ArticlePDF Available

The Intergenerational Transmission of Divorce: A Fifteen-Country Study with the Fertility and Family Survey

  • University of Leipzig and ETH Zurich

Abstract and Figures

Studies mainly from the United States provide evidence that children of divorced parents face a higher risk of divorce in their own marriages. We estimate and analyze the effects of divorce transmission using comparative individual data from the United Nations for 13 eastern and western European countries as well as for Canada and the United States. We find substantial and highly statistically significant transmission effects in all samples. This shows that the intergenerational transmission of divorce is a widespread phenomenon observed without a single exception in our data covering a large number of countries with differing historical, institutional, and cultural contexts.
Content may be subject to copyright.
© Koninklijke Brill NV, Leiden, 2013 DOI: ./-
Comparative Sociology 12 (2013) 1–14
The Intergenerational Transmission of Divorce:
A Fifteen-Country Study with the
Fertility and Family Survey*
Andreas Diekmannand Kurt Schmidheiny
 Swiss Federal Institute of Technology (ETH)
University of Basel
Studies mainly from the United States provide evidence that children of divorced
parents face a higher risk of divorce in their own marriages. We estimate and ana-
lyze the efects of divorce transmission using comparative individual data from the
United Nations for 13 eastern and western European countries as well as for Can-
ada and the United States. We nd substantial and highly statistically signicant
transmission efects in all samples. This shows that the intergenerational transmis-
sion of divorce is a widespread phenomenon observed without a single exception
in our data covering a large number of countries with difering historical, institu-
tional, and cultural contexts.
*The authors wish to thank the Advisory Group of the FFS program for its permission,
granted under identication number 81, to use the FFS data on which this study is based. We
are also very much indebted to Elisabeth Coutts and Hartmut Esser for valuable comments.
We started the analysis of the data of the Fertility and Family Survey (FFS) in 2001 when
we received approval by the U.N. Population Activities Unit to use the Fertility and Family
Survey data. Our ndings concerning a) the transmission efects in various countries and b)
a strong and signicant correlation of divorce rates and transmission efects were presented
on conferences (e.g. in Florence 2002) and in a working paper (2004). We are grateful to
Hartmut Esser who objected a causal interpretation of b) arguing that this relation might be
an artifact. Theoretical reasoning and robustness tests by the authors (see below) supported
this view. Dronkers and Härkönen (2008) replicated our ndings on country efects. Also,
they reported the seemingly divorce rate efect on the transmission efect using a multi-
level model. We believe this being an artefact having no causal meaning. For a critique of
Dronken and Härkönen (2008) see our comment Diekmann and Schmidheiny (2008).
COSO 12.2_1-14-Diekmann and Schmidheiny.indd 1COSO 12.2_1-14-Diekmann and Schmidheiny.indd 1 2/15/2013 5:50:31 PM2/15/2013 5:50:31 PM
2 A. Diekmann, K. Schmidheiny / Comparative Sociology 12 (2013) 1–14
divorce, divorce risk, intergenerational transmission, consequences of divorce,
child well-being
1 Introduction
Studies from the United States provide evidence that children of divorced
parents face a higher risk of divorce in their own marriages, although there
is controversy about the reasons for this efect (Amato 1996; Bumpass and
Sweet 1972; Glenn and Kramer 1987; Greenberg and Nay 1982; Keith and
Finlay 1988; McLanahan and Bumpass 1988; Mott and Moore 1979; Mueller
and Pope 1977; Pope and Mueller 1976; Teachman 1982; Wolnger 1999).
The efect has also been found in the Netherlands (Traag, Dronkers and
Vallet 2000), West Gemany (Diekmann and Engelhardt 1999), Great Britain
(Kiernan and Cherlin 1999) and France (Traag, Dronkers and Vallet 2000),
although no efect was found in East Germany (Diefenbach 1997). The
intergenerational transmission of divorce has contributed to the upward
trend in divorce rates, and a better understanding of its efect is crucial
not only to the analysis and prediction of divorce trends but also to insight
into the development of children after their parents’ divorce. Up to now it
has not been established whether the transmission efect and its proposed
explanations are generalizable to other Western countries and beyond. In
this study, we estimate and analyze the efects of divorce transmission for
13 eastern and western European countries as well as for Canada and the
United States. Our study is the rst to conduct a systematic investigation
covering a large number of countries with difering historical, institutional,
and cultural contexts, thus allowing us to answer the question of whether
divorce transmission is a stable and robust phenomenon observable across
diferent cultures. Also, the analysis of patterns in divorce-related charac-
teristics among children from divorced families sheds light on the explana-
tion of the transmission efect.
It is by no means natural to assume that transmission efects are a univer-
sal phenomenon. One can hypothesize that these efects depend on many
factors, some of them related to institutional setting (Amato and Keith
1991a,b). Custody regulations, nancial support by the non-resident parent,
and other aspects of divorce laws may have an impact on how well children
are able to cope with their parents’ divorce. Of course, these regulations
COSO 12.2_1-14-Diekmann and Schmidheiny.indd 2COSO 12.2_1-14-Diekmann and Schmidheiny.indd 2 2/15/2013 5:50:32 PM2/15/2013 5:50:32 PM
A. Diekmann, K. Schmidheiny / Comparative Sociology 12 (2013) 1–14 3
vary over time and across countries. The nations in our sample also difer
in terms of cultural patterns, religious beliefs and the degree to which a
divorce is considered a “normal” event in the life course.
In the current study, we therefore examine the divorce transmission
efect in countries that vary in terms of the above factors. We use the ret-
rospective data on family histories from the Fertility and Family Survey
to estimate transmission efects for fteen countries, including various
Western and Eastern European countries, Canada, and the United States.
We also control for additional independent variables that may well
afect the survey’s respondent’s risk of divorce. The set of control vari-
ables includes well-known divorce risk factors such as age at the start of
a union that led to marriage, birth of a child, the wife’s educational level
premarital cohabitation and membership in certain marriage cohorts
(e.g. White 1990). Previous studies have shown that divorce risks decrease
with the age at which the union that leads to marriage is commenced,
while childless couples and spouses who lived together before marriage
exhibit higher divorce risks than couples who have children and who did
not share a household before marriage. This set of control variables may
also mediate the transmission efect. Empirical studies show that if par-
ents divorce, their ofspring complete less education, marry earlier, have
a greater tendency to cohabit before marriage and may invest less in a
partnership than children from non-divorced families (e.g. Keith and Fin-
lay 1988, Amato 1996, Diekmann and Engelhardt 1999). A further strength
of the present study is that our data allow us to investigate whether the
means of possible mediating variables difer systematically across coun-
tries. Finally, we have also accounted for marriage cohort membership
because divorce risk in general has increased in Western countries in
recent decades.
Our estimation employs the techniques known as survival analysis in
statistics, duration modelling in econometrics, or event-history analysis
in sociology. First, we investigate the presence or absence of transmission
efects in the countries included in our study. Second, we examine whether
the ndings are explainable by the other divorce-related covariates men-
tioned above. Finally, we explore whether there are systematic diferences
in the means of divorce risk factors for respondents with divorced and non-
divorced parents, diferences which could explain at least part of the social
inheritance of divorce.
COSO 12.2_1-14-Diekmann and Schmidheiny.indd 3COSO 12.2_1-14-Diekmann and Schmidheiny.indd 3 2/15/2013 5:50:32 PM2/15/2013 5:50:32 PM
4 A. Diekmann, K. Schmidheiny / Comparative Sociology 12 (2013) 1–14
2 Data
The study is based on data from the Fertility and Family Survey. The FFS
comprises surveys in 21 countries, but the necessary information on either
the duration of the respondent’s marriage or whether the respondent’s par-
ents divorced is lacking for ve of those countries. Thus, with West and East
Germany analyzed separately, our estimates are based on 16 data sets col-
lected in the early 1990s in 13 European countries, Canada, and the United
States. We conne our analysis to female respondents who were married
or have previously married. With these restrictions, net sample sizes vary
from 1,279 (Czech Republic) to 6,844 (U.S.).
Table A.1 in the Appendix displays the variables used and their means.
Tables A.2 and A.3 show the means of the variables separately for respon-
dents whose parents did not divorce and those whose parents divorced.
The variable of main interest is duration of rst marriage in months. We
consider a marriage terminated when it ends in divorce or permanent
separation. For these purposes, we consider the termination date to be
the date of dissolution of a common household.
The main explanatory variable is the parents’ relationship during the
respondent’s childhood. The dummy variable parents’ divorce is set to 1 if
the respondent’s (natural or adoptive) parents divorced or separated after
her birth. In addition to the divorce/separation of the respondent’s parents,
the analysis includes the family structure of the home of origin, specically
information on whether the respondent grew up with both parents, one
parent or without either parent.
Further independent variables include the respondent’s marriage
cohort, age at start of the union that led to her rst marriage, birth of a
child, her educational level and her cohabitation history. We use ve-year
marriage cohorts from 1970 to 1990. The age at start of union is the age of the
Samples were drawn from the population within certain age limits. The Belgian sample
covers only Flanders and the region of Brussels. For more information on the FFS and its use
in comparative research, see Festy and Prioux (2002).
The FFS Standard Recode File does not distinguish between legal divorce and separation.
This denition seems reasonable as the time between the end of co-residence and the
date of the legal divorce varies substantially across the diferent jurisdictions. Furthermore,
the date of legal divorce is not reported in most FFS data sets. See Festy and Prioux (2002,
p. 32) for a discussion of the comparability of FFS partnership data across countries.
The information on composition of the home of origin varies across countries.
COSO 12.2_1-14-Diekmann and Schmidheiny.indd 4COSO 12.2_1-14-Diekmann and Schmidheiny.indd 4 2/15/2013 5:50:32 PM2/15/2013 5:50:32 PM
A. Diekmann, K. Schmidheiny / Comparative Sociology 12 (2013) 1–14 5
respondent at the time she began living with her rst marriage partner. The
birth of the respondent’s rst child is included as a time-dependent covari-
ate. The respondent’s educational level is the one attained by the date of the
interview, and is measured in accordance with the international standard
classication of education (ISCED). This scale covers seven educational
levels from pre-primary (0) to the second stage of tertiary (6). We distin-
guished among three levels: ‘lower’ (values 0, 1 or 2 of the ISCED classica-
tion), ‘medium’ (ISCED 3 or 4), and ‘higher’ (ISCED 5 or 6). Cohabitation
denotes whether the respondent had already shared a household with
her rst spouse before they married (see Appendix A1 for the means of
the covariates).
3 Methods
We use a proportional hazard rate model to estimate the efects of a respon-
dent’s parents’ divorce and sociodemographic covariates on the respon-
dent’s own divorce risk.
It is well known that divorce risk increases during the initial years of
marriage and decreases thereafter (see Figure 1). Because of this non-mono-
tonic duration dependence, we model the hazard rate of divorce risk as
r(t) = atet / λ (1)
with marriage duration t, parameter λ measuring time in months elapsed
until maximal risk and a = exp(β0 + x1β1 + ... + xkβk + ... xKβk); x1 is a dichot-
omous variable indicating whether parents remained married (x1=0) or
were divorced (x1=1), x2,..., xm are further covariates, and β1, β2,..., βm
are empirically estimated parameters. βk · 100% is approximately and
[exp(k β)1] · 100% is exactly equal to the percentage change in the respon-
dent’s divorce risk r(t) when the covariate xk increases by one unit. We use
Unfortunately, educational attainment at marriage is either not reported or very poorly
reported for most countries. See Festy and Prioux (2002, p. 32) for a discussion of the limited
comparability of education variables across countries in the FFS.
The other levels of the ISCED scale are: (1) primary education or rst stage of basic educa-
tion, (2) lower secondary or second stage of basic education, (3) (upper) secondary educa-
tion, (4) post-secondary non-tertiary education, and (5) rst stage of tertiary education. See
UNESCO (1997) for more details.
COSO 12.2_1-14-Diekmann and Schmidheiny.indd 5COSO 12.2_1-14-Diekmann and Schmidheiny.indd 5 2/15/2013 5:50:32 PM2/15/2013 5:50:32 PM
6 A. Diekmann, K. Schmidheiny / Comparative Sociology 12 (2013) 1–14
maximum likelihood to estimate the β parameters of covariate efects and
the parameter λ. Apart from the birth of the rst child, all independent
variables are treated as time constant. We estimate the parameters of the
time-dependent covariate in the likelihood function using the method of
episode splitting (Blossfeld and Rohwer 1995).
The complete length of the episode can be observed only in marriages
that ended in divorce before the interview took place. Marriages still in
efect at the time of the interview or those ended by the death of a spouse
are treated as censored data. Both the complete episodes and the censored
ones were used to estimate the β and λ-parameters. In the presence of
censored data, the maximum likelihood method provides consistent and
asymptotically normally distributed estimates of the parameters. Partial
likelihood estimation (Cox 1972) of β1, β2,..., βm which does not require
specication of the baseline hazard rate function r(t) leads to practically
identical estimates as the ones reported.
Episode splitting is a method for decomposing an episode like marriage duration into
subintervals. Covariates remain constant within subintervals, and the likelihood function
can therefore be rewritten as a product of the subinterval-specic likelihoods. For technical
details see, for example, Blossfeld and Rohwer (1995).
Figure 1
The sickle model of the divorce risk function. Hazard rate curves for difer-
ent values of the parameters λ and a.
Divorce Risk
0 60 120 180 240
Time in Months
 = 180, a = 5.10-5
 = 120, a = 8.10-5
 = 120, a = 5.10-5
 = 120, a = 2.10-5
 = 60, a = 5.10-5
300 360 420
COSO 12.2_1-14-Diekmann and Schmidheiny.indd 6COSO 12.2_1-14-Diekmann and Schmidheiny.indd 6 2/15/2013 5:50:32 PM2/15/2013 5:50:32 PM
A. Diekmann, K. Schmidheiny / Comparative Sociology 12 (2013) 1–14 7
4 Results
We estimate three diferent models. The estimated transmission efects
from those three models are summarized in Table 1 and the respective per-
centage efects visualized in Figure 2. Model 1 controls only for marriage
cohorts, Model 2 controls for these cohorts and home of origin, and Model 3
includes the additional covariates of date of the rst child’s birth, cohabita-
tion, age at start of union, and educational level (see Figure 2 and Table 1, for
more detailed information see Tables A.4 to A.6 in the Appendix).
Transmission efect, [exp()‒1]
Model 1: controlled for cohorts
Model 2: additionally controlled for parental home
Model 3: additionally controlled for education, age at start of union, cohabitation, children
Figure 2
The intergenerational transmission efect of divorce. Summary of estimates
of β1 from country specic maximum likelihood estimations of hazard rate
models with diferent sets of control variables. Plotted is the percent efect
exp(β1)–1. Reading the transmission efect (e.g., Austria): the percent efect
of 101% means that children whose parents divorced have a 101% higher
risk of divorce than children whose parents did not divorce. 95% con-
dence interval based on (non-linear) transformation of condence bounds
of the estimated β1. Full estimation results are provided in Tables A.4 to A.6
in the Appendix.
COSO 12.2_1-14-Diekmann and Schmidheiny.indd 7COSO 12.2_1-14-Diekmann and Schmidheiny.indd 7 2/15/2013 5:50:32 PM2/15/2013 5:50:32 PM
8 A. Diekmann, K. Schmidheiny / Comparative Sociology 12 (2013) 1–14
Austria Belgium Canada Czech Estonia E-Germ W-Germ Hungary Italy Latvia Lithu. Slovenia Spain Sweden Switz. USA
Intergenerational transmission effect in model 1: controlled for cohorts
Parents divorced 0.70
(0.11) (0.17) (0.09) (0.15) (0.12) (0.13) (0.17) (0.12) (0.29) (0.09) (0.13) (0.24) (0.31) (0.14) (0.12) (0.05)
Intergenerational transmission effect model 2: additionally controlled for parental home
Parents divorced 0.62
(0.12) (0.21) (0.09) (0.18) (0.13) (0.15) (0.19) (0.14) (0.32) (0.11) (0.16) (0.27) (0.33) (0.20) (0.13) (0.05)
Intergenerational transmission effect model 3: additionally controlled for education, age at start union, cohabitation, children
Parents divorced 0.55
0.28 0.32
0.59 0.42
(0.12) (0.21) (0.1) (0.18) (0.13) (0.15) (0.19) (0.14) (0.32) (0.11) (0.16) (0.27) (0.33) (0.2) (0.13) (0.05)
All marriages in sample
Number 3217 2373 2429 1268 1237 1733 1289 2756 3163 2041 2145 1958 2674 1839 3059 6518
Of which divorced 18% 11% 29% 21% 27% 20% 18% 17% 5% 29% 15% 8% 5% 19% 15% 35%
Marriages with non-divorced parents
Number 2868 2192 2043 1058 955 1436 1141 2317 3070 1579 1779 1822 2564 1628 2705 4915
In % of all marriages 89% 92% 84% 83% 77% 83% 89% 84% 97% 77% 83% 93% 96% 89% 88% 75%
Of which divorced 17% 10% 27% 19% 25% 18% 16% 16% 5% 28% 14% 7% 5% 18% 15% 33%
Marriages with divorced parents
Number 349 181 386 210 282 297 148 439 93 462 366 136 110 211 354 1603
In % of all marriages 11% 8% 16% 17% 23% 17% 11% 16% 3% 23% 17% 7% 4% 11% 12% 25%
Of which divorced 28% 22% 41% 30% 34% 28% 31% 21% 14% 35% 21% 15% 10% 29% 23% 42%
Notes: Reported is the parameter β
of the maximum likelihood-estimations of the hazard rate r(t) = a t exp(-t/λ) where a = exp(β
+ x
+ x
+ … + x
) and x
indicates whether
the parents have been divorced. Standard errors in parentheses. Parameters with (***,**,*) are significantly different from 0 at p < .001 resp. p < .01, p < .05. The complete maximum
likelihood-estimations of the sickle models for each country are reported in Tables A.4 to A.6 in the Appendix.
Table 1
Survival analysis of divorce risk, summary of estimations by countries
COSO 12.2_1-14-Diekmann and Schmidheiny.indd 8COSO 12.2_1-14-Diekmann and Schmidheiny.indd 8 2/15/2013 5:50:32 PM2/15/2013 5:50:32 PM
A. Diekmann, K. Schmidheiny / Comparative Sociology 12 (2013) 1–14 9
Our rst nding is the universal existence of the transmission efect. Model
1 in Figure 2 shows the percentage transmission efect, i.e. exp(β1)-1, for chil-
dren from divorced families versus respondents from non-divorced families
for all samples. The efects range from 50% for Hungary and Latvia to 220%
for Italy with an average efect of 103%. This means that children whose
parents divorced face – on average – a 103% higher risk of getting divorced
than children whose parents did not divorce. The efects for all 16 data sets
are substantially larger than 0 and highly signicant (Spain at p < 0.05, Slo-
venia at p < 0.01; all other countries at p < 0.001). In contrast to a previous
analysis (Diefenbach 1997), we also nd a highly signicant transmission
efect for East Germany. Our analysis clearly shows that the intergenera-
tional transmission of divorce is a widespread phenomenon observed with-
out a single exception in formerly communist Eastern Europe, Southern
(Catholic) Europe, Western Europe and North America.
Children of divorced parents usually grow up with one parent only. In
the next step we therefore disentangle the efect of the parent’s divorce
from that of the parental home. Model 2 controls for whether the person
grew up with one parent only. The two efects can be separately identied
as some children lose a parent for reasons other than divorce. Our results
show that the transmission efect is only moderately changed in all coun-
tries but Italy, where it is substantially increased. The transmission efect
remains highly signicant in all countries.
In model 3 we include educational level, age at start of union, cohabita-
tion and the birth of children as additional control variables. Transmission
efects are again only moderately reduced compared to the unconditional
efects in model 1, except in Italy (where few of the respondents’ parents
were divorced), and remain signicant at p < 0.05 in all countries but Spain
(p < 0.1) and the Czech Republic. This reduction is a result of controlling for
cohabitation and age at start of the union, since in all countries the chil-
dren of divorce cohabit more and begin unions earlier.
The pattern of union formation and cohabitation sheds light on the
explanation of the transmission efect. In all countries but one, the esti-
mated coecient indicates a higher risk of divorce for couples that lived
together before marriage (see Table A.6). In addition, the well-known nega-
tive efect on divorce of age at union formation shows up in all countries. At
the same time, the children of divorce cohabit more frequently than respon-
dents from non-divorced families and enter into unions at younger age in
COSO 12.2_1-14-Diekmann and Schmidheiny.indd 9COSO 12.2_1-14-Diekmann and Schmidheiny.indd 9 2/15/2013 5:50:32 PM2/15/2013 5:50:32 PM
10 A. Diekmann, K. Schmidheiny / Comparative Sociology 12 (2013) 1–14
all countries (compare Tables A.2 and A.3). Although cohabitation does
likely not exert a positive causal efect on divorce, the covariate is corre-
lated with the divorce risk. Previous studies have shown that spouses living
together before marriage form a selective group that is more divorce-prone
than those entering into traditional marriages (Brüderl and Kalter 2001).
For older marriage cohorts in particular, cohabitation signals that spouses
are less committed to marriage than spouses who did not live together
before marriage. The observed patterns t well with the “low-commitment
hypothesis” (Amato and DeBoer 2001, Wolnger 2005). Accordingly, one
causal pathway is that children of divorce develop more sceptical attitudes
toward a long-lasting partnership, therefore choose less binding commit-
ments when starting a union, and are ultimately more likely to divorce.
The systematic efect of parental divorce on the pattern of educational
attainment is also noteworthy. In all countries except one, the children
of divorced parents participate in advanced education (ISCED 5 or 6) to
a lesser extent than respondents with non-divorced parents. Yet, as far as
possible efects of this disparity on marriage stability are concerned, edu-
cation is not consistently related to the respondent’s divorce risk and is
therefore not a mediator variable in explaining part of the transmission
efect in almost all the countries studied. Women’s education has mul-
tiple consequences on such divorce-related risk factors as labour-force
participation, personal income, household income and cultural prefer-
ences. Some of these factors increase divorce risks while others may have
a positive impact on marital stability. Thus, it is no surprise that we do
not nd a robust efect for a woman’s education on marital stability for all
the national data sets. Previous studies have also not reported consistent
ndings on the efect of a woman’s level of education on her divorce risk
(Dourleijn and Lieroer 2002).
5 Explaining Cross-Country Variation of Transmission Efects
Our estimations show substantial variation in the magnitude of the divorce
transmission efect ranging from 0.38 in Hungary to 1.34 in Italy. Can these
diferences across counties be explained by historical, institutional and cul-
tural diferences?
We did not nd any systematic pattern that related the magnitude of the
transmission efect to, for example, catholic religion, former communist
COSO 12.2_1-14-Diekmann and Schmidheiny.indd 10COSO 12.2_1-14-Diekmann and Schmidheiny.indd 10 2/15/2013 5:50:32 PM2/15/2013 5:50:32 PM
A. Diekmann, K. Schmidheiny / Comparative Sociology 12 (2013) 1–14 11
countries, etc. However, there is one striking relationship (see Figure 3):
the transmission efect is highly negatively correlated with the divorce
rates of the parent population (slope = 3.46, t-value = 4.38, R2 = 0.58). This
result is especially important as it seems to conrm the hypothesis that the
detrimental efects of divorce on children are stronger in societies where
divorce is rare and thus more likely stigmatised.
While highly appealing, we believe that this nding is an artefact for two
reasons. First, the negative correlation between transmission efect and
parental divorce rates found across countries is not found within countries
over time: while divorce rates increased in all countries, the transmission
efect decreased in only half of the countries and increased in the others.
Our scepticism is in line with Li and Wu (2004) who show that Wolnger’s
(1999) nding of a decrease in the U.S. divorce transmission efect was
awed. Second, the negative correlation between transmission efect and
Detailed estimation results are available from the authors on request.
Figure 3
Intergenerational divorce transmission and divorce rates by country.
Divorce Rate
Transmission Efect, beta
0 0.5 1 1.5
COSO 12.2_1-14-Diekmann and Schmidheiny.indd 11COSO 12.2_1-14-Diekmann and Schmidheiny.indd 11 2/15/2013 5:50:32 PM2/15/2013 5:50:32 PM
12 A. Diekmann, K. Schmidheiny / Comparative Sociology 12 (2013) 1–14
parental divorce rates found for the relative increase in divorce risk is not
found for the absolute increase in divorce risk: while the relative efects dif-
fer by a factor of almost 3 across countries, the absolute efects are of simi-
lar magnitude. Relative increases are the result of dividing absolute efects
by the baseline divorce risk. The signicant negative relationship between
the relative transmission efect and parental divorce rates can simply be
explained by the self-evident positive relationship of the left-hand side
numerator (the divorce rate of the observed generation) with the explana-
tory variable (the divorce rate of the parents’ generation).
6 Conclusions
This study investigates the intergenerational transmission of divorce in
13 European countries, the United States and Canada. We analyze the
cross-national data from female respondents in the Fertility and Family
Survey, applying techniques of event history analysis. We nd substantial
and highly statistically signicant transmission efects in all samples. This
shows that the intergenerational transmission of divorce is a widespread
phenomenon observed without a single exception in our data covering a
large number of countries with difering historical, institutional, and cul-
tural contexts.
Our study also demonstrates the presence of systematic patterns in the
consequences of divorce on children’s marital behaviour. Women whose
parents had divorced were, in all countries, also more likely to cohabit with
the men they eventually married than women who grew up with both of
their parents. This nding is in keeping with previous studies, which have
also noted greater rates of cohabitation among respondents with divorced
parents (e.g. Amato 1996, Diekmann and Engelhardt 1999, Kiernan and
Cherlin 1999). Our study adds to this nding by showing that the increased
tendency to cohabit is a rather universal phenomenon, one observed – at
least for female children – in all the countries we studied. This result ts
with Amato and DeBoer’s (2001) suggestion that the children of divorce
have less favourable attitudes toward marriage and therefore choose less
binding commitments when starting a union.
Detailed estimation results are available from the authors on request.
COSO 12.2_1-14-Diekmann and Schmidheiny.indd 12COSO 12.2_1-14-Diekmann and Schmidheiny.indd 12 2/15/2013 5:50:32 PM2/15/2013 5:50:32 PM
A. Diekmann, K. Schmidheiny / Comparative Sociology 12 (2013) 1–14 13
Amato, Paul R. (1996). Explaining the Intergenerational Transmission of Divorce. Journal of
Marriage and the Family 58: 628–640.
Amato, Paul R. and Danelle DeBoer (2001). The Transmission of Marital Instability Across
Generations: Relationship Skills or Commitment to Marriage? Journal of Marriage and
the Family 63: 1038–1051.
Amato, P. and B. Keith (1991a). Parental Divorce and the Well-Being of Children: A Meta-
Analysis. Psychological Bulletin 110: 26–46.
— (1991b). Parental Divorce and Adult Well-Being: A Meta-Analysis. Journal of Marriage
and the Family 53/1: 43–58.
Blossfeld, Hans-Peter and Goetz Rohwer (1995). Techniques of Event History Modeling. Hills-
dale, NY: Erlbaum.
Brüderl, Josef and Kalter, Frank (2001). The Dissolution of Marriages. The Role of Informa-
tion and Marital-Specic Capital. Journal of Mathematical Sociology 25: 403–421.
Bumpass, Larry L. and James A. Sweet (1972). Diferentials in Marital Instability. American
Sociological Review 37: 754–766.
Cox, D. R. (1972). Regression Models and Life Tables. Journal of the Royal Statistical Society,
Series B, 34: 187–220.
Diekmann, Andreas and Peter Mitter (1984). A Comparison of the ‘Sickle Function’ with
Alternative Stochastic Models of Divorce Rates. In Andreas Diekmann and Peter Mitter,
eds. Stochastic Models of Social Processes. Orlando, Fl.: Academic Press.
Diekmann, Andreas and Henriette Engelhardt (1999). The Social Inheritance of Divorce: Efect
of Parent’s Family Type in Postwar Germany. American Sociological Review 64: 783–793.
Diekmann, Andreas and Kurt Schmidheiny (2008). Comment on: Jaap Dronkers and Juho
Härkönen (2006), The divorce cycle in cross-national perspective: results from the Fertil-
ity and Family Surveys. Unpublished manuscript: ETH Zurich.
Dourleijn, Edith and Aart C. Lieroer (2002), Unmarried Cohabitation and Union Stability.
Testing the Role of Difusion Using Data from 16 European Countries. Paper presented at
the conference on “Divorce in a Cross-National Perspective. A European Research Net-
work”, Florence, November 2002.
Dronkers, Jaap and Juho Härkönen (2008). The intergenerational transmission of divorce
in cross-national perspective: Results from the Fertility and Family Surveys. Population
Studies 62: 273–288.
Festy, Patrick and France Prioux (2002). An Evaluation of the Fertility and Family Surveys
Project in the Countries of the UNECE Region. New York and Geneva: United Nations Eco-
nomic Commission for Europe.
Glenn, Norval D. and Kathryn B. Kramer (1987). The Marriages and Divorces of the Children
of Divorce. Journal of Marriage and the Family 49: 811–825.
Greenberg, Ellen F. and W. Robert Nay (1982). The Intergenerational Transmissionof Marital
Instability Reconsidered. Journal of Marriage and the Family 44: 335–347.
Keith, Verna M. and Barbara Finlay (1988). The Impact of Parental Divorce on Children’s
Educational Attainment, Marital Timing, and Likelihood of Divorce. Journal of Marriage
and the Family 50: 797–809.
COSO 12.2_1-14-Diekmann and Schmidheiny.indd 13COSO 12.2_1-14-Diekmann and Schmidheiny.indd 13 2/15/2013 5:50:32 PM2/15/2013 5:50:32 PM
14 A. Diekmann, K. Schmidheiny / Comparative Sociology 12 (2013) 1–14
Kiernan, Kathleen E. and Andrew J. Cherlin (1999). Parental Divorce and Partnership Disso-
lution in Adulthood: Evidence from a British Cohort Study. Population Studies 53: 39–48.
Li, Lui-Chung Allen and Lawrence L. Wu (2004), No Trend in the Intergenerational Transmis-
sion of Divorce. Department of Sociology, New York University. Mimeo.
McLanahan, Sara and Larry Bumpass (1988). Intergenerational Consequences of Family Dis-
ruption. American Journal of Sociology 94: 130–152
Mott, Frank L. and Sylvia F. Moore (1979). The Causes of Marital Disruption among Young
American Women: An Tnterdisciplinary Perspective. Journal of Marriage and the Family
41: 355–365.
Mueller, Charles W. and Hallowell Pope (1977). Marital Instability: A Study of its Transmis-
sion between Generations. Journal of Marriage and the Family 39: 83–92.
Pope, Hallowell and Charles W. Mueller (1976). The Intergenerational Transmission of Mari-
tal Instability: Comparisons by Race and Sex. Journal of Social Issues 35: 112–125.
Schoenmaeckers, Ronald, and Edith Lodewijcky (1999). Changes in Demographic Behaviour
in Europe: Some Results from FFS-Country Reports and Suggestions for further Research.
European Journal of Population 15: 207–240.
Teachman, Jay D. (1982). Methodological Issues in the Analysis of Family Formation and
Dissolution. Journal of Marriage and the Family 44: 1037–1053.
Traag, Tanja, Jaap Dronkers and Louis-André Vallet (2000). The intergenerational transmis-
sion of divorce risks in France. ASSR working paper 68, presented to the Research Com-
mittee 28 – Social Stratication of the International Sociological Association in Libourne,
France May 11–14, 2000.
UNESCO (1997). ISCED. International Standard Classication of Educatio. UNESCO Institute
for Statistics.
White, Lynn K. (1990). Determinants of Divorce: A Review of Research in the Eighties. Jour-
nal of Marriage and the Family 52: 904–912.
Wolnger, Nicholas H. (1995). Trends in the Intergenerational Transmission of Divorce.
Demography 36: 415–20.
Wolnger, Nicholas (1999). Trends in the Intergenerational Transmission of Divorce.
Demography 36: 415–420.
COSO 12.2_1-14-Diekmann and Schmidheiny.indd 14COSO 12.2_1-14-Diekmann and Schmidheiny.indd 14 2/15/2013 5:50:32 PM2/15/2013 5:50:32 PM
... V Evropě se mezigeneračním přenosem rozvodu zabývali například ve Francii [Traag, Dronkers, Vallet 2000], Švédsku [Gähler, Härkönen 2014], Německu [Engelhardt, Trappe, Dronkers 2002] či Velké Británii [Kiernan, Cherlin 1999] nebo Norsku [Lyngstad, Engelhardt 2009]. Výzkum byl v posledních letech rozšířen i o mezinárodní srovnání [Dronkers, Härkönen 2008;Diekmann, Schmidheiny 2013]. ...
... Hlavním předmětem výzkumů ovšem není samotná existence mezigeneračního přenosu rozvodu. Například Amato označuje rozvod rodičů za "dobře zdokumentovaný rizikový faktor pro ukončení manželství" [Amato 1996: 628], který můžeme pozorovat napříč různými kulturami [Diekmann, Schmidheiny 2013;Dronkers, Härkönen 2008]. Výzkumníci se proto snaží především přijít na vysvětlení vztahu mezi rozvodem rodičů a rozvodem dětí, jež zatím zůstává nejasné [Amato 1996]. ...
... Nízký věk při první svatbě je přitom jedním z dobře zdokumentovaných rizikových faktorů rozvodu [Booth, Edwards 1985;Härkönen, Dronkers 2006]. Zároveň jsou to právě děti rozvedených rodičů, kteří před vstupem do manželství častěji žijí v kohabitaci [Thornton 1991;Diekmann, Schmidheiny 2013;Härkönen, Brons, Dronkers 2020], což je další faktor, který zvyšuje riziko rozvodu [Dush, Cohan, Amato 2003]. Kromě toho dosahují tyto děti nižšího vzdělání [Kreidl, Štípková, Hubatková 2017], přičemž nízký socioekonomický status je opět spojen s vyšším rizikem rozvodu [McLeod 1991;Gähler, Palmtag 2015]. ...
... Divorce is a commonly studied example of relationship instability, and the research on this topic has found that individuals who experience a parental divorce are more likely to divorce in adulthood themselves (e.g., Amato, 1996;Bumpass et al., 1991;Pope & Mueller, 1976). This intergenerational transmission has been consistent in studies across the United States, a variety of European countries, and Canada (e.g., Diekmann, & Schmidheiny, 2013;Dronkers, & Härkönen, 2008). Additionally, the number of parental relationships an offspring witnessed during childhood is positively associated with their own number of partners in early adulthood, demonstrating an intergenerational transmission of union transitions (Amato & Patterson, 2017;Kamp Dush et al., 2018). ...
This study extends prior research on the intergenerational transmission of relationship instability by examining parents' history of on‐off relationships as a predictor of emerging adults' own cycling (i.e., breaking up and renewing with the same romantic partner). Data were collected at a large mid‐western university from 702 emerging adults (18–25 years old). Multinomial logistic regression was used to predict the likelihood that participants had cycled in a past or current relationship. Results show that parental cycling increased the likelihood of offspring cycling in a past or current relationship relative to never cycling, and greater uncertainty about the future of the relationship was a mechanism through which such transmission occurred. Findings from this study demonstrate that parental relationship instability can even be consequential for the transient relationships within emerging adulthood, making family history a productive area to explore for practitioners working with cyclical partners and/or emerging adults. 上一代人的关系不稳定对下一代人关系的代际传递影响已有研究,本研究将扩展该研究课题。我们研究调查了父母之间时分时合的关系发展史如何能作为一个预告因子来预测刚踏入成人阶段的青年人自己的感情发展循环 (即与同一情侣分手又和好)。研究人员从美国中西部一所大型大学收集了702名18‐25岁的年轻人的数据。多项逻辑回归被用来预测参与者在过去或现在的关系中分分合合的可能性。结果表明,与从不没有分分合合历史的人相比,父母的分分合合关系史增加了子女过去或现在的爱情关系中分分合合的可能性,而对未来关系发展的更大的不确定性是这种代际传递发生的机制。这项研究的结果表明,父母关系的不稳定甚至可能对刚踏入成年期的年轻人较短的情侣关系产生影响,这使得家族的关系发展情况值得特定的一类行医者们视其为一个亟待深入探索研究并多产结果的领域,他们的治疗对象主要是和关系时好时坏的伴侣们和/或刚踏入成年期的年轻人。
We use variation in the intergenerational persistence across social assistance benefits over 18 years to study the drivers of intergenerational disadvantage. Young people are more likely to receive social assistance if their parents received disability, caring, or single parent benefits, and less likely if they received unemployment benefits. Disparity in intergenerational persistence across benefit types suggests that parental bad luck has broader consequences for youth disadvantage than do their personal choices. Using the intensive margin and timing of parental social assistance to account for unobserved heterogeneity indicates that intergenerational disadvantage is more likely driven by poverty traps than welfare cultures.
This article examines the transmission of older women's relationship quality with their mothers and fathers to their relationship quality with their own adult children in midlife. We also investigate how the transmission is moderated by the dimension of relationship quality (closeness vs. strain) and the gender of both the older women's parents and their adult children. Prior research has primarily examined parents' transmission of relationship quality to young children with little attention to whether and when this pattern occurs in later‐life families. We conducted multilevel analyses using data collected from 249 older women and 643 of their adult children as part of the Within‐Family Differences Study‐I. We found evidence for transmission of older women's reported closeness and tension with their mothers and fathers to their reported closeness and tension with their adult children. Adult children's reports also revealed that older women's closeness with their own mothers was transmitted to their adult children's reported closeness with the older women themselves. Mother–child closeness was transmitted more strongly than mother–child tension, and mother–child closeness was transmitted more strongly to daughters than sons, based on adult children's reports. This study demonstrates the continuity of intergenerational influence in later‐life families and highlights the essential roles that selective social learning and social structural position (i.e., gender) play in conditioning the socialization process.
Over the past few decades, the increasing divorce rate has been one of the most prominent behavioral changes influencing Chinese families and the nurturing and socialization of children. Research has found that parental divorce exerts only a limited negative impact on children's socioeconomic achievement in China relative to that in Western societies. However, few studies have explored the long-term consequences of parental divorce on children's demographic outcomes in China. Therefore, how parental divorce influences the timing of offspring's first sexual intercourse and marriage, as well as its impact on their sexual and marital well-being, were investigated in this study. Based on findings obtained using data from the Chinese Private Life Survey, children from divorced families were more likely to initiate sexual intercourse at younger ages than those from intact families, although the two groups entered their first marriage at similar ages. Regarding sexual and marital well-being, married men and women who experienced parental divorce during childhood were less satisfied with their current marriage and marital sex and exhibited a higher level of divorce proneness and more sexual dysfunction symptoms than those from intact families. The effect of parental divorce on marital well-being was also largely mediated by the onset of sexual intercourse at a younger age. Having more years of premarital sexual experience was associated with lower sexual satisfaction, more sexual dysfunction symptoms, and more liberal sexual attitudes and behaviors. The worsening of sexual life also further undermined marital well-being.
This study is a randomized controlled trial of an asynchronous, fully web‐based program for divorced and separated parents, the electronic New Beginnings Program (eNBP). This program is an adaptation of a group, in‐person program for divorced parents, the New Beginnings Program (NBP), which has been shown in randomized trials to reduce a wide range of offspring problems and improve a wide range of competencies up to 15 years after participation. The 10‐module, 5‐h program uses evidence‐based, highly interactive strategies to teach skills designed to strengthen parenting after divorce and reduce interparental conflict. Participants were 131 parents (63% mothers) and 102 adolescent offspring. Parents were randomly assigned to the eNBP or a wait‐list control condition. Parents and their children completed pre‐ and post‐tests. Analyses showed that at post‐test, parents and children in the eNBP reported significantly higher parent–child relationship quality, more effective discipline, lower interparental conflict and lower child mental health problems than did those in the wait‐list control condition. These are the strongest findings in the literature on the effects of web‐based programs to reduce interparental conflict, strengthen positive parenting and reduce children's post‐divorce mental health problems. Given that parental divorce has significant individual and societal costs, widespread implementation of this program could have significant public health implications. A fully web‐based parenting‐after‐divorce program reduced interparental conflict, increased quality of parenting and reduced children's post‐divorce behavior problems as reported by both parents and children. Web‐based programs to improve parenting can be a cost‐effective approach to promote children's post‐divorce adjustment. Widespread implementation of easily accessible, web‐based parenting‐after‐divorce programs can significantly reduce the negative impact of divorce on children's outcomes.
Numerous studies have shown that parental divorce is associated with an increase in adult children's divorce risk. We extend this literature by assessing how parental divorce on both sides of a couple is related to their partnership dynamics, specifically, whether there is parental divorce homogamy and whether a history of parental divorce for both partners is associated with increased dissolution risks for cohabiting and married unions. We use Finnish Census Panel data on 28,021 cohabiting and marital partnerships to conduct event-history models that follow individuals between ages 18 and 45. Findings show substantial parental divorce homogamy. Children with experience of parental divorce have 13% greater odds of cohabiting with and 17% greater odds of marrying a fellow child of divorcees, compared with those whose parents have not divorced. Moreover, contrary to evidence from the United States and Norway, our findings for Finland support an additive—rather than multiplicative—association between parental divorce homogamy and union dissolution. Parental divorce homogamy increases offspring's union dissolution risk by 20% for cohabitation and 70% for marriage, compared with couples for whom neither partner's parents are divorced. In Finland, the sizes of these associations are notably weaker than in the United States and Norway, likely because cohabitation and separation are more widespread and socially accepted in Finland, and an expansive welfare state buffers the socioeconomic consequences of divorce.
This study investigated whether the Abandonment schema mediates the relationship between a history of parental separation or divorce and attachment styles in adulthood. Participants (N = 426; Mage = 30.78 years) completed online measures, using Qualtrics. Results demonstrated that a history of parental separation or divorce was positively associated with anxious and avoidant attachment styles in adulthood. Mediation analyses revealed that the Abandonment schema mediated these relationships (p <.001). This suggests that adults with a history of parental divorce or separation are more likely to report anxious and avoidant attachment styles, and this is partially accounted for by an expectation that they will be abandoned by others. These findings have implications for case formulation and can be used to inform treatment. Schema therapy focused on healing the Abandonment schema could promote attachment security in individuals with a history of parental separation or divorce. For example, limited reparenting (i.e., the therapist seeking to meet the client’s unmet emotional needs, within the professional boundaries of the relationship), could be tailored to address the individual’s underlying unmet need for security and stability. Further research is needed to advance our understanding of the implications of these findings for treatment.
Partners in intimate relationships, because they have each other to rely on, have generally been considered safe from the negative consequences of social isolation. Here we question this assumption, suggesting instead that social isolation may pose a threat to couples by depriving them of the tangible and emotional support that couples are likely to need, especially when confronted by stress. After briefly reviewing theoretical frameworks relevant to this idea, this article summarizes existing research documenting: (1) associations between network ties and relationship outcomes, (2) mediators of these associations, e.g., support and approval, and (3) moderators of these associations, e.g., relationship qualities and cultural differences. We conclude by describing a research agenda to address methodological limitations in existing research and the policy implications of this line of work.
Full-text available
The social inheritance of divorce is one factor contributing to the upward trend in marriage dissolution rates during the last few decades. Several studies confirm the transmission hypothesis for U.S. marriages. We investigate the intergenerational transmission of divorce risk among German first marriages using multivariate event-history techniques. Our data are from the 7,200 respondents of the German Family Survey. The historical circumstances of postwar Germany allow a comparison between families dissolved by war and familities dissolved by divorce. Respondents whose parential families dissolved by the death of a parent have only slightly higher divorce risks than respondents who grew up in two-parent families. There is, however, a large gap in marital instability for respondents from divorced-parent families compared with respondents from two-parent families and families with a widowed parent. Hence, the inheritance of divorce cannot be explained simply by the absence of a parent. The data suggest that differences in personal investments in the marriage partnership may partially explain the transmission effect.
Growing up in a divorced family can cause the children to have difficulties in maintaining relationships. Nicholas Wolfinger demonstrates the significant impact of parental divorce upon people's lives and society. The divorce cycle phenomena ensures the transmission of divorce from one generation to the next. This book examines how it has transformed family life in contemporary America by drawing on two national data sets. Compared to people from intact families, the children of divorced parents are more likely to marry as teenagers, but less likely to wed overall. They are more likely to marry other people from divorced families, but more likely to dissolve second and third marriages, and less likely to marry their live-in partners.
A meta-analysis was conducted of studies dealing with the long-term consequences of parental divorce for adult well-being. Effect sizes were calculated for 15 outcome variables across 37 studies involving over 81,000 individuals. Mean effect sizes were significant and negative for all outcomes; this indicated that adults who experienced parental divorce exhibited lower levels of well-being than did adults whose parents were continuously married. The strongest estimated effects occurred in the areas of one-parent family status, psychological adjustment, behavior/conduct, and educational attainment. Effect sizes were significantly stronger in clinical studies than in studies based on community samples. In addition, effects sizes tended to be stronger for whites than for blacks, stronger in earlier studies than in more recent studies, and stronger in studies that did not use statistical controls than in studies that did.
The argument presented here is that parental divorce diminishes the economic and social resources available to children, which in turn has negative consequences for children's educational attainment, marital timing, marital probability, and divorce probability. Based upon a combined sample of national data, for white respondents only, the analysis shows that parental divorce is associated with lower educational attainment and earlier age at marriage for both sexes. Daughters of divorced parents have a higher probability of being divorced. For sons of divorced parents, the probability of ever marrying is lower and of divorcing is higher only if they have lower social class backgrounds.
The determinants of marital disruption for a nationally representative cross-section of black and white women aged 14 to 24 in 1968 are examined from an interdisciplinary perspective. Utilizing data from the National Longitudinal Survey of Labor Market Behavior of Young Women, a multivariate analysis incorporating a full range of economic, social and demographic variables examined the determinants of a first marital disruption between 1968 and 1973 for young women who were married at any point during that interval. Whereas economic factors were found to have some importance, other social and demographic factors appeared to have a more significant independent effect on the probability of disruption.
Family formation and dissolution is a process by which individuals make transitions between different statuses, such as married or not married. These family life-course transitions can be characterized by their number, timing, and sequence. While significant changes in these three dimensions of life-course transitions have been documented in recent decades, appropriate methodological tools for uncovering the determinants of such changes have not been generally available. This paper describes a methodological technique, proportional hazards models, which allows researchers to investigate variations in the number, timing, and sequence of life-course transitions in a multivariate framework. The technique is illustrated by an analysis of marital dissolution in a sample of white women.
A 1970 national sample of white ever-married females is used to explore the process of the intergenerational transmission of marital instability. The research examines the possibility that mate-selection outcomes operate as intervening variables between parent and child generation marital instability. Partial support is found for this: about one-half of the effect of parent instability is mediated by mate-selection out-comes, with the high-risk circumstances of early and limited-education marriages being the most important. The relevance of these mate selection circumstances in the transmission process is interpreted within the framework of social control and economic rationales.
This study uses national longitudinal data to explain the intergenerational transmission of divorce. Parental divorce is associated with an increased risk of offspring divorce, especially when wives and both spouses have experienced the dissolution of their parents' marriage. Offspring age at marriage, cohabitation, socioeconomic attainment, and prodivorce attitudes mediate modest proportions of the estimated effect of parental divorce. In contrast, a measure of interpersonal behavior problems mediates the largest share of the association. The findings suggest that parental divorce elevates the risk of offspring divorce by increasing the likelihood that offspring exhibit behaviors that interfere with the maintenance of mutually rewarding intimate relationships.
The effect of parental divorce on the divorce-proneness of offspring was estimated separately for white males, white females, black males, and black females through analysis of pooled data from 11 U.S. national surveys conducted from 1973 to 1985. The estimated effect for white females was substantial and statistically significant, but any effects in the other race-sex categories appear to have been moderate. Analyses performed to test some common and plausible explanations for an intergenerational transmission of divorce-proneness yielded indirect support for a "lower-commitment-to-marriage" explanation and revealed that a small proportion of the estimated transmission effect can be explained by a tendency for the children of divorce to marry at an early age.