423061
2012
ISS0010.1177/0268580911423061Katrnak et al.International Sociology
Article
The relationship between
educational homogamy and
educational mobility in 29
European countries
International Sociology
27(4) 551­–573
© The Author(s) 2012
Reprints and permission: sagepub.
co.uk/journalsPermissions.nav
DOI: 10.1177/0268580911423061
iss.sagepub.com
Tomáš Katrňák
Masaryk University, Czech Republic
Petr Fučík
Masaryk University, Czech Republic
Ruud Luijkx
Tilburg University, The Netherlands
Abstract
This article deals with the relationship between educational homogamy and educational mobility
in 29 European countries. It answers three interrelated questions: (1) Is there any relationship
between educational mobility and educational homogamy? (2) Does educational homogamy
diverge from educational mobility (negative relationship) or does educational heterogamy
strengthen educational mobility (positive relationship)? (3) If there is any positive relationship,
do educational mobility and educational heterogamy indicate the level of educational inequality
to the same degree? To answer these questions the authors use data from three waves of the
European Social Survey (2002, 2004 and 2006). The answers are given in absolute (percentages)
as well as in relative measures (log-multiplicative effects). The results show that there is a positive
relationship between educational mobility and educational heterogamy. For all countries analysed,
relative educational heterogamy is a stronger indicator of educational inequalities than relative
educational mobility. The systematic deviation of educational heterogamy from educational
mobility is explained by a number of factors that the authors discuss in the last part of the article.
Keywords
Assortative mating, educational homogamy, educational mobility, European Social Survey,
log-linear models, log-multiplicative models
Corresponding author:
Tomáš Katrňák, Masaryk University, Faculty of Social Studies, Joštova 10, Brno 602 00, Czech Republic.
Email: katrnak@fss.muni.cz
552
International Sociology 27(4)
In social stratification research, the analysis of educational homogamy supplements the
analysis of social or educational intergenerational mobility because educational homogamy is also considered to be an indicator of ‘societal openness’ (Smits et al., 1998a,
1998b, 2000; Ultee and Luijkx, 1990).
High social or educational mobility implies high social or educational motion.
Representatives of social classes move within the class structure; barriers between them
are low and easy to overcome. Solidarity and social ties are expected across the class
structure. When social or educational mobility does not exist, social movement by members of society is negligible and people’s class status remains unchanged throughout the
course of their lives. Descendants live and die within their parents’ class background;
barriers between classes are too high to be overcome and there are no mobility channels
between classes. Social ties and solidarity primarily exist within the respective classes.
This class structure is defined as closed and tight.
Low educational homogamy (high educational heterogamy) implies that the coupling
of people with different educational statuses is widespread. Differently educated people
enter into marriage because societal distances between educational levels are not as large
and significant as they assert. Marital bonds do exist across the educational structure. On
the other hand, there is high educational homogamy (low educational heterogamy) when
marriages between people with different educational levels are not very common.
Societal barriers between educational levels are reflected in partner choices. Marriages
tend to take place within educational categories rather than across them. In this case,
educational levels must be considered as predictors of assortative mating.
Social mobility refers to the labour market, educational mobility to the educational
structure. High social and educational mobility is considered to be a condition for the
prosperity of national states (Blau and Duncan, 1967; Breen, 2004). Socially and educationally open societies work on the basis of the merit principle and are stratified according to the acquired and not prescribed qualities of a person, which has a beneficial effect
on the functioning of their economies. Homogamy refers to conditions in the marriage
market and indicates to what extent the choice of a spouse is influenced by social, cultural and economic criteria.
All of these concepts draw on the ‘openness’ of the social and educational structure;
thus, a relationship should exist between them. According to Ultee and Luijkx (1990),
high social mobility should go hand in hand with low educational homogamy. In this
case people overcome class differences relatively easily (whether inter- or intragenerationally); they similarly overcome educational differences when entering into marriage.
In contrast, low social mobility is linked to high educational homogamy (Ultee and
Luijkx, 1990). Here class barriers are too high to be easily overcome, just as educational
inequalities are too large and decisive for assortative mating. In such societies, people are
selected on the basis of ascribed characteristics, such as gender, race, religion or social
origin, which retards the economic development of a society (Blau and Duncan, 1967).
In this article, we focus on educational mobility and study the relationship between
this mobility and educational homogamy. Investigating the relationship between these
two phenomena is important for the study of social stratification. Both concepts are seen
by social stratification researchers as indicators of distances between levels of education.1 However it is not clear yet which of these indicators is going to be more important
Katrňák et al.
553
for the study of educational inequalities. Which is it: educational mobility or educational
homogamy? Another problem is whether these indicators measure the distances between
degrees of education by the same yardstick. It is quite possible that each of these indicators ‘captures’ a different dimension of educational inequality, and that each indicates
something different about the distances between levels of education. These problems are
more methodological than practical; nevertheless they are important for the study of
social stratification, and so far have not been convincingly answered. We hope this study
will shed some light on these problems.
The subject of this article therefore is the relationship between educational mobility
and educational homogamy. The article specifically addresses three interrelated questions. (1) Is there a relationship between educational mobility and educational homogamy? (2) Is this relationship positive or negative? Alternatively, does educational
homogamy decrease (compensate for) educational mobility (negative relationship) or
does educational heterogamy strengthen educational mobility (positive relationship)?
(3) If there is any positive relationship, do educational mobility and educational heterogamy indicate the level of educational inequality with the same intensity? Alternatively,
is one of these indicators systematically stronger than the other?
In the first section of the article we focus on the relationship between educational
mobility and educational homogamy as proposed in four generations of social stratification research (Ganzeboom et al., 1991; Treiman and Ganzeboom, 2000). The next section of the article presents two contradictory hypotheses on the relationship between
educational mobility and educational homogamy that involve our three research questions. The methodological section deals with an analytical approach to measuring educational mobility and educational homogamy. Further on, we analyse the relationship
between these two social phenomena, first in absolute figures and then on the relative
level, test our hypotheses and offer answers to our research questions.
Educational mobility and educational homogamy in
social stratification research
Four generations of researchers have distinguished themselves in the field of social stratification analysis (Ganzeboom et al., 1991; Treiman and Ganzeboom, 2000). The first
generation (in the 1950s and 1960s) did not empirically test the relationship between
educational mobility and educational homogamy; nonetheless, data on marriages were
used several times as an indicator for the openness of the social structure (see Berent,
1954; Hall, 1954; Lipset and Zetterberg, 1959). Nor was the relationship between educational mobility and educational homogamy empirically tested by the second generation
(active in the 1960s and 1970s). Blau and Duncan (1967) measured only spouses’ closeness in terms of social background (indicated by the employment status of the spouses)
and the difference in the spouses’ educational level. The association between the spouses’
education turned out to be stronger than that between their social backgrounds. Thus
education structured the marital choice more than the social background in American
society.
In the third generation (dominant in the 1970s and 1980s), the relationship between
social mobility and educational homogamy was theoretically laid out (see Hout, 1982;
554
International Sociology 27(4)
Sixma and Ultee, 1984) and empirically tested as well. Ultee and Luijkx (1990) analysed
social mobility along with educational heterogamy in 23 countries, and found a correlation of .51 between the proportion of intergenerationally socially mobile men and the
proportion of educationally heterogamous couples. The correlation was even higher (.67)
between the relative measures (net changes in the marginal distributions of mobility and
in educational homogamy) of both types of data: relative mobility opportunities and relative opportunities for educational heterogamy seem to be interrelated. A relationship
between social mobility and educational homogamy was found. Unfortunately, none of
the social stratification researchers of this generation analysed the relationship between
educational mobility and educational homogamy.
A fourth generation is still in the process of establishing itself (Treiman and
Ganzeboom, 2000) and continues the legacy of the three previous generations of social
stratification research. They base their work on the statistical-methodological tools
developed by previous generations. Nevertheless, they do not focus so much on developing these tools further, but instead use them to answer specific and down-to-earth research
questions. The turn towards the substantive issues of the social stratification process
instead of its methodological and statistical specification characterizes this fourth generation of researchers. The relationship between social mobility and educational homogamy as indicators of ‘societal openness’ was conceptually developed by the representatives
of this generation (see Smits et al., 1998a, 1998b). However, it has not yet been empirically
tested. This generation of researchers, in particular those focusing on log-multiplicative
modelling, measure social and educational mobility and compare their trends within
nation-states (see Breen, 2004; Ganzeboom et al., 1989; Wong, 1990, 1992), or measure
educational homogamy and analyse its trends in distinct countries (see e.g. Blossfeld
and Timm, 2003; Kalmijn, 1991; Mare, 1991; Qian, 1998; Qian and Preston, 1993; Uunk
et al., 1996) or compare various countries (see e.g. Raymo and Xie, 2000; Smits et al.,
1998a, 1998b, 2000). They regard social and educational mobility as well as educational
homogamy as the main indicators of the openness of the social structure. Still, these
indicators have not been examined together at the same time within the framework of
nation-states.
What kind of a relationship can we expect between
educational mobility and educational homogamy?
According to some theoretical assumptions (Blau and Schwarz, 1983; Goody, 1983;
Jones, 1987; Schumpeter, 1951), we would expect a positive relationship between educational mobility and the assortative mating process. In this case educational reproduction and educational homogamy (or educational mobility and educational heterogamy)
go hand in hand (Blau and Schwarz, 1983; Goody, 1983; Ultee and Luijkx, 1990). But
according to other theoretical assumptions (Jones, 1987), we expect the relationship to
be negative. Educational mobility and educational homogamy (or educational reproduction and educational heterogamy) act against one another, or neutralize each other. More
educational mobility in a society means less educational heterogamy and vice versa.
Let us focus on these two relationships in more detail. In the first case, restricted
educational mobility and strong class and educational reproduction also imply a
Katrňák et al.
555
considerable social and educational similarity between partners (Ultee and Luijkx,
1990). Between educational reproduction and educational homogamy there is a positive
correlation. Class and educational structure is rigid, and barriers and differences between
social classes are high. It is difficult to overcome them from one generation to another.
The majority of descendants end up in the same social class and have the same education
as their parents. At the same time, they take as a husband or wife a person belonging to
the same social class and having the same education. On a microstructural level this positive link between educational reproduction and educational homogamy means that it is
above all the norms, values and ‘habitus’ (patterns of behaviour) of a person (Bourdieu,
1989), together with family origin, which structures the course of his/her education and
influences the selection of his/her spouse. The educational reproduction of a person also
limits his/her range of opportunities to meet a potential partner with a different level of
education (Kalmijn, 1991, 1998; Mare, 1991). Not only that person’s values, but his/her
structurally conditioned opportunities tend to greatly increase the chances of an educationally homogamous marriage.
A positive correlation also means that there is a link between educational mobility
and educational heterogamy. In educationally open societies, where the chances to
obtain education do not depend on social origin, descendants from various social origins acquire various levels of education, and pass through various social environments.
For them, besides advancement over their parents, educational mobility is a way of
broadening their environment, and a way of adopting new patterns of behaviour
(Bourdieu, 1989). The structurally conditioned opportunities to meet various possible
marriage partners are thus greater (Kalmijn, 1991, 1998). An educationally mobile person is freed from the environment of the origin family, and takes on a new ‘habitus’
(patterns of behaviour) in the educational system (Bourdieu, 1989). His/her cultural
horizon is broader than that of a person remaining intergenerationally at the same level
of education. Models of behaviour adopted during the mobility movement expand the
acceptability of potential partners from various cultural environments, and allow the
person to pass through diverse social environments which structure the selection of a
mate (Kalmijn, 1991, 1998; Mare, 1991).
In the second case we should expected that high educational mobility means a low
educational heterogamy and vice versa. The correlation between educational mobility
and educational heterogamy is negative. The barriers between social classes are small.
The descendants of the parents belonging to different social classes end up with different
educations (regardless of whether their educational mobility is upward or downward)
and their education matches that of their husband or wife. Educational mobility weakens
the survival of social classes from one generation to another, but educational homogamy
strengthens the social classes and contributes to the current inequalities in the labour
market (Jones, 1987). On a microstructural level this means that social origin does not
predict the achieved education level. Expansion of the educational system weakens the
effect of the parents’ education on the education of their offspring. However, because it
is above all the educational system that structures the selection of a spouse (Blossfeld,
2009; Blossfeld and Timm, 2003; Mare, 1991), the same expansion of the educational
system assures that young people will encounter one another mainly within its bounds,
and thus form educationally homogamous marriages.
556
International Sociology 27(4)
Nevertheless, it can be also the other way around; the mechanism of social and educational mobility can strengthen the intergenerational coherence of social classes, yet the
mechanism of assortative mating damages the coherence of social classes. In this case
there is a negative link between educational reproduction and educational homogamy
(i.e. educational reproduction is high and educational homogamy is low). The descendants have the same education as their parents; however, they take as a husband or wife a
person of another educational level. Educational mobility acts to benefit the stability of
the social structure, but the choice of a husband or wife disrupts that stability. On a
microstructural level this means that the offspring of families with less education do not
have the chance to gain higher education. The educational system is not opened. This of
course means that the educational system does not determine the selection of a spouse.
Instead people tend to meet one another through employment or other social or cultural
places, where the chance of finding a mate with the same level of education is less than
in the educational system (Kalmijn, 1991).
When we take into account these two possible views on the relationship between
educational mobility and educational heterogamy, we can formulate two hypotheses:
H1: High educational mobility (low educational reproduction) indicates ‘societal openness’
and is linked to high educational heterogamy (low educational homogamy). The relationship
between educational mobility and educational heterogamy (or between educational reproduction
and educational homogamy) is positive.
H2: High educational mobility (low educational reproduction) indicates ‘societal openness’,
nevertheless is not positively linked to high educational heterogamy (low educational
homogamy). The relationship between educational mobility and educational heterogamy (or
between educational reproduction and educational homogamy) is negative.
Description of the data
We test the relationship between the educational mobility and the educational homogamy as the relationship between the parents’ educational background and the respondent’s educational background (educational mobility), and the respondent’s education and
the education of his/her spouse (educational homogamy).
Within the nation-state framework, we begin with the respondent, first focusing on
mapping the relationship between his/her educational background and his/her educational level, and second on the relationship between the respondent’s education level and
his/her spouse’s educational level (see Figure 1). We do not analyse either the relationship between the parental educational background of the respondents’ spouses and the
spouse’s educational level, or the relationship between the respondent’s educational
background and the educational background of his/her spouse.
We answer our three research questions (and test the two interrelated hypotheses)
using data from the European Social Survey (ESS). In our investigation, we analysed 29
countries from three waves of ESS data collection (years 2002, 2004 and 2006).2 For
each country, we created a four-way contingency table combining the respondent’s gender, education of his/her parents (always the higher education of one of the parents),
respondent’s education and spouse’s education. We constructed this table only for people
Katrňák et al.
557
Figure 1. Analysed relationships.
on the labour market (age boundary 25–65 years). In cases where a country was repeatedly included in separate waves of ESS (in all three years or in any two years), we
merged this four-way table into one.3 This contingency table design allows us to combine
the measures of educational mobility and educational homogamy appropriate for loglinear and log-multiplicative modelling.
Education was originally measured using the seven categories of ISCED97.4 We collapsed this into a three-category educational variable by merging categories (0), (1) and
(2); categories (3) and (4); and finally categories (5) and (6). This was a necessary step
because of the sufficient number of cases in table cells. Thus, we work with primary,
secondary and tertiary education in our analysis.
Figures 2 and 3 present the relationship between absolute educational reproduction
and absolute educational homogamy in the 29 countries analysed. On the x-axis, we
present the proportion of educationally immobile in each country (those people who did
not change their educational level in comparison to their parents), while on the y-axis the
proportion of those who entered into educationally homogamous marriages is represented. The data are presented separately for men (Figure 2) and women (Figure 3).
In both cases, we can see a clear tendency for the countries analysed to aggregate
around the regression line. In countries where we identify a lower proportion of educational shifts between parents and their descendants, there are also fewer marriages
between partners with different educational levels and vice versa; absolute higher educational reproduction indicates a higher proportion of educationally homogamous marriages. This relationship is slightly stronger for women than for men (the value of the
correlation coefficient is .60 for men and .69 for women). There are three outlier countries that differ significantly from the others (Ukraine, Bulgaria and Iceland). If we use
the dataset without these countries, the correlation rise to .78 for men and .82 for women,
so the gender differences tend to diminish.
A relationship between absolute educational mobility and absolute educational heterogamy exists (it is relevant to our first research question), and it is positive (relevant to
our second research question). Our first hypothesis (H1) assumed that the relationship
between educational mobility and educational heterogamy is positive (higher educational mobility means higher educational heterogamy); our second hypothesis (H2)
educational homogamy (%)
558
International Sociology 27(4)
84
82
80
78
76
74
72
70
68
66
64
62
60
58
56
54
52
50
48
46
44
42
40
fitted values
y = 40.27 + 0.46x
UA
PT
TR
CZ
BG
GR
IT
SK
ES
AT
CY
SI
IE
CH
GB FR
SE
DE
LU
IL
NO
EE
DK
BE
FI
NL
HU
PL
IS
correlation 0.60
34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88
educational reproduction (%)
educational homogamy (%)
Figure 2. Absolute educational reproduction and absolute educational homogamy in 29
countries (men)
84
82
80
78
76
74
72
70
68
66
64
62
60
58
56
54
52
50
48
46
44
42
40
PT
fitted values
y = 36.35 + 0.51x
TR
BG
UA
GR
IT
SK
CZ
ES
AT
CH
IE
IL
CY
SE
EE
FI
NO
DK
BE
NL
PL
FR
LU
HU
SI
GB
DE
correlation 0.69
IS
34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88
educational reproduction (%)
Figure 3. Absolute educational reproduction and absolute educational homogamy in 29
countries (women)
suggested that this relationship is negative (higher educational mobility means lower
educational heterogamy). On the absolute level, we have to reject the second hypothesis.
In societies where a higher proportion of people have different educational levels than
559
Katrňák et al.
85
80
75
70
%
65
60
55
50
45
40
35
30
CY UA IS EE IL NO NL BG FI BE DK SK LU SE GB IE FR SI CH DE HU GR PL ES IT AT CZ TR PT
countries
edu reproduction
edu homogamy
Figure 4. Educational reproduction and educational homogamy in 29 countries – sorted by
educational reproduction (men)
their parents (educational mobile people), we also found a higher proportion of educational heterogamous marriages. It seems that the social mechanisms that influence absolute educational mobility and provide descendants with different educational levels than
their parents (mechanisms making societies more educationally open) may influence the
proportion of achieved marriages by different educational levels of spouses.
Figures 4 (for men) and 5 (for women) show the same data as Figures 2 and 3, but
from a different perspective. In both figures, the countries analysed are situated on the
x-axis. We can see the proportion of educationally reproductive respondents and educationally homogamous respondents (y-axis) for each country. Countries are sorted from
the lowest to the highest proportion of educational reproduction, from the left side to the
right side of the graphs. With the exception of Germany (men as well as women) and
Austria (women), we observe that the proportion of educationally homogamous respondents is higher then the proportion of respondents mirroring the educational level of their
parents in all the countries studied. To answer our third research question, we can assume
that the probability for the same level of absolute educational mobility and absolute educational heterogamy is not the same. With the exception of one or two countries, absolute
educational homogamy is higher then absolute educational reproduction.
Analysis and results
Absolute educational mobility and absolute educational homogamy are to some extent
structurally determined (Breen, 2004; Erikson and Goldthorpe, 1992). Therefore, our
next step is to use log-linear and log-multiplicative models that analyse the associations
in the frequency tables net of marginal distributions. In this case, the results describe the
560
International Sociology 27(4)
85
80
75
70
%
65
60
55
50
45
40
35
30
CY IS IL EE FI UA NO NL SE BG BE IE DK PL LU FR SI SK GB HU CH DE IT GR ES CZ AT PT TR
countries
edu reproduction
edu homogamy
Figure 5. Educational reproduction and educational homogamy in 29 countries – sorted by
educational reproduction (women)
relative educational mobility and relative educational homogamy, which refer to unequal
educational opportunities in the educational system (educational mobility) or the marriage market (educational homogamy) (for more information on these analyses, see
Goodman and Hout, 1998, 2001; Hout, 1983; Powers and Xie, 2000; Xie, 1992;
Yamaguchi, 1987).
We analysed a four-way contingency table. The structure of the data was 29 × 3 × 3 ×
3 (Country × Parents’ education × Respondent’s education × Spouse’s education). The
analysis was run separately for men and women. We tested whether a relationship
between educational mobility and educational homogamy exists (our first question),
what form it takes (our second question) and measured the relative intensity for both
concepts (our third research question) against empirical reality. In terms of our hypotheses, we have examined if educational mobility is linked to educational heterogamy
(positive relationship – H1) or if it is connected to educational homogamy (negative
relationship – H2).
Figure 6 shows a graphic representation of the possible relationships between educational reproduction (PR – the association between parents’ and respondent’s educational
level) and educational homogamy (RS – the association between the respondent’s and
his or her spouse’s educational level) in every analysed country (C), tested using loglinear and log-multiplicative models. The model equation of table frequencies completes
each graphic representation for these relationships. If we do not take into consideration
the model 0, in which we assume the non-existence of educational reproduction as well
as educational homogamy, the graphic representations of relationships (and their model
equations) are sorted by numbers according to the concept of parsimony (for model 1, we
need to estimate the most parameters, for the last model 5, the least parameters).
561
Katrňák et al.
CPRS
Model 0: log( Fhijk
) = λ + λhC + λiP + λ jR + λkS + λhiCP + λhjCR + λikCS
CPRS
Model 1: log( Fhijk
) = λ + λhC + λiP + λ jR + λkS + λhiCP + λhjCR + λikCS + ϕh µi µ j +ψ h µ j µk
level of association
RS
PR
C - countries
CPRS
Model 2: log( Fhijk
) = λ + λhC + λiP + λ jR + λkS + λhiCP + λhjCR + λikCS + ϕ h µi µ j + cϕ h µ j µ k
level of association
RS
PR
C - countries
level of association
CPRS
Model 3: log( Fhijk
) = λ + λhC + λiP + λ jR + λkS + λhiCP + λhjCR + λikCS + ϕ h µi µ j + ϕ h µ j µ k
PR = RS
C - countries
level of association
CPRS
Model 4: log( Fhijk
) = λ + λhC + λiP + λ jR + λkS + λhiCP + λhjCR + λikCS + ϕµi µ j +ψµ j µk
RS
PR
C - countries
(Continued)
562
International Sociology 27(4)
Figure 6. (Continued)
level of association
CPRS
Model 5: log( Fhijk
) = λ + λhC + λiP + λ jR + λkS + λhiCP + λhjCR + λikCS + ϕµi µ j + ϕµ j µk
PR = RS
C - countries
Figure 6. Relationships between educational reproduction (PR) and educational homogamy
(RS) by countries (C) and equations of these relationships in log-linear and log-multiplicative
models.
In the first case (model 1), we assume that there are differences in the association between
parents’ and respondents’ education level (PR – educational reproduction) and in the
association between respondents’ and their spouses’ education level (RS – educational
homogamy) among countries. In the second case (model 2), we suppose that there are
differences in both associations among countries as well. Nevertheless, the distance
between educational reproduction (PR) and educational homogamy (RS) is constant in
each country. In the third case (model 3), the countries differentiate in the association of
educational reproduction and homogamy. However, the intensity of educational reproduction and educational homogamy is the same in every country (PR = RS). In the next
model (model 4), we assume that the educational reproduction (PR) and educational
homogamy (RS) are the same in all countries, and the distance between them is also the
same. In the last case (model 5), countries do not diverge either by educational reproduction (PR) or by educational homogamy (RS), and there are no divergences of the two
measures among the countries analysed (PR = RS).
Table 1 presents the goodness-of-fit statistics of all estimated models for men and
women.5 Model 0 is a null association model. It assumes that the educational associations between parents and respondent (PR) as well as between respondent and his or her
spouse (RS) do not exist. In both tests (men and women), this model poorly represents
the data – it has a positive BIC value (Raftery, 1986, 1995) and it misclassifies about
27% of the cases.
The structure of association for all the next five models is the same for both sexes:
row and column scores for educational mobility (PR) and educational homogamy (RS)
are estimated as the same. This is the row and column association model RCII (Goodman,
1979; Clogg, 1982a; 1982b), where row and column scores are equal (mi = mj = mk). In
model 1, we assume that the levels of association PR (educational reproduction − φ
parameters) and association RS (educational homogamy − ψ parameters) are different in
all countries (C), and distances between these levels of association PR and RS are also
different in all countries (C) (equation 1, Figure 6). In model 2 (compare to model 1), we
captured 28 more parameters. This implies that the 29 parameters ψ for the levels of
3675.32 10.96% −2343
578
Notes: C: country; P: parents’ education level; R: respondent’s education level; S: spouse’s education level; L2: log-likelihood ratio chi-square statistic; d.f.: the degrees
of freedom; BIC: Bayesian information criterion; Δ: the index of dissimilarity that indicates the proportion of cases misclassified by the model.
3740.32 10.62% −2315
3639.23 10.27% −2405
8.95% −2957
8.54% −3083
3498.63 10.43% −2509
2804.22
2668.25
577
9.64% −2781
9.12% −2970
2945.64
BIC
550
Δ
2746.18
L2
549
BIC
18449.76 27.40% 12411 19999.17 27.34% 13923
2685.83 8.93% −2739 2586.20 8.28% −2872
Δ
Women
580
521
Men
0 CP CR CS
Null association PR and RS
1 model 0, PR = RS, RCII model for PR and RS association, equal row and column
jPR|C yRS|C
distances PR and RS, levels of association PR and RS (phi and
psi parameters) are heterogeneous by C, distances between
levels of associations PR and RS are different by C
2 model 0, PR = RS, RCII model for PR and RS association, equal row and column
distances PR and RS, levels of association PR and RS (phi and
jPR|C c*jRS|C
psi parameters) are heterogeneous by C, distances between
levels of associations PR and RS are constant by C (it means
yRS|C = c*jRS|C)
3 model 0, PR = RS, RCII model for PR and RS association, equal row and column
jPR|C = yRS|C
distances PR and RS, levels of association PR and RS (phi and
psi parameters) are heterogeneous by C, distances between
levels of associations PR and RS do not exist by C
4 model 0, PR = RS, RCII model for PR and RS association, equal row and column
distances PR and RS, levels of association PR and RS (phi and
jPR yRS
psi parameters) are homogeneous by C, distances between
levels of associations PR and RS are constant by C
5 model 0, PR = RS, RCII model for PR and RS association, equal row and column
distances PR and RS, levels of association PR and RS (phi and
jPR = yRS
psi parameters) are homogeneous by C, distances between
levels of associations PR and RS do not exist by C
d.f.
L2
Description of model
Model
Table 1. Statistics of models fit for the analysis of the relationship between educational mobility and educational homogamy.
Katrňák et al.
563
564
International Sociology 27(4)
association RS (educational homogamy) in every country from model 1 are replaced by
1 parameter c that shows a constant distance among the levels of PR association (educational reproduction) and RS association (educational homogamy) in each country (equation 2, Figure 6). Model 3 originates from model 2, but presumes that there is no distance
between educational mobility and educational homogamy (levels of associations PR and
RS are the same). In this model, we capture one more parameter that describes the distance between the levels of educational mobility (PR) and educational homogamy (RS)
in each country (equation 3, Figure 6) in model 2. Model 4 is the same as model 2, but
more restrictive. Like model 2, it assumes that the distances between the levels of PR and
RS associations are constant by c. Moreover, it assumes that these levels of associations
are constant in all countries analysed. To compare this model to the second one, we saved
28 more parameters in this model, which implies that the 29 parameters for the identification of the levels of association PR (educational reproduction) in each country (C) are
replaced by one parameter. In order to compare to model 3, we captured 27 more parameters in model 4. It suggests that the 29 parameters for the identification of the two levels
of associations PR and RS (educational reproduction and educational homogamy) in
each country are replaced by two parameters. One of them is given for the identification
of the PR level association for all countries; the other is given for the identification of the
RS level of association for all countries, thus the distance between PR and RS associations is the same in each country (see equation 4, Figure 6). Finally, model 5 is similar to
model 3. It supposes that there is no distance in the levels of PR and RS associations
(educational reproduction is the same as educational homogamy). Moreover, in comparison to model 3, it assumes that these associations are the same for all countries analysed.
When comparing to model 4, we saved one more parameter in this model showing the
level of RS association and indicating the distance between PR and RS association in
model 4 (equation 5, Figure 6)
The comparison of the BIC criteria for these five models suggests that we should
interpret the data based on model 2 (for men as well as for women). However, because
the conditional L2 test of model 2 to model 1 is statistically significant (for men and
women, alpha criterion 5%), and according to classical fit statistics (L2 and delta) the fit
of model 1 is better than the fit of model 2, we thus interpret the data with the help of both
these models.
Figure 7 shows the results from model 1 for men and Figure 8 for women (φ and ψ
parameters from equation 1, Figure 6). In this case, these parameters show a relative
deviation in the levels of association against the first parameter (here for Austria). In both
figures, we see that a stronger association between parents’ and respondent’s education
(higher relative educational reproduction) means a stronger association between respondent’s and his or her spouse’s education (higher relative educational homogamy). In both
figures, the correlation between relative educational reproduction and relative educational homogamy (the correlation between φ and ψ parameters from model equation 1,
Figure 6) is positive and high (for men .81, for women .86).
Given these results, we have to reject our second hypothesis (H2) assuming a negative
relationship between educational mobility and educational heterogamy. The results from
model 1 show that higher educational mobility opportunities go hand in hand with higher
educational heterogamy opportunities for both men and women. A positive relationship
565
Katrňák et al.
.65
BG
UA
fitted values
y = 0.01 + 1.34x
.60
educational homogamy (psi)
.55
CZ
.50
GR
.45
SK
.40
.30
.25
NL
.20
.15
GB
IS
EE
NO
FI
SE
IL
IT
PT
PL
TR
HU
SI IE
DE
DK
CY
ES
LU
BE
AT
.35
CH
FR
.10
correlation 0.81
.05
.10
.15
.20
.25
.30
.35
.40
.45
educational reproduction (phi)
Figure 7. Relative educational reproduction and relative educational homogamy in 29
countries (men – model 1)
.65
fitted values
y = -0.04 + 1.45x
.60
BG
educational homogamy (psi)
.55
.50
.45
UA
.40
.35
CH
LU
.25
.20
DK
NO
FI
SE
.15
IT
TR
IL
.30
.10
CZ
SK
GR
PL
PT
IE
EE
DE SI
ES
AT
BE
CY
HU
FR
NL
GB
IS
correlation 0.86
.05
.10
.15
.20
.25
.30
.35
.40
.45
educational reproduction (phi)
Figure 8. Relative educational reproduction and relative educational homogamy in 29
countries (women – model 1)
exists between relative educational mobility and relative educational heterogamy. In
societies where we find higher educational opportunities with regard to educational origin, we find also higher opportunities for marriage to a spouse of different educational
log of association (phi)
566
International Sociology 27(4)
1.3
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
-0.7
-0.8
IS GB SE FR NL FI AT NO EE ES DK BE CY LU IE SI HU DE TR PT PL IL SK IT CH GR UA CZ BG
countries
edu reproduction
edu homogamy
Figure 9. Relative educational reproduction and relative educational homogamy in 29
countries (men – model 2)
level. This indicates in everyday reality that descendants with different educational
attainments to their parents do not keep to their educational status by marriage with a
similar educational level partner.
Figures 9 and 10 show the results from model 2. In both figures on the x-axes, we can
see the analysed countries sorted by (log) levels of association (indicated by φ parameter
for relative educational reproduction and by multiplication of cφ parameters for relative
educational homogamy, cf. equation 2, Figure 6). The distance between relative educational reproduction and relative educational homogamy is estimated as a constant c in all
countries. For men (the log of) this distance is 0.27 (relative educational homogamy is
1.32 times higher then educational reproduction in all countries analysed); for women
(the log of) this distance is 0.22 (relative educational homogamy is 1.24 times higher
then educational reproduction in all countries analysed). For both men and women, educational homogamy indicates the extent of educational opportunities – relative openness
of educational structure – with a systematically higher intensity than educational reproduction. Between these two measures we find the association, but the intensity by which
they indicate the level of educational inequalities is different.
The order of the countries according to strength of association in Figures 9 and 10
show that Bulgaria, Ukraine, the Czech Republic, Slovakia, Italy, Greece and also
Switzerland are countries where there is a relatively high educational reproduction and
educational homogamy compared to other analysed countries. The lowest relative educational reproduction and homogamy are found in Iceland, Sweden, Finland and Norway,
but also in France and Great Britain. Domaňski and Przybysz (2007) came to a similar
conclusion in their analysis of educational homogamy in selected European countries.
Their analysis showed that the strongest association between couples’ education is in the
567
log of association (phi)
Katrňák et al.
1.3
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
-0.7
-0.8
IS GB SE FI NL FR NO BE LU ES AT DK EE SI DE HU PT IL IE PL CY TR GR IT SK CH UA CZ BG
countries
edu reproduction
edu homogamy
Figure 10. Relative educational reproduction and relative educational homogamy in 29
countries (women – model 2)
former socialist countries (Slovakia, Czech Republic, Poland, Ukraine, Hungary and
Estonia), the lowest association is in the Scandinavian countries, France and Luxembourg.
Conclusion and discussion
The main aim of this article was to test two contradictory hypotheses. The first one (H1)
considers the relationship between educational mobility and educational heterogamy as
an indicator of the concept of societal openness. The second hypothesis (H2) proposes a
more complicated relationship and states that educationally heterogamous marriages do
not go hand in hand with the level of educational mobility in societies. We refuted the
second hypothesis. A positive association exists between educational reproduction and
educational homogamy (or between educational mobility and educational heterogamy)
in the European countries studied in the article. We can conclude that a permeable educational system is related to a permeable marriage market, which is confirmed in our
work in both the results of the descriptive analysis, and the results of log-multiplicative
models.
We posed three interrelated questions in the introduction to our article: Is there any
relationship between educational mobility and educational heterogamy? Is this relationship positive or negative? Is one of these indicators systematically stronger than the
other? Our answer is that the relationship between educational mobility and educational
heterogamy exists, and is positive, but educational mobility and educational heterogamy
do not indicate the level of educational inequalities with the same intensity. Relative
educational homogamy is a systematically stronger indicator than relative educational
mobility. On a general level we can distinguish four explanations for this difference.
568
International Sociology 27(4)
First, this it can be due to the difference in the data analysed (see Blau and Duncan,
1967). While educational homogamy refers only to those respondents in the samples who
entered into marriage, educational mobility refers to all the respondents in the samples.
Second, the time factor can be behind this difference. Educational mobility happens
over time, and thus the process can be influenced by other variables such as the openness
of the educational system (Breen, 2004), peer group, or educational/employment aspirations (Hauser, 1972; Hauser et al., 1983; Sewell and Hauser, 1972, 1975), all of which can
weaken the educational relationship between parents and descendants. On the contrary, the
selection of a spouse is, in comparison with educational mobility, accomplished in a relatively brief period. Moreover, our research presumed that education does not change significantly in the course of one’s life, while there might occur a so-called educational ‘pull’
between the spouses, which remains hidden in the data we worked with (respondents stated
the spouses’ education at the time of data collection rather than at the time of entry into
marriage). These facts might distort the relative degree of educational homogamy.
Third, this systematic difference can be linked to intimacy. Educational homogamy
can indicate ‘societal openness’ more strongly because it is a more intimate issue than the
process of educational mobility. A number of factors must be taken into consideration
when we talk about the educational homogamy process: emotional closeness, similarity
of opinions, value orientations, cultural preferences and lifestyles. All these go hand in
hand with educational level (Bourdieu, 1989). As a result, the concept of educational
homogamy can capture more sensitively the inequalities in educational opportunities
than the concept of educational mobility.
Fourth, we should consider the different aspects of the concept of ‘societal openness’
(Ultee and Luijkx, 1990). Educational mobility refers to one aspect of the openness of
the social structure, yet educational homogamy refers to a different aspect of the same
thing. The two concepts are interrelated; although each of them indicates one and the
same thing to a different degree, they point to different aspects of the same phenomenon.
Educational mobility refers to inequalities of educational opportunity; it indicates (interor intragenerational) educational movements. Educational homogamy refers to educational inequalities and indicates how people approach each other with regard to the
educational levels in society.
Generally, we believe that each of these four explanations must be taken into account
when interpreting the different relative measures of educational mobility and educational
homogamy. Nevertheless, the first explanation does not apply to our analysis. We have
dealt with respondents who have experienced educational mobility and were also married. For this reason we turn to the second, third and fourth explanations to interpret why
educational homogamy is relatively higher than relative educational reproduction. But
our data cannot decide which of these explanations is correct. To determine this, a different type of data will be necessary and a different analytical approach (for more see
Blossfeld, 2009).
Since the first generation of social stratification research the question ‘which countries are among those with an open social structure’ has remained an open one (Breen and
Jonsson, 2005). Although today’s social stratification researchers have at their disposal a
number of studies and analyses of social and educational mobility and educational
homogamy, no consensus has been reached on the above question. Our study does not
Katrňák et al.
569
enable us to reject the hypothesis of a positive relationship between educational mobility
and educational heterogamy. Bearing in mind these facts, we suggest that the analysis of
the relative social or educational mobility should be supplemented with an analysis of
relative educational homogamy in further social stratification research. We believe that
the question of openness of social structures in individual countries will be better
addressed using both these measurements.
Funding
This research was supported by Czech Science Foundation (Grant No. 403/08/0542) as well as the
Centre for the Study of Social Stratification and Inequality at Tohoku University under the Global
COE Programme.
Notes
1.
Education is an ordinal variable wherein we can recognize which level of education is greater
and which lower. However we cannot say how far these levels are apart from one another, and
whether they move closer to one another over time or further apart. For example we know
that university education is higher than secondary school, but we do not know whether the
distance between them is less the difference between secondary school and trade school, and
how these distances develop over time.
2. With the first wave of ESS (2002) we use the Integrated File − Edition 6.0, which we supplemented with the national data file from the Czech Republic. The survey involved 22 countries,
and the total number of respondents in this file was 42,359. Data collection was conducted in
2002 and 2003. With the second wave of ESS (2004) we use the Integrated File − Edition 3.0,
which we supplemented with the national data file from Great Britain. In this wave the research
involved 25 countries (research conducted in 2004 and 2005). The number of respondents was
47,537. With the third wave of ESS (2006) we use data file Edition 1.0, containing 30,949
respondents from 17 countries. Samples in all of the countries were selected at random and
are representative of the given countries. The codes of the countries in the data file are as
follows: AT – Austria, BE – Belgium, BG – Bulgaria, CY – Cyprus, CZ – Czech Republic,
DE – Germany, DK – Denmark, EE – Estonia, ES – Spain, FI – Finland, FR – France, GB –
United Kingdom, GR – Greece, HU – Hungary, CH – Switzerland, IE – Ireland, IS – Iceland,
IL – Israel, IT – Italy, LU – Luxembourg, NL – Netherlands, NO – Norway, PL – Poland, PT
– Portugal, SE – Sweden, SI – Slovenia, SK – Slovakia, TR – Turkey, UA – Ukraine.
3. Before merging the tables from different years, we tested the differences in individual table
frequencies using the chi-square test. At a 90% level of statistical significance, the chi-square
test indicated the possibility of merging all the table frequencies from all countries we had
data for more then one year. When we adopted a 95% significance level, the differences
were statistically significant in two tables in Belgium, Spain and Finland, and in one table in
France, Great Britain and Greece. Because in all these cases the value of chi-square test was
just above the level 0.05 we decided to leave the data from these tables in analysis after merging them. The overall size of our analysed data is 68,720 respondents.
4. ISCED 1997 (International Standard Classification of Education) includes these educational
levels: (0) pre-primary education; (1) primary education or first stage of basic education;
(2) lower secondary or second stage of basic education; (3) (upper) secondary education;
(4) post-secondary and non-tertiary education; (5) first stage of tertiary education; (6) second
570
International Sociology 27(4)
stage of tertiary education (for more detail, see UNESCO, 1997; for an evaluation of ISCED
1997 see Schneider, 2007).
5. It is not possible to include the variable of sex in this analysis and estimate only one set of
hierarchical models where the effect of sex on the other variables is presented, because the
basic two-way interaction between respondent’s parents’ education and his or her sex does
not have substantive meaning. Similarly, a two-way interaction between respondent’s parents’ education and his or her spouse’s education does not have substantive meaning either.
All models were estimated using the LEM program (Vermunt, 1997). When analysing the
relationship between educational reproduction and educational homogamy, some cells in the
tables had zero frequencies. We added a small constant 0.1 to all these cells in order to achieve
the most reliable estimate of models (Knoke and Burke, 1980).
References
Berent J (1954) Social mobility and marriage: A study of trends in England and Wales. In: Glass
DV (ed.) Social Mobility in Britain. London: Routledge, pp. 321–338.
Blau PM and Duncan OD (1967) The American Occupational Structure. New York: Wiley.
Blau P and Schwarz JE (1983) Crosscutting Social Circles: Testing a Macrostructural Theory of
Intergroup Relations. New York: Academic Press.
Blossfeld HP (2009) Educational assortative marriage in comparative perspective. Annual Review
of Sociology 35: 513–530.
Blossfeld HP and Timm A (eds) (2003) Who Marries Whom? Educational Systems as Marriage
Markets in Modern Societies. Dordrecht: Kluwer Academic Publishers.
Bourdieu P (1989) Distinction: A Social Critique of the Judgement of Taste. London: Routledge.
Breen R (ed.) (2004) Social Mobility in Europe. Oxford: Oxford University Press.
Breen R and Jonsson JO (2005) Inequality of opportunity in comparative perspective: Recent research
on educational attainment and social mobility. Annual Review of Sociology 31(1): 223–243.
Clogg CC (1982a) Some models for the analysis of association in multiway cross- classifications
having ordered categories. Journal of the American Statistical Association 77(380): 803–815.
Clogg CC (1982b) Using association models in sociological research: Some examples. American
Journal of Sociology 88(1): 114–134.
Domaňski H and Przybysz D (2007) Educational homogamy in 22 European countries. European
Societies 9(4): 495–526.
Erikson R and Goldthorpe JH (1992) The Constant Flux: A Study of Class Mobility in Industrial
Societies. Oxford: Clarendon Press.
Ganzeboom HBG, Luijkx R and Treiman DJ (1989) International class mobility in comparative
perspective. Research in Social Stratification and Mobility 8(1): 1–29.
Ganzeboom HB, Treiman DJ and Ultee WC (1991) Comparative intergenerational stratification
research: Three generations and beyond. Annual Review of Sociology 17(1): 277–302.
Goodman LA (1979) Simple models for the analysis of association in cross-classifications having
ordered categories. Journal of the American Statistical Association 74(367): 537–552.
Goodman LA and Hout M (1998) Statistical methods and graphical displays for analyzing how
the association between two qualitative variables differs among countries, among groups or
over time: A modified regression-type approach. Sociological Methodology 28(1): 175–230.
Goodman LA and Hout M (2001) Statistical methods and graphical displays for analyzing how the
association between two qualitative variables differs among countries, among groups or over
time. Part II: Some explanatory techniques, simple models, and simple examples. Sociological
Methodology 31(1): 189–221.
Katrňák et al.
571
Goody J (1983) The Development of the Family and Marriage in Europe. Cambridge: Cambridge
University Press.
Hall J (1954) A comparison of the degree of social endogamy in England and Wales and the
U.S.A. In: Glass DV (ed.) Social Mobility in Britain. London: Routledge, pp. 338–354.
Hauser RM (1972) Disaggregating a social-psychological model of education attainment. Social
Science Research 1(2): 159–188.
Hauser RM, Tsai SL and Sewell WH (1983) A model of stratification with response error in social
and psychological variables. Sociology of Education 56(1): 20–46.
Hout M (1982) The association between husbands’ and wives’ occupations in two-earner families.
American Journal of Sociology 88(2): 397–409.
Hout M (1983) Mobility Tables. Beverly Hills, CA: Sage.
Jones FL (1987) Marriage patterns and the stratification system: Trends in educational homogamy
since 1930s. Australian and New Zealand Journal of Sociology 23(2): 185–198.
Kalmijn M (1991) Status homogamy in the United States. American Journal of Sociology 97(2):
496–523.
Kalmijn M (1998) Intermarriage and homogamy: Causes, patterns, trends. Annual Review of
Sociology 24(1): 395–421.
Knoke D and Burke PJ (1980) Log-Linear Models. Beverly Hills, CA: Sage.
Lipset SM and Zetterberg HL (1959) Social mobility in industrial societies. In: Lipset SM and
Bendix R (eds) Social Mobility in Industrial Society. Berkeley, CA: University of California
Press, pp. 11–75.
Mare RD (1991) Five decades of educational assortative mating. American Sociological Review
56(1): 15–32.
Powers DA and Xie Y (2000) Statistical Methods for Categorical Data Analysis. San Diego, CA:
Academic Press.
Qian Z (1998) Changes in assortative mating: The impact of age and education, 1970–1990.
Demography 35(3): 279–294.
Qian Z and Preston SH (1993) Changes in American marriage, 1972 to 1987: Availability and
forces of attraction by age and education. American Sociological Review 58(4): 482–496.
Raftery AE (1986) Choosing models for cross-classification. American Sociological Review 51(1):
145–146.
Raftery AE (1995) Bayesian model selection in social research. Sociological Methodology 25(1):
111–163.
Raymo JM and Xie Y (2000) Temporal and regional variation in the strength of educational
homogamy: Comment on Smits, Ultee, and Lammers. American Sociological Review 65(5):
773–781.
Schneider S (2007) Cross-nationally comparable measurement of educational attainment: An
evaluation of ISCED-1997 in the European Social Survey. In: RC 28 Spring Meeting, Brno,
24–27 May.
Schumpeter JA (1951) Economic theory and entrepreneurial history. In: Clemence RV (ed.)
Essays on Economic Topics of Joseph Schumpeter. Port Washington, NY: Kennikat Press.
Sewell WH and Hauser RM (1972) Causes and consequences of higher education: Models of the
status attainment process. American Journal of Agricultural Economics 54(5): 851–861.
Sewell WH and Hauser RM (1975) Education, Occupation and Earnings. New York: Academic
Press.
Sixma H and Ultee W (1984) Marriage patterns and the openness of society: Educational homogamy in the Netherlands in 1959, 1971, and 1977. In: Bakker BFM, Dronkers J and Ganzeboom
HBG (eds) Social Stratification and Mobility in the Netherlands. Amsterdam: SISWO, pp.
91–108.
572
International Sociology 27(4)
Smits J, Ultee W and Lammers J (1998a) Educational homogamy in 65 countries: An explanation
of differences in openness using country-level explanatory variables. American Sociological
Review 63(2): 264–285.
Smits J, Ultee W and Lammers J (1998b) Occupational homogamy in eight countries of the
European Union, 1975–89. Acta Sociologica 42(1): 55–68.
Smits J, Ultee W and Lammers J (2000) More or less educational homogamy? A test of different
versions of modernization theory using cross-temporal evidence for 60 countries. American
Sociological Review 65(5): 781–788.
Treiman DJ and Ganzeboom HBG (2000) The fourth generation of comparative stratification
research. In: Quah MSR and Sales A (eds) The International Handbook of Sociology. London:
Sage, pp. 98–121.
Ultee W and Luijkx R (1990) Educational heterogamy and father-to-son occupational mobility in
23 industrial nations: General societal openness or compensatory strategies of reproduction?
European Sociological Review 6(2): 125–149.
UNESCO (1997) International Standard Classification of Education – ISCED 1997. Available at:
www.unesco.org/education/information/nfsunesco/doc/isced_1997.htm
Uunk WJG, Ganzeboom HBG and Róbert P (1996) Bivariate and multivariate scaled association
models: An application to homogamy of social origin and education in Hungary between 1930
and 1979. Quality and Quantity 30(3): 323–343.
Vermunt JK (1997) LEM: A general program for the analysis of categorical data. Tilburg: Tilburg
University. Available at: www.kub.nl/faculteiten/fsw/organisatie/departementen/mto/software2.html
Wong RSK (1990) Understanding cross-national variation in occupational mobility. American
Sociological Review 55(4): 560–573.
Wong RSK (1992) Vertical and nonvertical effects in class mobility: Cross-national variations.
American Sociological Review 57(3): 396–410.
Xie Y (1992) The log-multiplicative layer effect model for comparing mobility tables. American
Sociological Review 57(3): 380–395.
Yamaguchi K (1987) Models for comparing mobility tables: Toward parsimony and substance.
American Sociological Review 52(4): 482–494.
Biographical notes
Tomáš Katrňák is Assistant Professor of Sociology at the Faculty of Social Studies, Masaryk
University, Brno (Czech Republic). His research interests include social stratification, class analysis, social mobility and categorical data analysis.
Petr Fučík is a researcher at the Institute for Research on Social Reproduction and Integration,
Masaryk University, Brno (Czech Republic). His research focuses on the sociology of family,
marital homogamy and divorce. He is currently writing his PhD thesis on divorce and changes in
reproduction strategies in the Czech Republic.
Ruud Luijkx is Associate Professor of Sociology at Tilburg University. He has contributed to international benchmark studies on social mobility and published further in the field of (educational)
heterogamy, social inequality, career mobility, labour market transitions and log-linear and latent
class analysis. He is currently serving as Data Director for the European Values Study.
Katrňák et al.
573
Résumé
Cet article traite du lien entre homogamie éducationnelle et mobilité éducationnelle dans 29 pays
d›Europe. Une réponse est apportée à trois questions liées entre elles : (1) Y a-t-il un lien entre
mobilité éducationnelle et homogamie éducationnelle ? (2) L›homogamie éducationnelle diverget-elle de la mobilité éducationnelle (lien négatif) ou bien l›hétérogamie éducationnelle renforce-telle la mobilité éducationnelle (lien positif) ? (3) En cas de lien positif, la mobilité éducationnelle
et l›hétérogamie éducationnelle indiquent-elles dans une même mesure le niveau des inégalités en
matière d›éducation ? Pour répondre à ces questions, les auteurs utilisent des données provenant
du European Social Survey sur trois périodes (2002, 2004 et 2006). Les réponses sont données en
termes absolus (pourcentages) ainsi que relatifs (effets log-multiplicatifs). Les résultats montrent
qu’il existe un lien positif entre mobilité éducationnelle et hétérogamie éducationnelle. Dans
l›ensemble des pays analysés, l›hétérogamie éducationnelle relative est, en matière d›éducation, un
indicateur d›inégalités plus prononcé que la mobilité éducationnelle relative. L›écart systématique
entre l›hétérogamie éducationnelle et la mobilité éducationnelle s›explique par un certain nombre
de facteurs que les auteurs analysent dans la dernière partie de l›article.
Mots-clés: European Social Survey, homogamie éducationnelle, homogamie sociale, mobilité
éducationnelle, modèles log-linéaires, modèles log-multiplicatifs
Resumen
Este artículo analiza la relación entre la homogamia educativa y la movilidad educativa en 29
países europeos. Responde a tres cuestiones relacionadas: (1) ¿Hay alguna relación entre movilidad
educativa y homogamia educativa? (2) ¿Existe divergencia entre homogamia educativa y movilidad
educativa (relación negativa) o la movilidad educativa es reforzada por la heterogamia educativa
(relación positiva)? (3) Si existe una relación positiva, ¿muestran la misma desigualdad educativa
la movilidad educativa y la heterogamia educativa? Para responder a estas preguntas, los autores
usan datos de tres oleadas de la Encuesta Social Europea (2002, 2004 y 2006). Las respuestas
se muestran en términos absolutos (porcentajes) y relativos (efectos log-multiplicativos). Los
resultados muestran que existe una relación positiva entre movilidad educativa y heterogamia
educativa. Para todos los países analizados, la heterogamia educativa relativa es un mejor indicador
de las desigualdades educativas que la movilidad educativa relativa. La desviación sistemática de
la heterogamia educativa con respecto a la movilidad educativa es explicada por diversos factores
que los autores discuten en la última parte del artículo.
Palabras clave: emparejamiento homogámico, Encuesta Social Europea, homogamia educativa,
modelos log-lineales, modelos log-multiplicativos, movilidad educativa