Transcript WHILE EDUCATIONAL INEQUALITIES DECREASED, DID

MORE UPWARD AND ALSO
MORE DOWNWARD MOBILITY?
COMPULSORY SCHOOLING IN THE
NETHERLANDS DURING THE 20TH CENTURY
Wout Ultee - Radboud Universiteit Nijmegen
Presentation in the Department of Sociology,
Tel Aviv University February 24, 2005
1
In the 1990s it was found that only in
Sweden and the Netherlands inequalities
in education had decreased
Some doubted the quality of the Dutch
data leading to this conclusion
New findings indicate that with sufficient
statistical power, a trend towards smaller
inequalities in education is present in
other (post)industrial countries too
2
Previous Research
on the Netherlands
Loglinear models with
educational distributions for fathers and
children (sons and daughters) fitted with
trend constraints on
relative chances
Educational inequality decreased according to:
uniform association and
diagonal densities
3
Present research: data

Stacked file from 31 surveys
 21,899 men/20,903 women; age over 25
 Fifteen 5-year birth cohorts (1900-1974)
 Father’s education 4 categories (low –
high)
 Daughter’s/Son’s education 4 categories
(low – high)
4
Are questions about diagonal densities and
uniform associations, or even questions
about odds ratio’s, pertinent research
questions given the theories at issue?
A lower odds ratio may result
from a higher odds to climb and
from a higher odds to fall.
What are current theories about?
5
Present research:
hypotheses

Four hypotheses from the field of
stratification
 One from each generation of
stratification studies
6
Zero-generation
non-hypothesis
There is a linear downward trend in
uniform association and diagonal density
This hypothesis at best is about some
unspecified macro force
7
First-generation
hypothesis
A rising standard of living makes for lower
uniform association and lower diagonal
density
This hypothesis is about the effect of a
country characteristic, that is, about a
macro force or factor
8
Second-generation
hypothesis
State intervention makes for lower uniform
association and diagonal density
This hypothesis is about effects of states
9
Third-generation
hypothesis
State intervention does not alter the density in
the diagonal cell for the highest level of
education
This hypothesis is about the fate of elites in the
hands of states
10
Embedding the four
hypotheses in general
sociology:
How to derive macro hypotheses from
micro hypotheses?
11
Zero generation hypothesis derived
(post)industrialization requires more and
more mobility, and this functional
prerequisite is met
Or: the increasing demand of employers for
highly educated employees makes it
worthwhile for employees to educate
themselves and they therefore do so
12
First-generation
hypothesis derived
Persons differ in financial resources; and the
relation between a country’s average
income and its educational inequalities is an
aggregate result of a general rise in income
13
Second-generation
hypothesis derived
Apart from individual actors there is an
all powerful corporate actor: the state
(Coleman about corporate actors)
14
Two possible derivations
from the assumption of
an intervening state:
The implicit old one and
The explicit new one
15
The implicit old one:
States provide stipends etc.
for children of low
backgrounds and with
outstanding intelligence to
attend university
16
My explicit new one:
States gradually raise the age
until which schooling is
compulsory for everyone:
first only primary school,
now also a long period in
secondary school
17
Third-generation hypothesis
derived
Highly educated parents deploy strategies
compensating for state policies raising the
age at which compulsory schooling ends
States are not all-powerful:
(Elias about a spiraling process as a
competitive outcome)
(Bourdieu about compensatory strategies)
(Hout & Raftery about maximally maintained
inequality)
18
All in all, the odds to move up for
persons at the lower end of the social
scale increase,
While at the time these odds increase,
the odds for persons at the upper end of
the social scale do not fall, but increase
too.
19
Theoretical lesson
Deriving macro hypotheses
from micro hypotheses,
makes clear that worthwhile
alternatives are around
20
The age until which schooling was
compulsory in the Netherlands
changed as follows
Since 1900
until the age of 12 years
After 1948
until the age of 14 years
After 1969
until the age of 16 years
After the mid 1970s
until the age of 18 years
21
Graph 1 Changes in Cohort Specific Association-Men
Models for linear trend, GDP and Compulsory Schooling
1
2
3
4
intergenerational educational mobility; the Netherlands 1900-1974
1900
1910
1920
1930
1940
birth cohort
cohort specific association
GDP per Capita
1950
1960
1970
linear trend
Age Compulsory Schooling
Model includes one general diagonal effect for each cohort
Homogeneous equal scalings for education
22
Graph 2 Changes in Cohort Specific Diagonal Effects-Men
Models for linear trend, GDP and Compulsory Schooling
0
.2
.4
.6
.8
1
intergenerational educational mobility; the Netherlands 1900-1974
1900
1910
1920
1930
1940
birth cohort
cohort specific association
GDP per Capita
1950
1960
1970
linear trend
Age Compulsory Schooling
Model includes one uniform associat ion effect for each cohort
Homogeneous equal scalings for education
23
Graph 3 Changes in Cohort Specific Diagonals 1 2 3-Men
Models for linear trend, GDP and Compulsory Schooling
0
.2
.4
.6
.8
1
intergenerational educational mobility; the Netherlands 1900-1974
1900
1910
1920
1930
1940
birth cohort
cohort specific association
GDP per Capita
1950
1960
1970
linear trend
Age Compulsory Schooling
Model includes one uniform associat ion effect for each cohort
Homogeneous equal scalings for education
24
Graph 4 Changes in Cohort Specific Diagonal 4-Men
Models for linear trend, GDP and Compulsory Schooling
0
.2
.4
.6
.8
1
intergenerational educational mobility; the Netherlands 1900-1974
1900
1910
1920
1930
1940
birth cohort
cohort specific association
GDP per Capita
1950
1960
1970
linear trend
Age Compulsory Schooling
Model includes one uniform associat ion effect for each cohort
Homogeneous equal scalings for education
25
Graph 5 Changes in Cohort Specific Association-Women
Models for linear trend, GDP and Compulsory Schooling
1
2
3
4
intergenerational educational mobility; the Netherlands 1900-1974
1900
1910
1920
1930
1940
birth cohort
cohort specific association
GDP per Capita
1950
1960
1970
linear trend
Age Compulsory Schooling
Model includes one general diagonal effect for each cohort
Homogeneous equal scalings for education
26
Graph 6 Changes in Cohort Specific Diagonal Effects-Women
Models for linear trend, GDP and Compulsory Schooling
0
.2
.4
.6
.8
1
intergenerational educational mobility; the Netherlands 1900-1974
1900
1910
1920
1930
1940
birth cohort
cohort specific association
GDP per Capita
1950
1960
1970
linear trend
Age Compulsory Schooling
Model includes one uniform associat ion effect for each cohort
Homogeneous equal scalings for education
27
Graph 7 Changes in Cohort Specific Diagonals 1 2 3-Women
Models for linear trend, GDP and Compulsory Schooling
0
.2
.4
.6
.8
1
intergenerational educational mobility; the Netherlands 1900-1974
1900
1910
1920
1930
1940
birth cohort
cohort specific association
GDP per Capita
1950
1960
1970
linear trend
Age Compulsory Schooling
Model includes one uniform associat ion effect for each cohort
Homogeneous equal scalings for education
28
Graph 8 Changes in Cohort Specific Diagonal 4-Women
Models for linear trend, GDP and Compulsory Schooling
0
.2
.4
.6
.8
1
intergenerational educational mobility; the Netherlands 1900-1974
1900
1910
1920
1930
1940
birth cohort
cohort specific association
GDP per Capita
1950
1960
1970
linear trend
Age Compulsory Schooling
Model includes one uniform associat ion effect for each cohort
Homogeneous equal scalings for education
29
Stipends etc. in the Netherlands
Not only for keen working class boys
Also for upper class girls,
They are for everyone
30
Testing the third-generation
hypothesis more severely

How do macro variables influence uniform
association and diagonal densities?
 Few states ever sought to increase the odds
for moving down.
 Uniform association and diagonal densities
derived from odds ratios
 Odds ratios made up from an odds for
moving up and an odds for moving down
31
Hypotheses about odds

Raising the age at which compulsory
schooling ends, increases the odds for
moving up.
 But what does it do to the odds for
moving down?
32
Observed Odds: Upward Move from Level 1 to Level 4
intergenerational educational mobility
8
10
Men in the Netherlands 1900-1974
upward trend
0
2
4
6

1900
1910
1920
1930
1940
birth cohort
1950
1960
1970
dots proportional to sample size
33
Observed Odds: Downward Move from Level 4 to Level 1
intergenerational educational mobility
.8
1
Men in the Netherlands 1900-1974
no trend or a downward trend
0
.2
.4
.6

1900
1910
1920
1930
1940
birth cohort
1950
1960
1970
dots proportional to sample size
34
Dutch
questions,
Dutch theories,
Dutch data collection
35
Dutch questions

Two large questions about societies;
– Various inequalities
– Several aspects of cohesion
36
Dutch questions: serach for
wrong ones, enrich existing
ones

Questions should not be about odds ratio’s
 Questions should be about separate odds
37
Dutch theories
Not factor sociology
Versus
Actor sociology
38
Dutch theories
Not factor sociology
Versus
Actor sociology:
There is a third
alternative
39
Dutch theories
* Factor or
macro forces
sociology
- Industrialization
- Technology
- Ideology
- Average income
- General rise in
education
* Actor or
individual
sociology
– Individuals
– One goal
– One shot
decisions
– (all powerful)
state
* ActorS sociology
– Egos and their
networks
– Several goals
– Sequential
decision processes
– (intermediate)
collective actors
40
Dutch data collection

Family Survey Dutch Population
– 1993
– 1998
– 2000
– 2003
41
Dutch data collection

Harry Ganzeboom
 Nan Dirk de Graaf
 Paul de Graaf
 Gerbert Kraaykamp
 Marcel Lubbers
 Ariana Need
 Wout Ultee
42
Dutch data collection

Multi-Moment data
 Multi-Actor data
 Multi-Context data
43
Dutch data collection

Multi-Moment data
 Multi-Actor data
 Multi-Context data
3 MAC data
44
Dutch data collection

Series moments primary respondents
– Job history
– Educational history
– Income history
– Housing history
– Religious history
– Smoking history
– Sport club history
45
Dutch data collection

Point moments primary respondents
– Parental, material, and cultural resources
– Leisure activities at age 18
– Weight at age 25 and now
– Relation with boss at every job spell
– Relation with co-workers at every job spell
46
Dutch data collection

Past and present attitudes of the primary
respondent
– Abortion
– Foreigners
– Homosexuals
47
Dutch data collection

Actors
– Primary respondent
– Partner
– Parent
– Child
– Sibling
48
Dutch data collection

Actors
– Partner same questionnaire as respondents to
answer questions about the extent to which
partner similarity is more than a byproduct of
educational homogamy, and whether spouse
effects are present
49
Dutch data collection

Actors
– Parental questionnaire as a check upon
retrospective answers of the primary respondent
– Additional questions about upbringing of
primary respondent
50
Dutch data collection

Actors
– Child
Questionnaire attuned to the issue of first
generation upon third generation effects
51
Dutch data collection

Actors
– Sibling
Questionnaire in main aspects the same as
questionnaire for primary respondents, so that
family effects can be estimated and the
statistical power of models can be increased
52
Dutch data collection

Contexts
– Municipalities
– Neighborhoods
– Schools
– Employers
– Economic sectors
– Sport clubs
53
Dutch data results




Wim Bernasco
Koen van Eijck
Tamar Fischer
 Maurice Gesthuizen
 Christiaan Monden
 Hester Moerbeek
 Inge Sieben
 Wilfred Uunk
Herman van der Werfhorst
 Maarten Wolbers
54