Intergenerational contact by european migrants in belgium

Lindsay Theunis1, Christine Schnor1,2, Didier Willaert1 & Jan Van Bavel1, 2
1 Interface
2
Demography, Vrije Universiteit Brussel
Family and Population Studies, University of Leuven
Divorce Conference 2014
Educational Assortative Mating and
Union Stability:
A Prospective Analysis Using Belgian
Census and Register Data
Motivation
• People tend to choose romantic partners who are
educationally similar to themselves
• Educational heterogamy, and especially hypogamy (woman
more educated), is expected to have a negative effect on
union stability
 as a consequence of frustrations and tensions related to
the dissimilarity
• Past research often confirmed this expectation
BUT inconsistencies remained because of variations in
legal, social and economic costs of divorce across time
and between and within countries
2
Motivation
Recent evidence from the US (Schwartz & Han 2014) suggest
changing patterns over time
 the stability of marriages between educational equals has
increased
 the negative association between hypogamy (W>M) and union
stability disappeared among marriages formed after 1990
 Because of the changed gender gap in education?
 A “diffusion of innovation story”?
Deviant union behavior (such as unions with a higher educated
woman) can be seen as an innovation to marriage market
constraints (‘not enough highly educated men on the mating
market’) --> when such marriages become more common, do they
become more accepted and less unstable?
3
Objectives & research questions
1. To examine the effect of relative education on union
stability in Belgium
 How are educational differences between partners
related to union stability, net of individual
educational effects?
2. To test the diffusion hypothesis by including
educational context factors
Does the effect of couples’ relative education on union
stability depend on the proportion of hypogamous
couples (W>M) in the neighborhood?
4
Data
• Belgian CENSUS 2001 – National Register 2006
 Covers the whole Belgian population
 Linkage on the individual level
 Changes in union status are calculated by comparing the dates
and destinations of residential migration of both partners
• Sample selection (N = 472,945 couples)
 Got married between 30/09/1986 and 30/09/2001
 Female age at marriage between 18 and 49
 Both man and woman had Belgian nationality of origin in
2001, were not enrolled in school in 2001 and did not
emigrate or die between 2001 and 2006
5
Absolute & relative education
• Operationalization education man and woman in 2001
 Low = lower secondary education
 Medium = higher secondary education
 High = tertiary education
• Distribution of the sample, in %
 Absolute education
woman
18.7
man
36.3
24.3
0
42.7
37.8
20
35.2
40
low
medium
60
high
missing
80
2.3
2.7
100
6
Absolute & relative education
 Relative education (Man-Woman)
homogamy (M=W)
10.4
0
19.3
10
low-low
hypogamy (M<W)
9.5
0
3.8
10
low-medium
hypergamy (M>W)
30
medium-medium
40
50
60
50
60
50
60
high-high
12.2
20
low-high
30
40
medium-high
6.0 1.6 7.0
0
10
medium-low
missing = 3,75%
20
26.4
20
high-low
30
40
high-medium
7
Belgium
8
Map 1: % homogamous couples, per municipality
9
Map 2: % hypergamous couples, per municipality
10
Map 3: % hypogamous couples, per municipality
11
Map 4: % divorced couples, per municipality
12
Method
• Piecewise constant multilevel hazard model
 Event = marital household dissolution between 2001 and 2006
 Time = marriage duration
 Left truncated data, entry in 2001
 Control for a municipality-level random residual
• Control covariates
 Age = female age at marriage, relative age
 Children = parity and age youngest child (TV)
 Occupational status = male occupational-related income, relative
occupational-related income (2001)
 Marital residence = ownership, comfort level (2001)
 Place of residence = region, degree of urbanization (2001)
13
Results
Hazard ratios of relative education*
(95% confidence interval)
1.4
1.3
1.2
1.1
1.0
0.9
0.8
homogamy (W=M)
hypergamy (W<M)
hypogamy (W>M)
14
*Controlled for female education and controls
= homogamy (W=M)
= hypergamy (W<M)
= hypogamy (W>M)
Results
Hazard ratios of female education*male education
(95% confidence interval)
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
low
medium highly
low
medium highly
low
medium highly
educated educated educated educated educated educated educated educated educated
man
man
man
man
man
man
man
man
man
low educated woman
medium educated woman
highly educated woman
15
Results
Hazard ratios for % hypogamous couples in
municipality* (95% confidence interval)
1.1
1.0
0.9
0.8
<24%
24-26%
>26%
16
*Categorization of % hypogamous couples is based on tertiles
= homogamy (W=M)
= hypergamy (W<M)
= hypogamy (W>M)
Results
Hazard ratios of male education*% hypogamous
couples for couples with a low educated woman
(95% confidence interval)
1.3
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
<24% 24-26% >26%
<24% 24-26% >26%
<24% 24-26% >26%
low educated man
medium educated man
highly educated man
17
= homogamy (W=M)
= hypergamy (W<M)
= hypogamy (W>M)
Results
Hazard ratios of male education*% hypogamous
couples for couples with a medium educated woman
(95% confidence interval)
1.3
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
<24% 24-26% >26%
<24% 24-26% >26%
<24% 24-26% >26%
low educated man
medium educated man
highly educated man
18
= homogamy (W=M)
= hypergamy (W<M)
= hypogamy (W>M)
Results
Hazard ratios of male education*% hypogamous
couples for couples with a highly educated woman
(95% confidence interval)
1.3
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
<24% 24-26% >26%
<24% 24-26% >26%
<24% 24-26% >26%
low educated man
medium educated man
highly educated man
19
Preliminary conclusion
• Simple homogamy-heterogamy distinctions cover more complex results
• The highest divorce risks were found among couples in which the
woman is low educated and the man is medium educated, the lowest
divorce risks were found among couples in which both man and woman
are highly educated
• Couples in which the woman is medium educated have a higher divorce
risk if the man is highly educated (hypergamy) and a lower divorce risk
if the man is low educated (hypogamy) than with a man who has a
similar educational level (homogamy)
• Living in a municipality with ‘more than average’ hypogamous couples
lowers the risk of divorce, especially of hypogamously married couples
20
Future prospects
• We are in the process of adding unmarried cohabitations: We will
determine dates of union formation (start of cohabitation) of
couples in 2001 by their residential movements between 1991
and 2001
• We will create place-varying covariates: If a couple moves during
the time at risk, we can adapt the values on the context variable
(% hypogamous couples) and the place of residence variables
(region and degree of urbanization) to those measured for their
destination
• We will consider other context covariates
• We will estimate similar analyses with the linked data of CENSUS
1991 and National Register 1996
21
Thank you for your
attention!
Suggestions and remarks are very welcome
Contact: Lindsay.Theunis@vub.ac.be