also exists in the form of social norms and networks,
in the community surrounding a school where parents know each other (and
possibly also the teachers). “A child’s friends and associates in school are sons
and daughters of friends and associates of the child’s parents.” (Coleman,
1990, p. 318). This affects the relationship of parents to the school. Parents
are able to get a lot more information about what is going on in the school,
and about the behaviour of their own child. Parents who talk to each other
can establish strong norms of behaviour for their children. Coleman argues
that the community around Catholic schools in the US, which is created by
the church, creates and enforces strong norms against dropping out of high
school, and that this explains the lower drop out rate in Catholic schools
than in other private schools.
Coleman also uses the concept of social capital within the school. Coleman,
in “The Adolescent Society” (Coleman, 1961) discusses the effects of peer
groups on pupils. He asks why academic prowess does not lead to popularity in school, whereas sporting prowess does. For Coleman, the answer
is that sporting competitions take place between schools, whereas academic
competition takes place between individuals. Therefore, individuals who do
well academically are chided for “rate-busting”, whereas pupils who excel in
sports gain popularity and status.
Whatever the explanation for the social norms prevailing in schools, it seems
likely that these norms may have an important effect on educational attainment, and, more broadly, on students’ well-being while in school. Smith and
Tomlinson (Smith and Tomlinson, 1989) find that having a high proportion
of high achievers in a school improves the achievement of pupils in the school
as a whole. (This could of course work through teachers’ expectations as well
as peer group norms). Power et. al. (Power et al., 1998) present an interesting finding that, whereas academically able pupils at a grammar school were
likely to worry about not being able to keep up with the work, academically
181
able pupils at a comprehensive were much more likely to worry about other
pupils thinking they were too clever. The sample size of Whitty et al’s study
are far too small to generalise from. However, I think that the norms imposed
by peers may well have as much impact on attainment as the rules imposed
by the school, yet studies focusing on attainment have not generally looked
at these norms. In addition, the desire for a peer group that will have a
positive effect on the child seems to be an important factor in school choice,
(Gewirtz et al., 1995; Walford, 1994) and may well be an explanatory factor
in any “private school effect”.
For all these reasons, I think that if we are to gain an understanding of the
processes underlying inequalities in educational attainment and participation within the contemporary British education system, we need a large and
representative data-set, containing both state and private schools.
7.6
General Theoretical Conclusions
Throughout this thesis, I have stressed the need to integrate theory with
empirical research. The need to operationalise a theory in order to test
it against empirical evidence almost inevitably leads to the abandonment
of “grand theory” (in C. Wright Mills’ phrase (Mills, 1959)) in favour of
something more precise. The empirical work will also be unlikely to sustain
the theory in whole. Some concepts will prove fruitful, others unhelpful or
unfounded.
The grand theory of cultural reproduction is unsupported by this thesis, as by
other work that has sought to unpack the theory in order to give an empirical
analysis (DiMaggio, 1982; DiMaggio and Mohr, 1985; DiMaggio and Mohr,
1995; Crook, 1997; De Graaf et al., 2000). Instead, we return to the much
earlier idea that middle class families have cultural resources, as well as
economic and social resources, which put them at an educational advantage.
Furthermore, cultural resources seem to account for only a modest proportion
of the educational advantage enjoyed by pupils from the higher social class
categories. In addition, participation in the “formal culture” of art galleries,
museums, etc. has no association with educational attainment at GCSE,
controlling for social class and other background variables. So, rather than
the grand theory of cultural reproduction we are left with a much more
modest story — reading and TV viewing habits are associated with abilities
(linguistic and cultural knowledge) which aid academic performance.
182
Rational choice theorists have been torn between strong (and obviously false)
and weak (and vacuous) versions of rational choice theory. I have pointed out
the need for data on agents’ motivations — something that rational choice
theorists have generally shied away from. Rational choice theorists seem to
hold a rather macho conception of the nature of science, according to which
it is characterised by the rejection of “soft” data, rather than by the testing
of hypotheses.
In the sociology of education, rational choice theorists have focused on the
“secondary effects” of stratification since these effects can readily be seen as
due to choices. In my view, this has led to an exaggeration of the importance
of the secondary effects of stratification, and a neglect of the primary effects.
Only Murphy (Murphy, 1990) attempts to characterise social class differences in educational performance prior to any key decision point in terms of
choices. As I have argued, this attempt is highly problematic. Not all of social life can be readily or usefully characterised in terms of decision-making.
This is a complex area. It is clear, for instance, that desires, attitudes and
beliefs are not straightforwardly chosen, and yet it would be wrong to suggest
that human will and agency can be can be entirely absent from shaping an
individuals’ desires, attitudes and beliefs. However, I would suggest that, to
the extent that a sociological question cannot usefully be seen in terms of
“decision-making”, rational choice theory will not be a useful tool for tackling
the question.
Overall, I would argue that sociologists, regardless of theoretical perspective,
should strive firstly for clarity. Vague and obscure theories have no use
except to inflate the author’s reputation among the gullible. (It is a sad and
perplexing fact that the humanities and social sciences are full of people who
are willing to be awed by anything incomprehensible). This criticism must
be most strongly directed against French and French-inspired sociology (for
illustration of this point with authors other than Bourdieu, see Sokal and
Bricmont (Sokal and Bricmont, 1998)). However, it should be remembered
that formality is not the same thing as clarity, and one can hide behind
formulae as well as words.
Secondly, it is crucial to abandon the distinction between theoretical and
empirical work. Speculation about the social world without any real engagement with it ranks alongside speculation as to the existence of an external
world for sheer futility. The existence of people styling themselves sociological theorists, or “social theorists” without ever engaging in empirical work is
absurd. Such people should be banished to departments of philosophy, or if
183
they won’t have them (as seems probable) to departments of cultural studies.
Appendix A
Correlations
184
Occupational
aspirations
Language
Graduate
parent
GCSEs
FE
Cultural
knowledge
Educational
attitudes
A levels
σ
p
σ
p
σ
p
σ
p
σ
p
σ
p
σ
p
σ
p
Alevels
0.48
0.00
0.28
0.00
0.44
0.00
0.60
0.00
0.45
0.00
0.39
0.00
0.46
0.00
Cultural knowledge
0.16
0.00
0.21
0.00
0.63
0.00
0.43
0.00
0.57
0.00
0.39
0.00
0.48
0.00
Educational attitudes
0.23
0.00
0.23
0.00
0.06
0.24
0.10
0.03
0.19
0.00
0.28
0.00
0.16
0.00
FE
0.33
0.00
0.14
0.01
0.16
0.00
0.28
0.00
0.44
0.00
0.21
0.00
0.23
0.00
GCSEs
0.30
0.00
0.58
0.00
0.44
0.00
0.60
0.00
0.63
0.00
0.23
0.00
0.33
0.00
Graduate parent
0.25
0.00
0.29
0.00
0.45
0.00
0.43
0.00
0.06
0.24
0.14
0.01
0.30
0.00
Language
0.31
0.00
0.39
0.00
0.57
0.00
0.10
0.03
0.16
0.00
0.58
0.00
0.25
0.00
Occupational
aspirations
0.46
0.00
0.39
0.00
0.19
0.00
0.28
0.00
0.44
0.00
0.29
0.00
0.31
0.00
Parents’ cultural
participation
0.43
0.00
0.49
0.00
0.22
0.00
0.22
0.00
0.45
0.00
0.54
0.00
0.33
0.00
0.30
0.00
Pupils’ cultural
participation
0.41
0.00
0.44
0.00
0.24
0.00
0.22
0.00
0.41
0.00
0.41
0.00
0.36
0.00
0.37
0.00
Service class parent
0.41
0.00
0.34
0.00
0.13
0.01
0.15
0.00
0.37
0.00
0.51
0.00
0.29
0.00
0.31
0.00
Sex
0.00
0.95
−0.02
0.71
0.11
0.02
0.14
0.00
0.09
0.05
−0.03
0.56
0.04
0.36
0.05
0.32
Single or step family
−0.11
0.02
−0.07
0.12
−0.14
0.00
−0.19
0.00
−0.24
0.00
0.02
0.75
−0.07
0.13
−0.02
0.67
Sibling number
−0.14
0.00
−0.29
0.00
0.01
0.87
−0.06
0.17
−0.24
0.00
−0.19
0.00
−0.23
0.00
−0.16
0.00
0.27
0.00
0.24
0.00
0.12
0.01
0.13
0.00
0.29
0.00
0.27
0.00
0.14
0.00
0.20
0.00
Social capital
Social capital
Sibling
number
Single or step
family
Sex
Parents’ cult.
participation
Pupils’ cult.
participation
Service class
parent
σ
p
σ
p
σ
p
σ
p
σ
p
σ
p
σ
p
Alevels
0.43
0.00
0.41
0.00
0.41
0.00
0.00
0.95
−0.14
0.00
−0.11
0.02
0.27
0.00
Cultural knowledge
0.49
0.00
0.44
0.00
0.34
0.00
−0.02
0.71
−0.29
0.00
−0.07
0.12
0.24
0.00
Educational attitudes
0.22
0.00
0.24
0.00
0.13
0.01
0.11
0.02
0.01
0.87
−0.14
0.00
0.12
0.01
FE
0.22
0.00
0.22
0.00
0.15
0.00
0.14
0.00
−0.06
0.17
−0.19
0.00
0.13
0.00
GCSEs
0.45
0.00
0.41
0.00
0.37
0.00
0.09
0.05
−0.24
0.00
−0.24
0.00
0.29
0.00
Graduate parent
0.54
0.00
0.41
0.00
0.51
0.00
−0.03
0.56
−0.19
0.00
0.02
0.75
0.27
0.00
Language
0.33
0.00
0.36
0.00
0.29
0.00
0.04
0.36
−0.23
0.00
−0.07
0.13
0.14
0.00
Occupational
aspirations
0.30
0.00
0.37
0.00
0.31
0.00
0.05
0.32
−0.16
0.00
−0.02
0.67
0.20
0.00
Parents’ cultural
participation
0.62
0.00
0.44
0.00
0.09
0.06
−0.20
0.00
−0.07
0.16
0.45
0.00
Pupils’ cultural
participation
0.34
0.00
0.09
0.07
−0.08
0.09
−0.05
0.34
0.32
0.00
0.62
0.00
Service class parent
0.03
0.53
−0.18
0.00
0.00
0.97
0.20
0.00
0.44
0.00
0.34
0.00
Sex
0.04
0.42
−0.07
0.13
0.03
0.57
0.09
0.06
0.09
0.07
0.03
0.53
Sibling number
0.06
0.18
−0.11
0.02
−0.20
0.00
−0.08
0.09
−0.18
0.00
0.04
0.42
Single or step family
−0.03
0.53
−0.07
0.16
−0.05
0.34
0.00
0.97
−0.07
0.13
0.06
0.18
0.45
0.00
0.32
0.00
0.20
0.00
0.03
0.57
−0.11
0.02
−0.03
0.53
Social capital
Appendix B
Questionnaire
187
Appendix C
Educational Attitudes Scale
Table C.1 shows the questions on attitudes to education to which pupils
responded either agree strongly, agree, neither agree nor disagree, disagree or
disagree strongly. Those items that are marked with an asterisk are phrased
negatively, and the scales for these items have been reversed.
C.1
Reliability
The reliability for the whole 16-item scale is standardised item α = 0.80. The
reliability for the 7-item sub-scale for attitudes to education as an intrinsic
good is standardised α = 0.73. The reliability for the 5-item sub-scale for
attitudes to education as an instrumental good is less high, at standardised
item α = 0.43. Nevertheless, I think that use of this sub-scale is justifies
in that it meets my theoretical purpose. Also, one must bear in mind that
reliability analysis is based on an assumption that the responses for each
item are normally distributed. This assumption is strongly violated in the
case of this educational attitude scale.
Factor analysis shows (see table C.2) one factor accounting for 27% of the
variance, with further factors being much less important. (The second factor
accounts for 9% of the variance, the third for 8%). I have limited the number
of factors extracted to three. The first component can be seen as reflecting
students’ general attitude towards education. The highest loading items are
‘I don’t like studying’ and ‘I enjoy studying’. The highest loading items in
the second component are ‘The more qualifications you get, the better the job
204
APPENDIX C. EDUCATIONAL ATTITUDES SCALE
STUDY3
STUDY4
STUDY8
STUDY9
STUDY14
STUDY15
STUDY16
STUDY2
STUDY5
STUDY6
STUDY7
STUDY1
STUDY1
STUDY10
STUDY11
STUDY12
205
Intrinsic
Studying is worthwhile for its own sake
I enjoy studying
Studying increases your confidence
Studying increases your ability to think clearly
*I don’t like studying
Studying improves your ability to be creative
*Studying is only ever worthwhile if it leads to a
job
Instrumental
The more qualifications you get, the better the
job you are likely to get.
*Qualifications are useless for getting jobs
Studying can help you to gain skills which will be
useful at work
*These days, it doesn’t matter how many
qualifications you get, you still won’t get a job
These days, if you are unqualified, you can’t get a
job
Neither
*School is a waste of time
*I muck about in lessons
*Studying is irrelevant to real life
I take school seriously
Table C.1: Attitudes to education scale
you are likely to get’, ‘Qualifications are useless for getting jobs’, and ‘These
days, it doesn’t matter how many qualifications you get, you still won’t get a
job’. This may suggest that component 2 reflects the ‘instrumental’ attitudes
to some degree. However, the overall picture for components 2 and 3 is rather
messy, and may simply reflect differences in the distributions of responses for
each item.
APPENDIX C. EDUCATIONAL ATTITUDES SCALE
STUDY1
STUDY2
STUDY3
STUDY4
STUDY5
STUDY6
STUDY7
STUDY8
STUDY9
STUDY10
STUDY11
STUDY12
STUDY13
STUDY14
STUDY15
STUDY16
1
0.592
0.420
0.403
0.711
0.300
0.499
0.305
0.631
0.637
0.557
0.377
0.698
0.175
0.725
0.515
0.336
Component
2
−0.009
0.474
0.285
−0.428
0.413
0.037
0.394
0.153
0.171
−0.456
0.258
−0.236
0.169
−0.373
0.271
−0.145
206
3
0.305
−0.252
0.103
−0.166
0.120
0.043
0.478
−0.251
−0.153
0.024
0.220
0.122
−0.652
−0.134
−0.151
0.479
Table C.2: Factor analysis of attitudes to education
Appendix D
Coding of occupational
expectations and aspirations
1. Business Owner
15. Architect
2. Manager
16. Associate Professional
3. Professional Other
4. Engineer
17. Teacher (Secondary or
freelance)
5. Doctor
18. Teacher (Primary)
6. Dentist
19. Nurse
7. Vet
20. Technician
8. Barrister
21. Skilled manual
9. Solicitor
22. Mechanic
10. Lawyer
23. Electrician
11. Computer
Programmer/Analyst
24. Engineer (non-professional)
12. Academic/Researcher
25. Other
13. Scientist
26. Unskilled manual
14. Accountant
27. Factory Work
207
APPENDIX D. CODING
208
28. Cleaner
53. Something in computing
29. Other
54. Computer operator
30. Armed Forces
55. Other
31. Army
56. Working with animals
32. RAF pilot
57. Veterinary Nurse
33. Other
58. Working with children
34. Routine non-manual other
35. Clerical/Office work
36. Secretary
37. Protective Services
38. Police officer
59. Nanny
60. Nursery nurse
61. Childminder
62. Other
63. Carer
39. Fire fighter
64. Old people’s home
40. Other
65. Something in law
41. Travel and Tourism
42. Holiday Rep
43. Air Hostess
44. Other
45. Hair and Beauty
46. Own salon
66. Something in the arts
67. Designer
68. Artist
69. Writer
70. Journalist
47. Waiter
71. Actor
48. Shop work
72. Musician
49. Stacking shelves
73. Photographer
50. Chef/Catering
74. Other arts or media job
51. Own restaurant
75. Something in sport
52. Profession/Head/Top Chef
76. Sports person
APPENDIX D. CODING
209
77. coach/trainer/instructor
84. Model
78. Work in leisure centre or gym
85. Student
79. PE teacher
86. Unemployed
80. Other
87. Stay at home
81. Fantasy other
88. Too vague to categorise
82. Film star
89. Other
83. Pop star
90. Don’t know
Bibliography
Arnot, M. (1994). In Stone, L., editor, The Education Feminism Reader.
Routledge and Kegan Paul, New York, London.
Becker, G. S. (1964). Human Capital. University of Chicago Press, Chicago.
Bernstein, B. (1973). Class, Codes and Control, volume 1. Paladin, London.
Boudon, R. (1974). Education Opportunity and Social Inequality. Wiley, New
York.
Boudon, R. (1994). The Art of Self Persuasion. Polity Press, Cambridge.
Bourdieu, P. (1967). Systems of education and systems of thought. International Social Science Journal, 19(3).
Bourdieu, P. (1974). The school as a conservative force: Scholastic and
cultural inequalities. In Eggleston, J., editor, Contemporary Research in
the Sociology of Education. Methuen, London.
Bourdieu, P. (1977a). Cultural reproduction and social reproduction. In
Karabel, J. and Halsey, A. H., editors, Power and Ideology in Education.
OUP, Oxford.
Bourdieu, P. (1977b). Outline of a Theory of Practice. CUP, Cambridge.
Bourdieu, P. (1989). Language and Symbolic Power. Polity Press, Cambridge.
Bourdieu, P. (1990). In Other Words. Polity Press, Cambridge.
Bourdieu, P. and Boltanski, L. (1981). The education system and the economy: Titles and jobs. In Lemert, C., editor, Rupture and Renewal Since
1968. CUP, New York.
Bourdieu, P. and Passeron, J.-C. (1990). Reproduction in Education, Society
and Culture. Sage, London, second edition.
210
BIBLIOGRAPHY
211
Bourdieu, P. and Saint-Martin, M. (1974). Scholastic excellence and the
values of the educational system. In Eggleston, J., editor, Contemporary
Research in the Sociology of Education. Methuen, London.
Bowles, S. and Gintis, H. (1976). Schooling in Capitalist America. Basic
Books, New York.
Breen, R. and Goldthorpe, J. (1997). Explaining educational differentials: Towards a formal rational action theory. Rationality and Society,
9(3):275–305.
Brennan, G. (1990). What might rationality fail to do? In Cook, K. S.
and Levi, M., editors, The Limits of Rationality. University of Chicago
Press, Chicago.
Carmines, E. and Zeller, R. (1979). Reliability and Validity Assessment.
Sage, Beverley Hills.
Cheung, S.-Y. (1997). Meritocracy Revisited: A Disaggregated Approach to
the Study of Educational and Occupational Attainment in Britain. PhD
thesis, Oxford University.
Coleman, J. (1961). The Adolescent Society. The Free Press, Glencoe, IL.
Coleman, J. (1990). Equality and achievement in education. Westview, Boulder, Colorado.
Collins, R. (1971). Functional and conflict theories of educational stratification. American Sociological Review, 36:1002 – 1019.
Cooper, B. (1976). Bernstein’s codes: a classroom study. volume 6 of Occasional Paper. University of Sussex Education Area.
Crook, C. J. (1997). Cultural Practices and Socioeconomic Attainment: The
Australian Experience. Greenwood Press, Westport, Connecticut.
De Graaf, N. D., De Graaf, P., and Kraaykamp, G. (2000). Parental cultural
capital and educational attainment in the Netherlands: A refinement of
the cultural capital perspective. Sociology of Education, 73:92–111.
De Graaf, P. (1986). The impact of financial and cultural resources on educational attainment in the Netherlands. Sociology of Education, 5(9).
De Graaf, P. (1988). Parents’ financial and cultural resources, grades and
transition to secondary school in the federal republic of germany. European Sociological Review, 4(3):209–231.
BIBLIOGRAPHY
212
DiMaggio, P. (1982). Cultural capital and school success. American Sociological Review, 47(2):189–201.
DiMaggio, P. and Mohr, J. (1985). Cultural capital, education attainment
and martial selection. American Journal of Sociology, 90(6):1231–1261.
DiMaggio, P. and Mohr, J. (1995). The intergenerational transmission of
cultural capital. Research in Social Stratification and Mobility, 14.
Egerton, M. (1997). Occupational inheritance: The role of cultural capital
and gender. Work, Employment and Society, 11(2):263–282.
Elliott, J. and Richards, M. (1991). Children and divorce: Educational performance and behaviour before and after parental separation. International Journal of Law and the Family, 5:258–76.
Elster, J. (1987). Ulysses and the Sirens. Cambridge University Press, Cambridge.
Elster, J. (1990). When rationality fails. In Cook, K. S. and Levi, M., editors,
The Limits of Rationality. University of Chicago Press, Chicago.
Erikson, R. (1984). Social class of men, women and families. Sociology,
18(4):500–514.
Erikson, R. and Jonsson, J. (1996). Can Education be Equalized? Westview
Press, Boulder.
Farkas, G., Grobe, R. P., Sheehan, D., and Shuan, Y. (1990). Cultural resources and school success: Gender, ethnicity and poverty groups within
an urban school district. American Sociological Review, 55:127–42.
Floud, J., Halsey, A., and Martin, F. (1956). Social class and educational
opportunity. Heinemann, London.
Gambetta, D. (1987). Were they Pushed or did they Jump?
University Press, Cambridge.
Cambridge
Gewirtz, S., Ball, S., and Bowe, R. (1995). Markets, Choice and Equity in
Education. Open University Press, Buckingham.
Gipps, C. and Murphy, P. (1994). A fair test? : assessment, achievement
and equity. Assessing assessment. Open University Press, Buckingham.
BIBLIOGRAPHY
213
Giroux, H. (1982). Power and resistance in the new sociology of education:
Beyond theories of social and cultural reproduction. Curriculum Perspectives, 2(3).
Graetz, B. (1988). The reproduction of privilege in australian education.
British Journal of Sociology, 39(3):358–76.
Halsey, A. H., Heath, A., and Ridge, J. (1980). Origins and Destinations.
OUP, Oxford.
Heath, A. and Britten, H. (1984). Women’s jobs do make a difference. Sociology, 18(4):475–490.
Hellevik, O. (1997). Class inequality and egalitarian reform. Acta Sociologica,
40(4):377–98.
Hurrel, P. (1995). Do teachers discriminate? Sociology, 29(1):59–72.
Jencks, R. (1972). Inequality: A reassessment of the effect of family and
schooling in America. Basic Books, New York.
Jenkins, R. (1992). Pierre Bourdieu. Routledge, London.
Jonsson, J. and Mills, C. (1993a). Social class and educational attainment
in historical perspective: A swedish english comparison. part 1. British
Journal of Sociology, 44(2):213–47.
Jonsson, J. and Mills, C. (1993b). Social class and educational attainment
in historical perspective: A swedish english comparison. part 2. British
Journal of Sociology, 44(3):403–28.
Jonsson, J. O. (1987). Class origin, cultural origin and educational attainment. European Sociological Review, 3(3):229–242.
Kahneman, D., Slovic, P., and Tversky, A. (1982). Judgement Under Uncertainty. Cambridge University Press, Cambridge.
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.
Katsillis, J. and Rubinson, R. (1990). Cultural capital, student achievement
and educational reproduction: The case of Greece. American Sociological
Review, 55(2):270–279.
BIBLIOGRAPHY
214
Kiernan, K. (1992). The impact of family disruption in childhood on transitions made in young adult life. Population Studies, 46:213–34.
Labov, W. (1972). The logic of nonstandard english. In Giglio, P. P., editor,
Language and Social Context. Penguin.
Lawton, D. (1968). Social Class, Language and Education. Routledge and
Kegan Paul.
Layard, R. and Psacharopoulos, G. (1974). The screening hypothesis and
the returns to education. Journal of Political Economy, 82:985–98.
Levy, P. and Goldstein, H. (1984). Tests in Education. Academic Press,
London.
Lindsay, K. (1926). Social Progress and Educational Waste. Routledge,
London.
Manski, C. F. (1993). Adolescent econometricians. In Clotfelter, C. T.
and Rothschild, M., editors, Studies of Supply and Demand in Higher
Education. University of Chicago Press, Chicago.
Marshall, G., Swift, A., and Roberts, A. (1997). Against the Odds? Social Class and Social Justice in Industrial Societies. Clarendon Press,
Oxford.
Micklewright, J. (1989). Choice at sixteen. Economica, 56:25–39.
Mills, C. W. (1959). The Sociological Imagination. Oxford University Press,
New York.
Mohr, J. and DiMaggio, P. (1995). The intergenerational transmission of
cultural capital. Research in Social Stratification and Mobility, 14.
Murphy, J. (1981). Class inequality in education: Two justifications, one
evaluation but no hard evidence. British Journal of Sociology, 32(2):182–
201.
Murphy, J. (1990). A most respectable prejudice: Inequality in educational
research and policy. British Journal of Sociology, 41(1):29–54.
Nash, R. (1990). Bourdieu on education and social and cultural reproduction.
British Journal of Sociology of Education, 11(4).
NCDS (1993). National Child Development Survey. ESRC Data Archive,
University of Essex.
BIBLIOGRAPHY
215
Nisbet, J. (1953). Family environment and intelligence. Eugenics Review,
XLV:31–42.
Office of Population Censuses and Surveys (1991). Standard Occupational
Classification, volume 1-3. HMSO, London.
Payne, J. (1987). Unemployment, apprenticeships and training: Does it
pay to stay on at school? British Journal of Sociology of Education,
8(4):425–445.
Power, S., Whitty, G., Edwards, T., and Wigfall, V. (1998). Schools, families
and academically able students: contrasting modes of involvement in secondary education. British Journal of Sociology of Education, 19(2):157–
76.
Robinson, R. and Garnier, M. (1985). Class reproduction among men and
women in France. American Journal of Sociology, 91(2):250–280.
Savage, M. and Egerton, M. (1997). Social mobility, individual ability and
the inheritance of class inequality. Sociology, 31(4):645–72.
Shavit, Y. and Blossfeld, H.-P. (1993). Persistent Inequality : changing educational attainment in thirteen countries. Social inequality series. Westview, Boulder, Colo.
Smith, D. and Tomlinson, S. (1989). The School Effect: A Study of MultiRacial Comprehensives. Policy Studies Institute, London.
Sokal, A. and Bricmont, J. (1998). Intellectual Impostures. Profile Books,
London.
Sullivan, A. and Cheung, S.-Y. (2000). The british case. In Ballarino, G.,
editor, Innovation, Flexibility, Training and Education: Link with the
Family Situation. E.U. Commission.
Walford, G. (1994). Choice and Equity in Education. Cassell, London.
Waquant, L. (1989). Towards a reflexive sociology: A workshop with Pierre
Bourdieu. Sociological Theory, 7.
Weiner, G., Arnot, M., and David, M. (1997). Is the future female? In Halsey,
A. H., Lauder, H., Brown, P., and Wells, A. S., editors, Education:
Culture, Economy and Society. Oxford University Press, Oxford.
Willis, P. (1977). Learning to Labour. Saxon House, Farnborough.
BIBLIOGRAPHY
216
Willis, P. (1983). Cultural production and theories of reproduction. In Barton, L. and Walker, S., editors, Race, Class and Education. Croom Helm,
London.
Questionnaire – Private and Confidential
Section 1
Education
1.
Are you:
Male
Female
2.
How many subjects are you studying for GCSE?
1
1
2
………….
2
What subjects are you studying for GCSE?
……………………………………………………………
……………………………………………………………
……………………………………………………………
……………………………………………………………
……………………………………………………………
……………………………………………………………
……………………………………………………………
3-20
……………………………………………………………
……………………………………………………………
……………………………………………………………
3.
What grades do you think you are most likely to get in your GCSE’s?
Tick one box only
Mostly As
1
Mostly As and Bs
2
Mostly Bs and Cs
3
Mostly Cs and Ds
4
Mostly Ds and Es
Mostly Es and Fs
Mostly Fs and Gs
21
5
6
7
Mostly Gs and ungraded
8
1
4.
5.
6.
What do you expect to do after your GCSE’s? Tick one box only.
Study for a BTEC
1

If so, which subject are you likely to do?
……………………………………………
22
Study for a GNVQ
2

If so, which subject are you likely to do?
……………………………………………
23
Retake GCSEs
3
Study for A levels
4

If so, which subjects are you likely to do?
……………………………………………
……………………………………………
Get a job
5
Join a training scheme
6
Other (please specify)
7

……………………………………………
If so, what sort of job?
……………………………………………

……………………………………………
Do you intend to go to university?
Yes
1
No
2
24
How much do you agree or disagree with each of these statements about studying?
Agree
strongly
(a)
School is a waste of time
(b)
The more qualifications you
get, the better the job you are
likely to get
(c)
Studying is worthwhile for its
own sake
(d)
I enjoy studying
(e)
Qualifications are useless for
getting jobs
(f)
Studying can help you to gain
skills which will be useful at
work
Agree
Neither
agree
nor
disagree
Disagree
Disagree
strongly
25
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
26
27
28
29
30
2
Agree
strongly
Agree
Neither
agree
nor
disagree
Disagree
strongly
Disagree
These days, it doesn’t matter
how many qualifications you
get, you still won’t get a job
1
2
3
4
5
(h)
Studying increases your
confidence
1
2
3
4
5
(i)
Studying increases your
ability to think clearly
1
2
3
4
5
(j)
I muck about in lessons
1
2
3
4
5
(k)
Studying is irrelevant to real
life
1
2
3
4
5
(l)
I take school seriously
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
(g)
(m)
These days, if you are
unqualified, you can’t get a
job
I don’t like studying
(n)
31
32
33
35
36
37
38
(o)
Studying improves your
ability to be creative
1
2
3
4
5
(p)
Studying is only ever
worthwhile if it leads to a job
1
2
3
4
5
7.
34
39
40
In general, how do you rate your academic abilities as compared
to other pupils at your school?
Excellent
Above average
Average
Below average
1
41
2
3
4
Poor
5
Employment
1.
What sort of job would you most like to be doing in 10 years time?
42
………………………………………………………………..
2.
What sort of job do you think you will actually be doing in 10 years time?
43
………………………………………………………………..
3
3.
In choosing a job, what things about it do you think are important?
Tick as many boxes as you like.
(a)
The job should involve working with my hands
(b)
It should make others respect me
(c)
It should be an outdoor job
(d)
It should involve using my head and need thought and concentration
(e)
It should allow me to excel
(f)
It should not get in the way of my family responsibilities
(g)
It should give me control over others
(h)
It should involve variety
(i)
It should be well paid
(j)
It should allow me to be creative
(k)
It should have convenient hours and conditions
(l)
It should give me the chance of being with other people
(m)
It should be a clean job
(n)
It should offer chances of promotion
(o)
The job should let me be my own boss
(p)
It should be a secure job
(q)
It should not have too much responsibility
(r)
It should give me the opportunity of helping others
44
1
45
1
46
1
47
1
48
1
49
1
50
1
51
1
52
1
53
1
54
1
55
1
56
1
57
1
58
1
59
1
60
1
61
1
Now look at the list again. Which of the items is:
Most important
2nd most important
3rd most important
………
………
………
(just give the letter of the item)
62
63
64
4
Activities
1.
How often do you do each of these spare time activities (not organised by the school).
Often
(a)
Watching television
(b)
Playing outdoor games and sports
(c)
Playing indoor sports
(d)
Going to parties at friends’ homes
(e)
Going to art galleries or museums
(f)
Voluntary work to help others
(g)
Going to the cinema
(h)
Going to see plays
(i)
Going to classical concerts
(j)
Going to pop concerts
(k)
Playing an instrument
(l)
Listening to classical music
(m)
2.
Listening to pop music
Sometimes
Hardly
ever
Never
65
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
66
67
68
69
70
71
72
73
74
75
76
77
Which TV programmes do you watch regularly? Name as many as you can
……………………………………………………………………………………..
……………………………………………………………………………………..
……………………………………………………………………………………..
78-100
……………………………………………………………………………………..
……………………………………………………………………………………..
……………………………………………………………………………………..
……………………………………………………………………………………..
……………………………………………………………………………………..
……………………………………………………………………………………..
……………………………………………………………………………………..
5
3.
What films have you seen at the cinema recently? (If you have not been to the cinema
recently, move on to the next question).
……………………………………………………………………………………..
……………………………………………………………………………………..
……………………………………………………………………………………..
……………………………………………………………………………………..
4.
About how often do you read books that are not connected to your schoolwork?
Never / Hardly ever
1 per month
1 per fortnight
1 per week
2 per week
1
101-120
121
2
3
4
5
3 or more per week
6
5.
State the titles and/or authors of any books you have read recently that are not connected
with schoolwork.
……………………………………………………………………………………..
122-130
……………………………………………………………………………………..
……………………………………………………………………………………..
……………………………………………………………………………………..
……………………………………………………………………………………..
6.
7.
Are you a member of a public library?
Yes
1
No
2
Yes
1
No
2
Do you have any favourite authors?
If so, who are they?
……………………………………………………………………………………..
……………………………………………………………………………………..
……………………………………………………………………………………..
8.
131
Do you sometimes read a national or local newspaper? If so which one(s)?
……………………………………………………………………………………..
……………………………………………………………………………………..
132
133-150
151-180
……………………………………………………………………………………..
6
Family
1.
Who do you live with?
Adults:
mother and father
1
mother and stepfather
2
father and stepmother
3
just mother
4
just father
other (please specify)
……………………………………………
181
5
6
Brothers and sisters (including step/half brothers and sisters) who are living with you,
or who have left home in the last 3 years:
182
Number of brothers ……………….
Number of sisters ……………….
Ages of brothers …………………….
Ages of sisters …………………….
183
184-190
191-200
In the following questions, I will be asking you about your mother and father.
For some of you, your “natural” parents may not be relevant – answer in terms
of who you see as your mother and father.
2.
3.
Is your mother employed?
Is your father employed?
Yes
1
No
2
Yes
1
No
2
201
202
7
4.
(a)
(b)
(c)
(d)
Please tell me about your parents’ jobs. If they are not working at the moment,
tell me about their most recent jobs.
Father
What is the name of the job? …………………
…………………
What kind of work do they do? …………………
…………………
What sort of organisation do they work for? …………………
…………………
Do they have their own business?
Yes
No
(e)
Do they work full or part time?
Full time
Part time
5.
6.
Mother
…………………
…………………
…………………
…………………
…………………
…………………
Yes
1
205
206
2
Full time
1
207
1
Part time
2
204
1
No
2
203
208
2
If you live with only one parent, or with a stepparent, are you in regular
contact with the other parent (i.e. at least once a fortnight)?
Yes
1
No
2
209
What is the highest qualification that each of your parents has?
Father
No qualifications
CSE
O level
A level / Scottish higher
Vocational qualification
(please specify)
Mother
210
1
1
2
2
3
3
4
4
5
5
211
212
213
………………
Degree
Post graduate qualification
(masters, Ph.D. or D. Phil)
7.
……………….
6
6
7
7
Please state if either of your parents is studying for any of the above qualifications
…………………………………………………………………….
…………………………………………………………………….
214
215
8
8.
Which of the following have you heard your parents discuss?
Often
(a)
Sport
(b)
Art
(c)
Politics
(d)
Books
(e)
TV
(f)
Films
(g)
Work
(h)
Friends
(i)
Science
(j)
Current affairs
(k)
Music
9.
Sometimes
Never
216
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
217
218
219
220
221
222
223
224
225
226
Do your parents get a newspaper regularly (at least once a week)?
Yes
1
No
2
10. Which newspaper(s) do they get?
…………………………………………………
…………………………………………………
…………………………………………………
227
228-250
11. Do your parents listen to the radio?
If so, which station(s) do they listen to?
…………………………………………………
…………………………………………………
…………………………………………………
Yes
1
No
2
251
252-270
9
12. Approximately how many books are in your home?
None, or very few (0-10)
271
1
A few books (11-25)
2
One bookcase full (26-100)
3
Two bookcases full (101-200)
4
Three bookcases full (201-300)
5
Four bookcases full (301-400)
6
About a room full (401-600)
7
More than a room full (601 or more)
8
13. How many rooms are there in your home? Count all rooms including bathrooms ………..
272
14. Does your family own a car?
273
Yes
1
No
2
If your family owns more than one car, how many cars does it own? ………..
15. Does your family have a computer?
16. Did your parents go on holiday last year?
274
Yes
1
No
2
Yes
275
276
1
No
2
If so, where did they go? ……………………………………..
277
17. For each of these activities, do your parents do it often, sometimes, rarely or never?
Often
(a)
Outdoor sports
(b)
Indoor sports
(c)
Watching television
(d)
Listening to music
(e)
Parties at friends’ homes
(f)
Going to art galleries or museums
Sometimes
Rarely
Never
278
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
279
280
281
282
283
10
Often
(g)
Voluntary work to help others
(h)
Going to the cinema
(i)
Reading novels
(j)
Reading non fiction
(k)
Going to see plays
(l)
Going to concerts
(m)
Playing a musical instrument
(n)
Political meetings
(o)
Evening or daytime classes
Sometimes
Rarely
Never
284
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
285
286
287
288
289
290
291
292
18. If your parents listen to music, what kind(s) of music do they listen to?
…………………………………………………..
…………………………………………………..
293-310
19. Do your parents have any friends who have the following jobs?
(a)
Secretary
(b)
Nurse
(c)
Doctor
(d)
Manual worker
(e)
Unemployed
(f)
Academic
(g)
Teacher
(h)
Social worker
(i)
Cleaner
(j)
Student
(k)
Manager
(l)
Police
(m)
Foreman or supervisor
(n)
Clerical worker
(o)
Business owner
311
1
312
1
313
1
314
1
315
1
316
1
317
1
318
1
319
1
320
1
321
1
322
1
323
1
324
1
325
1
11
(p)
Technician
(q)
Accountant
(r)
Lawyer
(s)
Shop Worker
(t)
Housewife
326
1
327
1
328
1
329
1
330
1
Section 2
Famous people
Each of the following names is a person you may have heard of. For each person listed, which
do you associate him or her with most out of the following categories: politics, music, novels,
art or science? If you do not know, do not guess, just tick “don’t know”.
Politics
Example:
Tony Blair
(a)
Albert Einstein
(b)
Jane Austen
(c)
Bill Clinton
(d)
Graham Greene
(e)
Wolfgang Mozart
(f)
Claude Monet
(g)
Karl Marx
(h)
Andy Warhol
(i)
Virginia Woolf
(j)
Marie Curie
(k)
Rachmaninov
(l)
Martin Amis
(m)
Galileo
(n)
Pablo Picasso
(o)
Rembrandt
Music
Novels
Art
Science
Don’t
know

331
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
332
333
334
335
336
337
338
339
340
341
342
343
344
345
12
Politics
(p)
Miles Davis
(q)
Charles Dickens
(r)
Louis Pasteur
(s)
Mahatma Gandhi
(t)
George Gershwin
(u)
John F. Kennedy
(v)
Stephen Hawking
(w)
Vincent van Gogh
(x)
Gordon Brown
(y)
Johannes Brahms
Music
Novels
Art
Science
Don’t
know
346
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
347
348
349
350
351
352
353
354
355
Comprehension
Underline the word that correctly completes the sentence. If you do not know which word
is correct, just move on to the next question.
Example:
You (fill, milk, boil, match, paint) water to make tea.
1.
About a week after (ordering, receiving, emptying, managing, upsetting) this letter,
I had a telegram from my brother.
2.
You can arrange a (money, summer, journey, job, weather) through a travel agent.
3.
She had been dieting for a month, but her weight had not (shown, increased, shrunk,
decreased, grown).
4.
Despite the (convenience, expense, poverty, economy, pleasure) of living in this area,
I have managed to save some money.
5.
The building that was going to be demolished had been (derisive, animated, vicious,
derelict, derivative) for some months.
6.
The (process, proceeds, page, gain, progress) from the sale of the book were given to a
worthwhile charity.
356
357
358
359
360
361
13
7.
It is not sufficient to know the means of preventing and curing disease; it is equally
(bad, necessary, useful, sufficient, good) to provide these means and even to compel
their use.
8.
The town council wanted to improve the (amounts, limits, amenities, places, inmates)
of their district.
9.
As we both come from the same town, my wife and I have a great many (typical,
mutual, friable, arable, viable) friends.
10. The discovery made by the explorers in the Arabian desert far (overtook, tormented,
exceeded, exposed, passed) their wildest dreams.
11. When the speaker asked if there were any questions he was (involved, immured,
inundated, implied, instructed) with queries.
12. Having had her expectations so much raised, it was very (realistic, discrediting,
uplifting, disconcerting, discriminating) to have them suddenly shattered.
13. It took three men to (refrain, prohibit, restrain, catch, restrict) the youth from rushing
into the burning shed to rescue his pet dog.
14. The boss firmly denied any accusation of discrimination; he claimed that the policies of
his firm were not dictated by (shareholders, prejudice, incrimination, accusations,
profit).
15. Although I won the debate, my opponent failed to acknowledge that I had (remaindered,
362
363
364
365
366
367
368
369
370
refuted, disputed, conquered, denied) his argument.
14
Synonyms
Synonyms are words that mean the same or approximately the same as each other. For example
happy and cheerful are synonyms.
For each of the following questions think of as many synonyms as you can for each word (stop
at five for each word). Here are two examples:
Disgusting …………………………………………
repellent,
nasty
…………………………………………
vile
…………………………………………
unpleasant
…………………………………………
obscene
…………………………………………
1.
Big
enormous
…………………………………………
gigantic
…………………………………………
huge
…………………………………………
massive
…………………………………………
obese
…………………………………………
Small
…………………………………………
371
…………………………………………
…………………………………………
…………………………………………
…………………………………………
2.
Stupid
…………………………………………
…………………………………………
…………………………………………
…………………………………………
372
…………………………………………
3.
Angry
…………………………………………
…………………………………………
…………………………………………
…………………………………………
…………………………………………
373
15
5.
Sad
…………………………………………
374
…………………………………………
…………………………………………
…………………………………………
…………………………………………
6.
Odd
…………………………………………
…………………………………………
…………………………………………
…………………………………………
375
…………………………………………
You have reached the end of this questionnaire.
Please use any remaining time to go over your answers.
16
D.Phil
Note, the PDF file has the wrong date.
Correct citation:
Sullivan, A. 2000. Cultural Capital, Rational Choice, and Educational Inequalities.
D.Phil. Thesis, Oxford University.
Publications from the doctorate:
Sullivan, A. 2001. ‘Cultural Capital and Educational Attainment’ Sociology. 35(4)
893-912.
Sullivan, A. 2002. ‘Bourdieu and Education: How Useful is Bourdieu’s Theory for
Researchers?’ Netherlands Journal of Social Sciences. 38(2) 144-166
Sullivan, A. (2006) ‘Students as Rational Decision-Makers: the question of beliefs
and attitudes’ London Review of Education 4(3) 271-290.
Sullivan, A. 2007. ‘Cultural Capital, Cultural Knowledge and Ability’ Sociological
Research Online 12(6).

Cultural Capital, Rational Choice and Educational Inequalities

Transcription

Similar documents

dec 2010 newsletter - Hindley High School

March 2016 - Issue 128

G azzette - CHIJ Our Lady Queen of Peace

Parents` Newsletter November 2015

Winter 2014 - Crofton Academy

The continental jigsaw puzzle - Earth Science Teachers` Association

Teachers Pack

Key Stage 1 Multi Skills Festival

July 2013 - St Michaels

the entire PDF-Newsletter