Contents lists available at ScienceDirect
Advances in Life Course Research
journal homepage: www.elsevier.com/locate/alcr
Grandparenting after divorce: Variations across countries
Zuzana Žilinčíkováa,⁎
, Martin Kreidlb
a
Faculty of Social Studies, Masaryk University, Jostova 10, 60200 Brno, Czech Republic
b
Oﬃce for Population Research, Faculty of Social Studies, Masaryk University, Jostova 10, 60200 Brno, Czech Republic
A R T I C L E I N F O
Keywords:
Grandparents
Grandchildren
Divorce
Intergenerational relationships
Intergenerational exchange
Grandparental childcare
A B S T R A C T
We analyze the eﬀect of grandparental divorce on the odds of providing grandparental childcare and investigate
the variation of this eﬀect across countries. The analysis is based on three-level hierarchical linear models, using
data collected between 2004 and 2011 in 18 European countries as a part of the SHARE project. Being divorced
is clearly associated with a signiﬁcant reduction in the odds of providing any grandparental childcare in the past
12 months as well as in the odds of providing intensive (at least once a week) childcare. There is, however,
signiﬁcant variation in the divorce eﬀect across countries. Looking at any grandparental childcare, we see that
the negative eﬀect of divorce is signiﬁcantly weaker at higher divorce rates. The disruptive eﬀect of divorce
declines by almost 30 per cent when crude divorce rate increases by one standard deviation. We conclude that
the level of conﬂict typically associated with divorce is lower when family disruption is more common, and its
disruptive eﬀect is weaker: intergenerational contact is thus preserved more often, and grandparental childcare
provision is more common. Moreover, social institutions related to divorce and post-divorce arrangements may
be more developed in countries with a higher incidence of divorce and thus they partially mitigate the negative
eﬀect of divorce. We do not conﬁrm the same pattern when studying intensive grandparental childcare. Despite
the low statistical signiﬁcance, the trend seems to be the opposite: the eﬀect of divorce becomes stronger with
growing incidence of divorce. We attribute this latter trend to a complex re-organization of the lives of the
divorcees that constrain their availability for intensive caregiving.
1. Introduction: divorce and intergenerational caregiving
Two recent population trends – increasing longevity and increasing
rates of family dissolution – accentuate the importance of intergenerational
relations in advanced societies. Increasing longevity implies
that grandparents and grandchildren have – on average – more
years of shared lives (Bengtson, 2001; see also Connidis 2010; Suitor
et al. 2011; Uhlenberg 2004). The signiﬁcance of intergenerational
bonds stands out especially when contrasted with increasingly fragile
intra-generational ties (Bengtson et al., 2002). Grandparental childcare
– the focus of this paper – is a prominent example of intergenerational
help and it is an essential source of support for children (Gray, 2005).
Hank and Buber (2009) found – in a sample of several European
countries – that 58 percent of grandmothers and 49 percent of grandfathers
provide some kind of care for their grandchildren. Other research
has demonstrated that the availability of grandparental childcare,
as well as other forms of childcare, may positively inﬂuence
fertility (Aassve et al., 2012a,b; Brewster & Rindfuss, 2000; Del Boca,
2002; Hank & Kreyenfeld, 2003; Kaptijn et al., 2010) and maternal
employment (Aassve et al., 2012a,b; Del Boca, 2002; Gray, 2005;
Kaptijn et al., 2010).
This paper focuses on the eﬀects of grandparental divorce on
grandparental childcare in a comparative perspective. Detailed, context-sensitive
analyses of the consequences of grandparental separation
are needed not only because of the general increase in divorce rates, but
because increasing family dissolution rates have been observed in older
generations as well (Brown & Lin, 2012; Hammond & Muller, 1992; Wu
& Penning, 1997). Thus, there is a growing need to better understand
the eﬀects of family dissolution on inter-generational relations and
exchange. We know from existing research that parental break-up often
leads to disruption of intergenerational ties with adult children and
results in reduced levels of support between generations (Kalmijn,
2008; King, 2003; Lin 2008). It also lowers the odds of grandparentgrandchild
contact, since members of the middle generation often serve
as mediators and gatekeepers of inter-generational contact and caregiving
(Crosnoe and Elder, 2004; Monserud, 2008; Mueller & Elder,
2003; Whitbeck et al., 1993). In addition, divorced parents tend to live
farther from their adult children (Chan & Ermish, 2015) and geographic
distance per se functions as a barrier to both contact and caregiving
(Devine & Earle, 2011; Jappens & Van Bavel, 2012; Uhlenberg &
https://doi.org/10.1016/j.alcr.2018.08.003
Received 31 January 2017; Received in revised form 3 July 2018; Accepted 26 August 2018
⁎
Corresponding author.
E-mail addresses: zilincikova@mail.muni.cz (Z. Žilinčíková), kreidlm@fss.muni.cz (M. Kreidl).
Advances in Life Course Research 38 (2018) 61–71
Available online 08 September 2018
1040-2608/ © 2018 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/BY-NC-ND/4.0/).
T
Hammill 1998).
We observe two contradictory demographic trends related to
grandparenthood. Whereas increasing longevity means that there are –
on the one hand – more grandparents available to provide grandparental
childcare, growing family dissolution rates – on the other hand
– imply that this resource may be utilized less frequently. With these
two major population developments in mind, we investigate the disruptive
eﬀect of grandparental divorce on the odds of providing any (or
intensive) grandparental childcare. We examine whether (and to what
extent) the eﬀect of grandparental divorce varies across societies. More
speciﬁcally, we link the size of the divorce eﬀect to the prevailing divorce
rate. We argue that higher family dissolution rates are associated
with more liberal divorce laws, less parental conﬂict, weaker self-selection
into divorce, and more institutionalized post-divorce arrangements.
These circumstances lead us to hypothesize that the disruptive
eﬀect of divorce on care provision should be weaker when divorce is
more common. We test this innovative theoretical assertion using internationally
harmonized data from SHARE (Survey of Health, Ageing,
and Retirement in Europe) complemented by selected macro-variables
taken from other data sources.
In agreement with earlier research, we show that grandparental
divorce is indeed an important predictor of the odds of any (or intensive)
grandchild care being provided. At the same time, we show
that the eﬀect varies in a predictable manner by the incidence of divorce
in a country. Across countries, the eﬀect of divorce on any care
provision appears to be much weaker when divorce is more common;
more precisely, the disruptive eﬀect of divorce declines by almost 30
per cent when crude divorce rate increases by one standard deviation.
However, contrary to our hypotheses, we also observe that the disruptive
eﬀect of divorce on providing intensive grandchild care becomes
stronger at higher divorce rates. Generalizing from a cross-sectional
comparison to within-country trends, we might foresee two somewhat
contradicting trends: the disruptive eﬀect of divorce on grandchild care
provision should shrink as divorce rates increase, but the eﬀect of divorce
on intensive grandchild care might strengthen. That is, we might
expect that divorced grandparents would be available more frequently
to provide some care for their grandchildren in the coming decades, but
will be less frequently available to provide intensive care.
2. Grandparents’ marital status and grandparental childcare:
theories and empirical evidence
On the theoretical level, the disruptive eﬀect of divorce on intergenerational
contact and caregiving may be explained with reference to
high levels of stress and conﬂict associated with family dissolution. For
instance, poor parental adjustment to divorce might interfere with
parenting skills (Amato, 1993) and the parent-child relationship might
bear lasting consequences. Moreover, conﬂicting relationships not only
exist between the two divorcing individuals, but might develop also
between parents and children. Children are often directly involved in
parental quarrels and may also be forced to “choose sides”, the parentchild
relationship often deteriorating as a result (Amato, 1993, 2000).
Divorce occurring earlier in life usually constrains contact between the
child and one parent (in the case of a sole custody) or both parents (in
the case of shared custody) with far-reaching implications for future
patterns of inter-generational interaction (Bulcroft & Bulcroft, 1991;
Westphal et al., 2015); for instance, aﬀection – especially between the
child and the non-resident parent – erodes (Amato & Booth, 1996).
However, even divorce (and other partnership transitions) occurring at
a later age might impact intergenerational relationships to a signiﬁcant
degree and may, for example, lead to feelings of shock and/or guilt and
cause tension, conﬂict, and alienation (Ahrons, 2004; Aquilino, 1994;
Cain 1989; Cooney 1994; Connidis 2010; Cooney et al., 1995;
Danielsbacka & Tanskanen, 2016; Lin 2008; Reed et al., 2016; Schenk,
& Dykstra 2012; White 1992).
If any harm is done to parent-child relationships in splitting
families, it is likely to have persistent implications for intergenerational
contact and caregiving (including grandparent-grandchild interaction),
because the middle generation (children) usually serves as mediators of
contact between grandparents and grandchildren (Hodgson, 1998; King
& Elder, 1995; Rossi & Rossi, 1990; Uhlenberg & Hammill, 1998). Indeed,
several studies (Jappens & Van Bavel, 2016; King, 2003;
Uhlenberg and Hammill, 1998) have found that the quality of the relationship
between the oldest and the middle generation mediates
grandparental care. Caregiving by divorced grandparents is, we argue,
often impeded by the poor quality of the relationship between the older
(grandparents) and the middle (children) generation that results from a
grandparental divorce.
Consistent with the theoretical argument presented above, previous
studies have shown that the marital status of grandparents is an important
predictor of grandparent-grandchild contact (Albertini &
Garriga, 2011; Jappens & Van Bavel, 2012; King, 2003; Silverstein &
Marenco, 2001; Uhlenberg & Hammill, 1998) or care provision by
grandparents (Baydar & Brooks-Gunn, 1998; Danielsbacka &
Tanskanen, 2016; Hank & Buber, 2009; Knudsen, 2012; Luo et al.,
2012; Fuller-Thomson & Marenco, 2001; Thomson & Minkler, 2001;
Musil et al., 2006). Focusing more speciﬁcally on the diﬀerence between
married and divorced grandparents, the literature indicates that
married grandparents are more likely than unmarried grandparents to
have contact with grandchildren (Albertini & Garriga, 2011;
Danielsbacka & Tanskanen, 2016; Jappens and Van Bavel, 2012; King,
2003; Uhlenberg & Hammill, 1998), to provide care for grandchildren
(Danielsbacka & Tanskanen, 2016; Fuller-Thomson & Minkler, 2001;
Luo et al., 2012; Silverstein & Marenco, 2001), and are – in general –
more likely to engage in intergenerational support (Daatland, 2007;
White, 1992). For example, King (2003) found that the odds of everdivorced
grandparents having frequent contact with their grandchildren
are 40 percent lower than those of never-divorced grandparents.
Silverstein and Marenco (2001) found a similar pattern – unmarried
grandparents had approximately 50 percent lower odds of
babysitting their grandchildren than their married counterparts.
Whereas the majority of empirical studies consistently report a negative
association between divorce and intergenerational ties and support, a
recent Dutch study by Geurts et al. (2015) did not ﬁnd any such association,
highlighting the need to investigate this association across
contexts and explore its macro-social prerequisites and correlates, i.e.
something this analysis attempts.
Divorce typically brings greater restrictions on, and a greater
weakening of the father-child relationship than the mother-child relationship
(King, 2003). Women are not only more likely to have custody
of their children after divorce, but are in general more strongly
involved in maintaining relationships with family members (Bracke
et al., 2008). Thus, it is likely that divorced grandfathers are less likely
than divorced grandmothers to have intensive contact and a strong
relationship with their children and grandchildren. Empirical research
renders strong empirical support to this argument. While the negative
eﬀect of divorce has been shown to exist for both men and women, it
tends to be much stronger among grandfathers (Baydar & Brooks-Gunn,
1998, Danielsbacka & Tanskanen, 2016; Hank & Buber, 2009; King,
2003; Uhlenberg & Hammill, 1998; White, 1992).
Divorce also impacts the geographic proximity of children and
grandchildren (Cherlin & Furstenberg, 1985; Chan & Ermisch, 2015);
divorced grandparents – especially grandfathers – tend to live farther
away from their children and grandchildren (Chan & Ermisch, 2015;
King, 2003). Closer geographic proximity correlates with a higher frequency
of inter-generational contact (Uhlenberg & Hammill 1998;
Barnett et al. 2010; Cherlin & Furstenberg, 1985) and greater odds of
the existence of intergenerational caregiving (Jappens & Van Bavel,
2012; Devine & Earle, 2011). Geographic proximity, as well as contact
with children (and grandchildren), might be inﬂuenced by the timing of
a divorce. If the divorce occurs when the children are young the harm to
intergenerational ties tends to be more severe, especially for the nonZ.
Žilinčíková, M. Kreidl Advances in Life Course Research 38 (2018) 61–71
62
resident parent (Bulcroft & Bulcroft, 1991).
3. Marital status and grandparental childcare across countries
Married grandparents are, as we know from existing research, more
likely than unmarried grandparents to provide care for grandchildren.
Most of the studies reporting this ﬁnding employ data collected in the
United States and only some have a European focus (Albertini &
Garriga, 2011; Danielsbacka & Tanskanen, 2016; Hank & Buber, 2009;
Jappens & Van Bavel, 2012; Knudsen, 2012). Some other studies (Chen
et al., 2011; Ko & Hank, 2013; Lee & Bauer, 2010) investigated
grandparental caregiving in Asia, but a systematic comparative analysis
of the disruptive eﬀect of grandparental divorce on grandparental
childcare is lacking in the literature.
While divorce may disrupt inter-generational ties in any society, we
argue that the disruptive eﬀect may vary in size across contexts, reﬂecting,
for instance, how common divorce is in a country. Albertini &
Garriga (2011) proposed a “collective declining eﬀect hypothesis”,
stating that the negative eﬀect of divorce declines with the rising divorce
rate in society (see also Dronkers et al. 2006; Goode, 1993;
Kalmijn & Uunk, 2007; Wolﬁnger 1999). The authors argue that there
are three main underlying mechanisms behind the hypothesis. First, the
more common divorce is, the lower the stigma associated with divorce.
Second, people are better aware of the possible negative consequences
of divorce and develop (and later institutionalize) strategies to mitigate
them. Third, in contexts where divorce is more common, there are a
smaller proportion of high-conﬂict couples among those who divorce.
The level of within-family conﬂict associated with divorce has declined
over cohorts (Amato & Hohmann-Marriott, 2007; Gähler & Palmtag,
2015), while divorce rates have increased over the same periods. Since
there is less conﬂict associated with family dissolution at higher family
dissolution rates, the disruptive eﬀect of divorce should be less severe
and the potential for grandparent-grandchild contact should be more
likely to persist.
This conclusion might appear to be in contrast with ﬁndings from
several studies that have suggested that divorce of low-conﬂict families
might result in worse outcomes for children than dissolution of highconﬂict
families (Amato et al., 1995; Booth & Amato, 2001; Hanson,
1999; Jekielek, 1998, Kreidl et al., 2017). Yet, research focusing more
speciﬁcally on parent-child involvement suggest the opposite. With
rising levels of parental conﬂict, the post-divorce involvement between
father and child deteriorates (Kalmijn, 2010; Riggio, 2004). Thus, with
declining average levels of conﬂict in splitting families, one should also
expect higher post-divorce involvement of fathers. Indeed, this assertion
has been supported empirically. For instance, Gähler and Palmtag
(2015) have shown – using Swedish data – that it was very uncommon
for children in earlier birth cohorts to have contact with their noncustodial
parent after parental separation: only less than 20% of children
born between 1925–1949 report to have had contact with the
absent parent at least once a month. Children born between 1980–1991
do report such contact in 80% of cases (Gähler & Palmtag, 2015: 612).
The logic of the “collective declining eﬀect” (Albertini & Garriga,
2011) can also be applied to intergenerational relations and grandparental
caregiving. Lower stigma might mean – on the one hand –
greater acceptance of divorce in society, but also – on the other hand –
among children of divorce, who would be less inclined to disrupt their
relationships with parents on account of feelings of guilt, blame, and/or
alienation. Similarly, better developed institutions related to post-divorce
interactions between former spouses and between parents and
children (e.g. shared custody and the system of visitation rights, alimony
payments, but also less formalized institutions related to the role
of the non-coresident parent etc.) to cope with the consequences of
divorce can help parents and children maintain a closer post-divorce
relationship. Lastly, less conﬂict within the family provides a better
basis for the post-divorce interaction between parents and children:
when there are less conﬂict, tension and stress, intergenerational
relations are more likely to persist undisturbed and intergenerational
contact is more frequent thus creating a stronger foundation for future
intergenerational support and exchange. Overall, it appears that divorced
grandparents should be more frequently available to provide care
for their grandchildren under conditions of higher family dissolution rates.
So far, empirical evidence on the “collective declining eﬀect hypothesis”
is not congruent across a wide range of outcomes (Kalmijn,
2010; Kreidl et al., 2017; Sigle-Rushton et al., 2005; Wagner & Weiß,
2006) including intergenerational ties (Albertini & Garriga, 2011;
Daatland, 2007; Kalmijn, 2008). While Albertini and Garriga (2011)
and Daatland (2007) observed no signiﬁcant variation in intergenerational
ties across contexts with varying incidences of divorce, Kalmijn
(2008) observed a signiﬁcant variation in the frequency of contact
between divorced fathers and their children. However, none of these
studies explored the eﬀect of divorce on grandparental childcare.
One theoretical as well as methodological puzzle concerning the
“collective declining eﬀect” argument relates to the timing of divorce,
of the measurement of the outcome variable, and of the measurement of
the incidence of divorce. While there may be considerable time gap
between family dissolution and the measurement of the dependent
variable (e.g. grandparental childcare, in our case), analysts wanting to
test the “collective declining eﬀect” face a diﬃcult choice regarding the
best moment for the measurement of the contextual variable (i.e. incidence
of divorce, most of often in the form of the crude divorce rate,
CDR). CDR can refer to the period when the divorce occurred, or it can
refer to the period when the value of the outcome variable was ascertained.
Each option seems to highlight a diﬀerent element of the “collective
declining eﬀect” theory and is likely to aﬀect the outcome differently
in a multivariate statistical analysis. While CDR measurement
at the time of divorce is conceptually more strongly linked to the circumstances
immediately surrounding divorce (such as typical levels of
within-family conﬂict, or divorce-inﬂicted stigma), CDR measurement
referring to the period when the outcome variable was measured will
probably more strongly reﬂect social context at that later moment, such
as values, norms, and institutions governing the lives of divorcees and
their interaction with children and other kin. These circumstances can
inspire a change in inter-generational relationships regardless of the
conditions immediately preceding or following after divorce.
This consideration is of a particular importance in countries such as
Portugal or Belgium, where divorce rates developed dramatically
(Eurostat, 2013): divorcees who divorced at times of low divorce rates
probably experienced high levels of conﬂict and stigma (with serious
immediate implications for their lives and those of their children), but
present conditions of relative tolerance and prevailing emphasis on
post-divorce cooperation in the family network (i.e. new social institutions)
may help them overcome past conﬂict and reorganize patterns
of interaction with their ex-spouses and children. Thus, it remains
to be shown if the eﬀect of divorce on patterns of grandparental
childcare interacts at all (or more strongly) with past or present measures
of CDR. If the former is the case, our interpretation should focus
more on the change in circumstances surrounding the time of separation
(such as conﬂict and stigma). If the latter is the case, we should
shift our interpretation more towards current circumstances (such as
the more cooperative post-divorce arrangements in the family).
For this purpose of evaluating past levels of conﬂict and stigma (and
their eﬀect on intergenerational exchange), it may be useful to use a
CDR measure that would cover several years, not only the year of the
actual divorce. It is because this argument is conceptually rooted in the
process theory of divorce (Morrison & Cherlin, 1995; Sun, 2001), which
highlights a longer-term perspective on what happens before, during,
and after separation: conﬂict in the couple might begin several years
before actual separation/divorce. If children’s exposure to conﬂict
matters (and if the level of conﬂict varies by divorce rate), for instance,
it will be more adequately captured by a composite measure covering
several years surrounding family dissolution.
To summarize, we build on existing literature that has shown that
Z. Žilinčíková, M. Kreidl Advances in Life Course Research 38 (2018) 61–71
63
grandparental divorce decreases the odds of providing grandparental
childcare. We argue, however, that this eﬀect varies across countries. In
accordance with the declining divorce eﬀect hypothesis, we formulate
our central hypothesis as follows:
Hypothesis: the disruptive eﬀect of divorce on grandchild care
provision is weaker (i.e. closer to zero) in societies where family separation
is more common.
In addition, we want to explore if the variation in the disruptive
eﬀect of divorce is the same for any care and intensive care. It is a
common practice to diﬀerentiate any care and intensive care in this
ﬁeld (Hank & Buber, 2009) and it has been documented that their
correlates diﬀer to some degree (Hank & Buber, 2009). Furthermore, it
has been shown repeatedly, that patterns of intergenerational caregiving
vary signiﬁcantly across countries and regions (Hank & Buber,
2009; Igel & Szydlik, 2011): grandparental childcare is more common
(but typically less intensive) in Nordic countries, whereas in more traditional
southern countries childcare is, in general, less common, but
when it occurs, it tends to be more intensive. Since family dissolution
rates correlate also with the north-south family typology (as well as
with the strength of family values) examining any care and intensive
care separately is warranted.
4. Data, variables, and method
4.1. Data sources
The main data source is the Survey of Health, Ageing, and
Retirement in Europe (SHARE) from 2004, 2006, and 2011, i.e. waves
1, 2, or 4, respectively (wave 3 – also known as SHARELIFE – could not
be used as it employed a diﬀerent questionnaire focusing on the retrospective
mapping of life-course events). Survey data are complemented
by country-level variables (divorce rates, levels of familism,
and religiosity) taken from external sources (details are presented
below). SHARE is a cross-national panel survey targeting the 50+ noninstitutionalized
population in each country. Our sample consists of 18
European countries: Austria, Germany, Sweden, the Netherlands, Spain,
Italy, France, Denmark, Greece, Switzerland, Belgium, the Czech
Republic, Poland, Ireland, Hungary, Portugal, Slovenia, and Estonia.
We only utilized data from the ﬁrst data collection in each country.
A total of 49,688 respondents were interviewed in selected countries/waves.
Our sample was restricted to family respondents (i.e. those
who were asked questions about children and grandchildren).
Furthermore, the sample was limited to grandparents, i.e. respondents
with at least one grandchild (the “child of a child” in the terminology of
the SHARE questionnaire). In the ﬁrst two waves of SHARE, some
questions on children’s characteristics were asked only for a maximum
of four children. If a respondent (a grandparent) had more than four
children, several selection criteria were applied so that only four children
were reported on (with a preference for non-minor children, close
geographical proximity, and earlier year of birth). To make use of all
children’s characteristics we restricted our sample to these four children
and applied the same selection criteria for the children interviewed in
the fourth wave (if a country joined SHARE at a later stage and wave 4
happened to be the ﬁrst data collection in a given country).
We reshaped the initial dataset from wide to long format; thus,
every child of a grandparent constituted one line in the data matrix.
After omitting cases with missing or not applicable values in key variables
and applying the condition that the youngest grandchildren of a
particular child had to be younger than 16 years, our ﬁnal (“full”) dataset
consisted of 24,286 entries (“children of a child”) and 15,845
grandparents; this sample was used to investigate correlates on any
grandparent childcare provision (a smaller sample, however, was used
to investigate intensity of care if any care was provided; number of
grandchildren in this “reduced” sample is 11,404, number of grandparents
is 8991).
We see the data as hierarchically structured (and methods of multilevel
analysis are employed to reﬂect this clustering): multiple children
are nested within one primary respondent (grandparent) and care is
provided (and reported) with reference to the “child of the focal child”.
The child – or more precisely, the pair of the child and child of that
child – is treated as a level-1 observation, whereas grandparents are
level-2 observations. Grandparents are further nested in the countries
that are considered as level-3 units. We investigate the data employing
standard multi-level techniques that may incorporate variables measured
at all three levels.
4.2. Dependent variables
There are two questions on child care from which we derived our
dependent variables. First, the respondents were asked whether they
had provided care for a child or children of a particular child in the past
12 months: a binary indicator (0-no, 1- yes) of caregiving is, thus, our
ﬁrst dependent variable.1
Taken across all level-1 observations the full
sample (i.e. “children of children”) in our analysis, we see that 47 per
cent of them had been recipients of some grandparental childcare in the
previous year. This percentage varies from a low of 38 per cent (in Italy,
Portugal, and Spain, see Table A1 in the Appendix for details) to a high
of 56 per cent (in Belgium). A binary variable indicating whether or not
a particular child received any grandparental childcare in the year prior
to the interview is our ﬁrst dependent variable.
Further, if the respondent answered positively, she/he was asked
about the frequency of care using a 4-point scale: almost daily, almost
every week, almost every month, and less often. We dichotomized the
frequency scale to diﬀerentiate between intensive (“almost daily” and
“almost every week”) and non-intensive care.2
Among grandchildren
receiving any care, one half received intensive care (see Table A1 in the
Appendix). Prevalence of intensive care among care recipients ranges
from 21 per cent in Denmark and 25 per cent in Sweden to 77 per cent
in Italy, 69 per cent in Portugal and Greece, and 67 per cent in Poland
(see Table A1 in the Appendix).
4.3. Main explanatory variable: marital status of grandparents
The marital status of the grandparent at the time of interview is our
main explanatory variable. We follow the recent trend and employ a
detailed measure of marital status (see e.g. Albertini & Garriga, 2011;
Uhlenberg & Hammill, 1998; Danielsbacka & Tanskanen, 2016) using
four categories3
(1) married or in a registered partnership (with the condition
that spouses live together), (2) divorced or separated, (3) widowed,
and (4) never married. Unfortunately, information about the timing of
divorce and multiple partnership transitions was not recorded in the
SHARE questionnaire in our respective waves, and thus, we were not
1
The exact wording of the question reads as follows: “During the last twelve
months, have you regularly or occasionally looked after (your grandchild/your
grandchildren) without the presence of the parents?”
2
We chose to use two dependent variables, as some recent research suggests
that a simple binary indicator of care provision may be insuﬃcient to capture
the complexities of inter-generational caregiving (see e.g. Hank & Buber, 2009).
3
Marital status typically used to be represented by two categories (such as
married/unmarried or with/without a coresidential partner) in research on
intergenerational contact/caregiving (Fuller-Thomson & Minkler, 2001;
Silverstein & Marenco, 2001). Often, the primary focus was on living arrangements
such as living with a partner/living without a partner (Baydar &
Brooks-Gunn, 1998; Hank & Buber, 2009; Knudsen, 2012; Luo et al., 2012;
Musil et al., 2006). King (2003) and Jappens and Van Bavel (2012) also worked
with two categories but distinguished between ever-divorced and never-divorced
grandparents. Similarly, Geurts et al. (2015) distinguished between divorced
or not divorced grandparents, while White (1992) controlled for being
divorced or remarried. This prevailing simpliﬁcation of the measurement is
increasingly unsatisfactory, considering that contemporary grandparents have
more complex family and partnership histories than previous generations
(Manning & Brown, 2011).
Z. Žilinčíková, M. Kreidl Advances in Life Course Research 38 (2018) 61–71
64
able to employ a more nuanced categorization of the marital/partnership
status (e.g. diﬀerentiate continuously married and re-married re-
spondents).4
Nevertheless, the current categorization of marital status
enables us to diﬀerentiate currently divorced grandparents from other
marital statuses and provide a lower bound estimate of the true disruptive
eﬀect of divorce: we believe that re-married grandparents
would – on average – provide grandchild care less frequently and would
thus drive the average frequency of caregiving among presently married
grandparents down (for further discussion see Conclusion), close to
the caregiving pattern among divorced grandparents.
The descriptive results (presented in Table 1) show that approximately
68 percent of all grandparents in the full sample were married,
11 percent divorced or separated, almost 20 percent widowed, and
slightly below 2 percent never married (percentage distribution of
marital status in the “reduced” sample of grandparents providing some
childcare is also shown in Table 1, last column). The proportion of divorced
individuals diﬀered across countries, with the largest proportion
of divorced grandparents in countries with traditionally high divorce
rates (19% in Denmark and Estonia and 18% in the Czech Republic
when measured in the “full” sample) and the smallest proportion in
countries where divorce is quite rare (2% in Italy and 3% in Spain). A
detailed percentage distribution of marital status by country is reported
in Table A1 in the appendix for both the “full” and the “reduced”
samples.
4.4. Level-1 and level-2 control variables
The analysis employs independent variables that characterize (1)
the grandparents, and (2) the children (and by extension grandchildren),
thus allowing for covariates at all levels to be controlled.
While some of these variables require little explanation, the inclusion of
some others should perhaps be more explicitly justiﬁed. Within each
group of covariates, we ﬁrst discuss variables that appear to be exogenous
with respect to the divorce-childcare relationship and then we
turn our attention to other covariates, which are more likely to function
as mediating variables. The former group is utilized in the main analysis,
whereas the latter is only employed in supplementary models
reported in the appendices.5
The distribution of the independent variables
is reported in Table 1 (Table A2 presents their distribution according
to marital status).
Grandparents’ exogenous characteristics include: sex, age (3 categories:
50–59, 60–69, 70+; some earlier analyses suggested that the
eﬀect of age is non-linear and that grandparents are most likely to
provide childcare at around the age of 67 (Baydar & Brooks-Gunn,
1998)), and education (4 categories: none or primary, lower secondary,
upper- or post-secondary, tertiary; Baydar and Brooks-Gunn (1998) as
well as Silverstein and Marenco (2001) found that better educated
grandparents are more likely to provide childcare. Controlling for
education is especially important given historical changes in the socioeconomic
gradient of divorce (Härkönen & Dronkers, 2006).
The following variables are also controlled in the complementary
models: subjective perception of health (a dichotomous variable - excellent
or good vs. fair or poor; a grandparent’s poor health can reduce
the probability of providing childcare (Hank & Buber, 2009)), total
number of grandchildren (the more sets of grandchildren a grandparent
has, the lower the probability of having frequent contact with a particular
set (Uhlenberg & Hammill, 1998)), employment (3 categories:
non-working, working, and unemployed; working grandparents are –
all else being equal – less likely to provide care, see Hank & Buber,
2009; Lakomý & Kreidl 2015).
From the characteristics of children, we utilized child’s sex as an
exogenous variable and included it in the main analysis. The rest of the
characteristics were employed in complementary analyses: child’s employment
status (working, non-working, and unemployed; full-time
employment of mothers is positively associated with the probability of
grandparental childcare being provided (Vandell et al., 2003; Hank &
Buber, 2009)), and age of the youngest child of a particular child (categories
0–2, 3–5, 6–10, 11–16 years; the likelihood of grandparental
childcare provision also declines with the increasing age of the
youngest grandchild in a given set (Hank & Buber, 2009; Silverstein &
Marenco, 2001)). Child’s marital status is measured using the same four
categories utilized for grandparents (married, divorced/separated, widowed,
and never married).
Table 1
Descriptive statistics of variables used in the analysis. Selected SHARE countries
in.2004–2011.
Level-3 variables
Crude Divorce Rate (mean before standardization) 0.27
Familialism (mean before standardization) 2.04
Religiosity (mean before standardization) 0.46
N 18
Level-2 variables Any care Intensive care
Respondent male 42.7 % 39.3 %
Respondent’s marital
status
Married or registered
partnership
68.1 % 73.1 %
Divorced or separated 10.7 % 9.9 %
Widowed 19.6 % 15.8 %
Never married 1.6 % 1.2 %
Age 50-59 29.2 % 34.8 %
60-69 41.5 % 45.2 %
70+ 29.3 % 20.1 %
Employment status Out of the labor force 74 % 70.3 %
Employed 22.3 % 25.7 %
Unemployed 3.8 % 4 %
Education None or primary 29 % 25.5 %
Lower, upper or post-
secondary
52.4 % 54.1 %
Tertiary 18.6 % 20.4 %
Health Excellent or good 57.4 % 62.7 %
Fair or poor 42.6 % 37.4 %
Number of grandchildren
(mean)
3.9 3.6
Age of the youngest
grandchild
0 to 2 25.8 % 29 %
3 to 5 22.3 % 28 %
6 to 10 28.3 % 29.4 %
11 to 16 23.6 % 13.6 %
N 15,845 8991
Level-1 variables Any care Intensive care
Child´s marital status Married or registered
partnership
81.5 % 81.3 %
Divorced or separated 8.1 % 7.6 %
Widowed 0.6 % 0.5 %
Never married 9.8 % 10.6 %
Child male 47.9 % 42.6 %
Child´s employment
status
Employed 81.4 % 81.7 %
Out of the labor force 13.3 % 13.2 %
Unemployed 5.3 % 5.1 %
N 24,286 11,404
4
Merging responses from the standard SHARE questionnaire with data from
the retrospective SHARELIFE interview proved unfeasible due to a high proportion
of missing responses resulting from panel attrition.
5
Several covariates might be seen to play a dual role in determining
grandparental childcare (for instance grandparental employment status may
constrain caregiving, but can also be changed when a grandchild is in need of
care, see e.g. Lakomý & Kreidl, 2015). It is not our goal to adjudicate arguments
about the precise position of each covariate in the model, we simply present
several diﬀerently speciﬁed models to show that our results are robust vis-à-vis
partial model re-speciﬁcation.
Z. Žilinčíková, M. Kreidl Advances in Life Course Research 38 (2018) 61–71
65
4.5. Level-3 (country) variables
Three variables – crude divorce rate (CDR), familism, and religiosity
– are used to capture important country-level characteristics that might
shape patterns of inter-generational interaction. The main country-level
explanatory variable is crude divorce rate, which was proposed by
Albertini and Garriga (2011) to be the centerpiece of the “collective
declining eﬀect hypothesis” (see Section 3 above for an explanation),
and, it has been argued, divorce eﬀects should vary systematically with
the incidence of divorce in society. Thus, CDR is also used in interactions
with the individual-level measure of divorce. We use several alternative
CDR deﬁnitions all of which are based on the data from
Eurostat (2013). First, CDR corresponds to the year of SHARE data
collection (i.e. the year when the outcome variable was measured).
When SHARE interviews spanned a two-year interval in a given
country, CDR for the ﬁrst year was taken. However, for theoretical
reasons explained above, we also use two other CDR measures that refer
to earlier periods preceding the measurement of the outcome variable.
Since we were not able to use the standard SHARE database6
data to
ﬁnd out the exact year of respondents’ divorce for a signiﬁcant proportion
of our sample (mostly due to panel attrition), we can only approximate
CDR around the most typical divorce years using the following
procedures:
a For all SHARELIFE respondents, who belong to our sample of divorced
respondents, we computed the mean and median year of
divorce (both were 1988). Then, we use a country’s CDR for 19887
in our models. The correlation of this CDR measure with CDR
measured at the interview is 0.79.
b For all SHARELIFE respondents, who belong to our sample of divorced
respondents, we ascertained the range of divorce years. We
found out that 95% of all divorces took place between 1970 and
2005 and we computed the average CDR for this entire period. The
correlation of this mean CDR variable with CDR measured at the
interview is 0.75.
Some of our models also contain two other country-level controls,
namely familialism and religiosity. They were chosen to capture important
diﬀerences in prevailing patterns of family life in Europe, which
are reﬂected, for instance, in the distinction between weak and strong
family systems (e.g. Reher 1998; Kohli et al., 2005; Bolin et al., 2008;
Hank 2007; Kalmijn 2008). Reher (1998) associates weak family ties
with Central and Northern Europe (for example Scandinavia, Great
Britain, Benelux, Germany, and Austria), and strong family ties with
Mediterranean countries (see also Kohli et al., 2005; Hank, 2007).
Strongly familialistic countries are more conservative, have lower rates
of divorce or extramarital pregnancy (Reher 1998), and are also associated
with low gender equity in the family and in public provisions for
the family (Kohli et al., 2005). A similar gradient is also associated with
diﬀerent patterns of childcare. Hank and Buber (2009) found that the
probability of grandparents providing care for their grandchildren is
higher in northern Europe (the highest probability was found in Denmark,
the Netherlands, France, and Sweden) and lowest in Mediterranean
countries (Italy and Spain). However, when the intensity of care
was considered, the relationship was reversed: grandparents in
southern Europe tended to provide very intensive care. Moreover,
Kalmijn (2010) showed that some eﬀects of divorce vary with the level
of familialism (members of the family network mobilize, for instance, to
help overcome some negative consequences of family dissolution) and
religiosity (the negative eﬀect of divorce on well-being is stronger in
traditional societies, but only for religious individuals).
Measures of familism and religiosity were derived from European
Social Surveys from the years 2004–2010 (either ESS Round 2, 2004;
ESS Round 3, 2006; or ESS Round 5, 2010). We chose the ESS year that
was closest to the year of SHARE data collection in a given country.
Similarly to Kalmijn (2010), we measured familism as the percentage of
unmarried people aged 18–50 living with parents, and religiosity as the
percentage of Christians attending church at least once a month. All
country-level variables were standardized (the mean value was subtracted
and the result was divided by the standard deviation) in order to
make the estimated coeﬃcients in our analysis more readily interpretable.
Descriptive statistics of all level-3 variables are presented in
Table A1 in the Appendix.
4.6. Analytical strategy
We employ random-intercept three-level hierarchical linear models
with binary response variables. We chose to estimate linear probability
models (rather than multi-level logistic regressions) for computational
feasibility, stability of estimates, ease of interpretation, and to avoid
problems when comparing eﬀect sizes across diﬀerently speciﬁed logistic
models (Mood 2010). We estimated the parameters of these
models using STATA 14 M P (using the “mixed” procedure).
Our analysis of both dependent variables relied on the same procedure.
We started by utilizing all above-mentioned level-1 exogenous
explanatory variables in an additive model. Then we proceeded to add
the interaction between respondent’s (grandparent’s) divorced status
and current crude divorce rate, which corresponds to the “collective
declining eﬀect hypothesis” as proposed by Albertini and Garriga
(2011). We then evaluated this interaction using both the criteria of
classical statistical inference (likelihood ratio test) and Bayesian statistics
(AIC). We evaluated the nature of the (statistically) signiﬁcant
interactions and determined the size of the divorce eﬀect for various
levels of CDR. All models were estimated jointly for men and women;
however, we also report the results of some robustness checks to conﬁrm
whether (and to what extent) parameter estimates are similar for
both genders.
5. Results
5.1. Models of any grandparental childcare
First, we model a binary variable to show which grandparents
provide (any) care for grandchildren. Model 1 (estimated parameters
are shown in Table 2) shows that the probability of care provision, with
all independent variables held constant, is 63 per cent. Divorce is
clearly associated with lower odds of grandparental caregiving. The
probability of grandparental childcare being provided – net of other
variables in the model – is more than 11 percentage points lower among
divorced grandparents than among married grandparents. The other
parameters of Model 1 are consistent with earlier research. For instance,
the odds of grandparental childcare being provided decline with
increasing age (by 22 percentage points in the 70+ category) and increase
with education (grandparents with tertiary education are by 12.3
percentage points more likely to provide care than grandparents with
elementary education). Care is provided more often to children of
daughters than to children of sons – when the child is male, the probability
of his children receiving grandparental childcare is lower by
almost 8.9 percentage points. Current CDR level has – net of other
predictors – only a very weak and insigniﬁcant eﬀect.
When we add the interaction between grandparental divorce and
CDR into Model 1 to test our main hypothesis, we obtain Model 2
(goodness of ﬁt statistics for both models are presented in Table 2).
Statistical comparison of these two models reveals that the interaction
6
Wave three of SHARE (known as SHARELIFE) was a survey that mapped
retrospectively life course event of SHARE respondents. It was conducted in
2008-2009 in SHARE countries, typically as the third round of survey inter-
views.
7
The only exception to this rule was Ireland, where divorce was not legal
until 1997; we took CDR for 1997 instead of 1988.
Z. Žilinčíková, M. Kreidl Advances in Life Course Research 38 (2018) 61–71
66
should not be omitted from Model 2 (L2
for comparing these two
models is 4.67 with 1° of freedom, which implies p = 0.0306, see
Table 2). AIC for Model 2 is lower (32,450.78) than for Model 1
(32,453.45) also indicating superiority of Model 2. Thus, on the basis of
both the criteria of classical inference and Bayesian model comparison,
we should conclude that the size of the divorce eﬀect varies by current
divorce rate. The estimated parameters of Model 2 are presented in
Table 2. We see that the main eﬀect of divorce is negative (i.e. being
divorced reduces the probability of grandparental childcare provision),
while the interaction between divorce and CDR is positive, thus indicating
that the negative eﬀect of being divorced is reduced (i.e. closer
to zero) at higher CDR levels. Whereas divorce seems to reduce the
probability of grandparental childcare provision by 12.6 percentage
points at the average CDR level, this eﬀect is reduced to 9.4 percentage
points when CDR increases by one standard deviation (see Fig. 1). On
the other hand, the divorce eﬀect increases to 15.8 percentage points
when CDR is one standard deviation below its mean (this corresponds
to an increase of 25 percent).
5.2. Models of intensive grandparental childcare
Our analysis of intensive grandparental childcare follows the same
logic as the analysis of any care presented above; we again present and
compare 2 multi-level linear probability models, but these are estimated
on a smaller sample of those grandparents, who provided some
care in the last year (there are 8991 such grandparents – i.e. level-2
units – in our sample with 11,404 grandchildren). The estimated
parameters of these models are presented in Table 3. Model 3 shows
that when all independent variables are held constant, the probability
Table 2
Estimated parameters of selected multi-level linear probability models of any
grandparental childcare provision. Selected SHARE countries.2004–2011.
M1 M2
Level-3 variables
Crude Divorce Rate −0.002 −0.004
Level-2 variables
Respondent male −0.093*** −0.093***
Respondent’s marital status (married is reference
category)
Divorced −0.114*** −0.126***
Widowed −0.097*** −0.097***
Never married −0.136*** −0.136***
Age (50-59 is reference category)
60-69 −0.059*** −0.059***
70+ −0.222*** −0.222***
Education (none or primary is reference category)
Lower, upper or post-secondary 0.062*** 0.062***
Tertiary 0.123*** 0.123***
Level-1 variables
Child male −0.089*** −0.089***
Interaction
Divorced x CDR 0.032*
Constant 0.632*** 0.632***
Goodness-of-ﬁt statistics
Log-likelihood −16,212.73 −16,210.39
d.f 14 15
p-value < 0.00005 < 0.00005
AIC 32,453.45 32,450.78
Contrast M2-M1
Likelihood-ratio test 4.67
d.f. 1
p-value 0.0306
Note: Number of level-3 units (countries) = 18, level-2 units (grandparents)
= 15,845, number of level-1 units (grandchildren) = 24,286.
Fig. 1. Estimated net eﬀect of divorce on the probability of any grandparental childcare by level of crude divorce rate. Selected SHARE countries.2004–2011.
Note: estimates are based on Model 2.
Table 3
Estimated parameters of selected multi-level linear probability models of intensive
grandparental childcare provision. Selected SHARE coun-
tries.2004–2011.
M3 M4
Level-3 variables
Crude Divorce Rate −0.073* −0.070*
Level-2 variables
Respondent male −0.047*** −0.047***
Respondent’s marital status (married is reference
category)
Divorced −0.063*** −0.048*
Widowed −0.017 −0.017
Never married 0.010 0.010
Age (50-59 is reference category)
60-69 0.001 0.001
70+ −0.034* −0.033*
Education (none or primary is reference category)
Lower, upper or post-secondary −0.029* −0.029*
Tertiary −0.064*** −0.064***
Level-1 variables
Child male −0.053*** −0.053***
Interaction
Divorced x CDR −0.039+
Constant 0.609*** 0.609***
Goodness-of-ﬁt statistics
Log-likelihood −7,467.56 −7,465.84
d.f 14 15
p-value < 0.00005 < 0.00005
AIC 14,963.12 14,961.69
Contrast M2-M1
Likelihood-ratio test 3.43
d.f. 1
p-value 0.0640
Note: Number of level-3 units (countries) = 18, level-2 units (grandparents)
= 8991, number of level-1 units (grandchildren) = 11,404.
Z. Žilinčíková, M. Kreidl Advances in Life Course Research 38 (2018) 61–71
67
of intensive care provision is 61 per cent. Divorced grandparents are by
6.3 percentage points less likely to provide intensive childcare than
married grandparents. Interestingly, married, widowed, and never
married grandparents are equally likely to provide intensive childcare.
We furthermore observe a signiﬁcant inﬂuence of gender – grandfathers
tend to provide care less frequently than grandmothers and intensive
grandchild care is provided more often to the children of daughters
than to the children of sons. The eﬀect of age in Model 3 agrees with the
ﬁndings of the previous models of any care provision: intensive care is –
everything else being equal – by 3.4 percentage points less likely among
grandparents aged over 70. The eﬀect of education goes, however, in
the opposite direction: in Model 3 we see that higher education associates
with a lower probability of intensive grandchild care. For instance,
the probability that a grandparent with tertiary education will
provide intensive grandchild care is by 6.4 percentage points lower
than among grandparents with elementary education; the education
eﬀect in Model 3 most likely reﬂects stronger labor market attachment
and resulting constraints among educated grandparents.
Model 4 (see Table 3) adds the interaction between grandparental
divorce and crude divorce rate. A comparison of Model 3 and Model 4
(see lower panel of Table 3) gives only weak evidence that the interaction
is statistically signiﬁcant. When we compare these two models
using criteria of classical inference, we obtain L2
= 3.43 with 1 ° of
freedom (p = 0.064). AIC suggests that we should prefer the more
parsimonious model, i.e. Model 3 (AIC for Model 4 is only 1.43 points
lower than AIC for Model 3; see Table 3).
Since the interaction is borderline statistically signiﬁcant, we present
estimated parameters of Model 4 (in Table 3) to also assess its
substantive signiﬁcance. We can see that – at average CDR – divorce
reduces the probability of intensive caregiving by 4.8 percentage
points. However, the divorce eﬀect increases in size with growing CDR.
For instance, one standard deviation increase in CDR results in an increase
of 3.9 percentage points in the size of the negative divorce eﬀect,
which is an increase of about 81 percent. When CDR declines one
standard deviation, the divorce eﬀect is trivial (only 0.9 percent). These
comparisons are visualized in Fig. 2 and suggest that the divorce*CDR
interaction is substantively important.
5.3. Diﬀerences in the eﬀects of divorce by gender of the grandparent
As numerous previous studies, also our results (not shown) conﬁrmed
that the provision of childcare is more restricted by divorce for
men than for women. Our reading of the literature also suggests that
trends in the size of the disruptive eﬀect of divorce on grandparental
childcare might diﬀer by grandparental gender. Divorced non-custodial
parents (typically fathers) tend to stay in closer contact with children in
more recent cohorts than used to be the case in cohorts born in the ﬁrst
half of the 20th
century (Gähler & Palmtag, 2015). Therefore, patterns
of grandparental childcare provided by divorced grandparents may
converge. We made an eﬀort to test for three-way interactions between
grandparental gender, grandparental divorce, and incidence of divorce
in society (CDR) to scrutinize this assertion. We found no (statistically)
signiﬁcant interactions. Yet, the pattern in estimated parameters suggested
that the divorce eﬀect might change with CDR diﬀerently in
grandfathers and grandmothers (the divorce*CDR interaction appears
to be stronger among women than among men). As it was quite diﬃcult
to obtain stable robust estimates (with only 18 level-3 units, i.e.
countries, and a complex model structure), it is little surprise that the
results were insigniﬁcant.
5.4. Sensitivity tests
We present several sensitivity tests to show the robustness of our
statistical models and to answer ancillary research questions. Our discussion
in this section focuses on three issues: 1) inclusion of other
level-1 and level-2 covariates, 2) inclusion of additional level-3 covariates,
and 3) utilization of alternative divorce rate measures.
First, we explore how an expansion of the set of level-1 (child) and
level-2 (grandparent) covariates alters the results. We add the following
variables to the model: respondent employment status, respondent’s
health, number of grandchildren, age of the youngest grandchild,
child’s marital status, and child’s employment status. We add these
variables to Models 1–4 to create Models 5–8 (see Table A3 in the
Appendix for parameter estimates). Then we contrast Models 5 and 6 to
see if grandparental divorce interacts with CDR in determining any
grandparental childcare. Similarly, we contrast Models 7 and 8 to test
the interaction in modelling intensive childcare. Comparison of Models
5 and 6 suggest that the divorce*CDR interaction should not be omitted
from Model 6 by criteria of classical inference (L2
= 4.18 with 1 d.f.,
p = 0.04). Also AIC favors Model 6 over Model 5 (respective AIC values
are 31,216.76 and 31,218.94, i.e. the diﬀerence is 2.18, which agrees
with the likelihood ratio test). Similarly, comparison of Models 7 and 8
conﬁrms – albeit less decisively – that grandparental divorce and CDR
interact in determining intensive grandparental childcare (L2
= 3.46
with 1 d.f., p = 0.06; diﬀerence in AIC for Models 7 and 8 is 1.46, see
Table A3 in the Appendix). Thus, we conclude that we can conﬁrm the
existence of the divorce*CDR interaction even with a much larger set of
covariates.
Second, we expand our models to include two additional macrolevel
controls – level of familialism and religiosity in society.
Familialism and religiosity are obviously correlated with divorce rate
and it is not clear which of the three variables causally precedes the
Fig. 2. Estimated net eﬀect of divorce on the probability of intensive grandparental childcare by level of crude divorce rate. Selected SHARE countries.2004–2011.
Note: estimates are based on Model 4.
Z. Žilinčíková, M. Kreidl Advances in Life Course Research 38 (2018) 61–71
68
other two (and it is actually likely that they mutually reinforce each
other in the course of the modernization process). We add familialism
and religiosity to Models 2 (model for any care) and 4 (model for intensive
care) and create Models 9 and 12 (see Table A4 in the
Appendix). In both models, the main eﬀect of divorce as well as the
divorce*CDR interaction remain almost unchanged. While the
divorce*CDR interaction was 0.032 in Model 2, it is 0.033 in Model 9
(both eﬀects are statistically signiﬁcant at the 0.05 level). Similarly, the
interaction was -0.039 (signiﬁcant at the 0.1 level) in Model 4 and it is
-0.042 (signiﬁcant at the 0.05 level) in Model 12 (see Table A4 in the
Appendix). So, the eﬀect of divorce changes with CDR level even when
we control for country level of familialism and religiosity.
Third, we employ alternative deﬁnitions of CDR referring to historical
periods preceding the survey interview. CDR from 1988 is used
in Models 10 and 13 (see Table A4), which are otherwise identical to
Models 2 and 4, and mean CDR for the 1970–2005 period is used in
Models 11 and 14. When the CDR deﬁnition changes, the divorce*CDR
interaction weakens and is no longer statistically signiﬁcant (see Table
A4); it is 0.018, 0.017, 0.008, and 0.012 in Models 10, 11, 13, and 14,
respectively. Whereas it is somewhat problematic to draw robust conclusions
from these alternative CDR measures (because they only very
roughly approximate the timing of divorce among divorced grandparents),
we might tentatively conclude that the eﬀect of grandparental
divorce on grandparental childcare seems to change with current CDR
levels (i.e. divorce rates that reﬂect the situation when the outcome
variable is measured) more than it does with historical CDR measures
(which describe the situation around the time of divorce). While the
former level-3 variable is associated with existing social institutions
(e.g. contemporary social institutions and prevailing family organization
in divorced families), the latter is more likely to relate to the typical
level of conﬂict and the strength of the stigma that existed when
grandparents divorced.
6. Conclusions and discussion
Our analyses show that the marital status of grandparents is a signiﬁcant
predictor of grandchild care provision. Consistent with previous
research, we found that divorced or separated grandparents tend
to provide both any and intensive care for their grandchildren with a
lower probability than married grandparents. We provide evidence that
divorce in the oldest generation is, in general, associated with the disruption
of intergenerational ties, and might deprive the middle generation
of an important source of childcare. Using multilevel data for 18
countries, we show that the eﬀect of divorce varies signiﬁcantly with
the overall incidence of divorce in a country. Consistent with our expectations,
the disruptive eﬀect of divorce on provision of any grandparental
childcare is signiﬁcantly weaker when divorce occurs more frequently. In
fact, it declines by almost 30 per cent when crude divorce rate increases
by one standard deviation. Yet, contrary to our expectations, the eﬀect
of divorce on intensive grandparental childcare seems to follow an
opposite trend – it becomes stronger (i.e. more negative) when divorce
is more common.
A weaker divorce eﬀect on any care provision under conditions of
higher divorce rates may result from three circumstances. 1) Family
dissolution is associated with less stigma and consequently, inter-generational
relationships continue to function more smoothly after divorce.
Good parent-child relationships are more likely to persist and
grandparents are therefore more likely to interact with and care for
grandchildren. 2) Patterns of post-divorce interaction that involve
regular contact among family members become more institutionalized
(Thompson & Amato, 1999) and non-custodial parents lose contact with
their children to a lesser degree (see e.g. Gähler & Palmtag, 2015).
Regular intergenerational contact is, as we know from earlier research,
an important prerequisite for grandparental childcare provision
(Dykstra & Fokkema, 2011). 3) When divorce is more common, a higher
proportion of low-conﬂict families split up (see e.g. Amato & HohmannMarriott,
2007; Gähler & Palmtag, 2015), which also creates greater
potential for uninterrupted inter-generational interaction, including the
provision of inter-generational care, be it grandparental childcare (as in
our analysis) or other kinds of care.
While the “collective declining eﬀect” operate with all 3 abovementioned
arguments, our ancillary analyses indicated that the divorce
eﬀect changes with current CDR levels rather than with past CDR levels.
This ﬁnding (despite all reservations that we have regarding the measurement
of historical CDR levels and their alignment with the timing of
grandparental divorce) suggests that especially current social institutions
(rather than past levels of stigma and conﬂict) governing postdivorce
interaction between family members (ex-spouses, parents and
children, but also involvement of other relatives) are responsible for
observed trends in the size of the divorce eﬀect. If patterns of postdivorce
interaction become more strongly institutionalized in society,
they may initiate change in inter-generational interaction even among
grandparents who have been divorced for a long time and experienced
stigma and conﬂict at the time they were divorcing.
Our results indicating a declining negative eﬀect of divorce on intergenerational
caregiving appear consistent with other analyses that
have examined changes in the divorce eﬀect over time or variations in
the size of the divorce eﬀect across societies with varying divorce rates.
(Kalmijn, 2008; Kalmijn, 2010; Dronkers & Härkönen, 2008). In short,
we conﬁrm the “collective declining eﬀect hypothesis” as proposed by
Albertini and Garriga (2011) in the case of grandparental involvement
in childcare. Yet, it needs to be pointed out that the negative eﬀect of
family dissolution only appears to decline for some outcomes, but not
for others. For instance, Kreidl, Štípková, and Hubatková (2017) have
shown that the negative eﬀect of parental breakup on children’s probability
of completing a college education has increased over time and is
stronger when divorce is more common.
Our analysis of intensive grandparental childcare shows that the
eﬀect of divorce on intensive childcare seems to increase under conditions of
high family dissolution rates. We interpret this result with caution, as the
statistical signiﬁcance of the eﬀect was relatively low. Nevertheless, it is
a puzzling result. It suggests that the “collective declining eﬀect” thesis
does not apply to intensive grandparental childcare (here deﬁned as
care provided at least once a week without the presence of the child’s
parents) to the same extent as it does to any care provision. This might
result from the fact that intensive care is much more demanding logistically
and requires a higher level of personal resolution and more
favourable circumstances (i.e. in terms of geographic proximity, family
complexity, labor market attachment etc.). For instance, divorce correlates
rather strongly with values and behaviors that engage people
outside of the original family (this evidenced, for instance, by higher
repartnering rates, stronger labor market attachment as well as more
individualistic values of the divorcees that partly result from the divorce
experience, see e.g. Amato & Booth, 1991; Cunningham &
Thornton, 2006) and these engagements create additional obstacles for
and reduce interest in intensive caregiving. Thus, the observed trend
might stem from a complex re-organization of people’s post-divorce
lives and also from increased levels of post-divorce individualism and
lower commitment to the family. When family dissolution is common
and tolerance towards non-traditional family forms prevails, individual
experience of divorce (and the resulting shift towards more individualistic
attitudes) may result in a stronger disengagement from the
family and thus represent a particularly strong impediment to intensive
intergenerational caregiving. However, it is also possible that our CDR
measure captures some additional cross-country diﬀerences in the
analysis of intensive grandparental care. While we made an eﬀort to
measure other relevant macro-level variables, given the small sample of
countries, these controls are far from perfect.
To the extent that we can predict trends over time on the basis of a
cross-sectional analysis, our results indicate that increasing divorce
rates (which have been witnessed in many western countries over the
last few decades, see e.g. Spijker, 2012; Kennedy & Ruggles, 2014) may
Z. Žilinčíková, M. Kreidl Advances in Life Course Research 38 (2018) 61–71
69
not per se imply that grandparents are providing childcare less frequently
to their grandchildren. In fact, it appears that families cope
with divorce more successfully when family dissolution is more
common. Consequently, grandparental caregiving will likely continue
to be a valuable resource for both children and grandchildren, not to
speak of the satisfaction it brings to grandparents themselves (Hughes
et al., 2007; Pruchno & McKenney, 2002). They will, not, however,
become more frequently available for intensive grandchild care. Our
results indicate rather the contrary – intensive care is most likely to be
less commonly provided under conditions of elevated family dissolution
rates.
While we identiﬁed no diﬀerence in the interaction between being
divorced and CDR by gender, we still believe that this is a promising
topic for further research. We suggest that it should be revisited when
more data accumulate. In particular, a larger sample of countries is
desirable in order to obtain more robust estimates of the three-way
interaction between divorce, CDR, and gender.
There are some limitations of our study which stem from the
availability of the measures in the SHARE data. First, despite improvements,
our indicator of marital status is imperfect. SHARE only
measures marital status at the time of the interview; thus, we can only
distinguish currently divorced and currently married respondents. The
latter group, therefore, combines several categories of respondents including
continuously married and re-married (after divorce) respondents.
For this reason, we believe that our estimate of the disruptive
eﬀect of divorce upon intergenerational caregiving is only a
lower bound estimate of the true eﬀect. This is because parental remarriage
tends to intensify conﬂict over personal issues between adult
children and their parents and lowers the probability of support exchange
(Schenk & Dykstra, 2012). If we were able to separate continuously
married and re-married respondents, the diﬀerence in the
frequency of intergenerational caregiving between the former group
and divorced respondents would very likely be signiﬁcantly larger than
is reported in this paper. Another promising extension of this research
would diﬀerentiate between married and cohabiting grandparents.
Whereas the eﬀect of adult children’s union status (married, cohabiting,
single) upon contact with parents has already been examined (Yahirun
& Hamplová, 2014) and signiﬁcant diﬀerences have been found, an
analogous analysis by parental union status has not been carried out.
Second, we were not able to retrieve suﬃciently detailed information
about the timing of divorce as this was not part of the standard
SHARE questionnaire. By not diﬀerentiating the timing of divorce,
which is shown to be an important covariate in the intensity of the
disruption of intergenerational support (Seltzer, 1991), the results represent
an averaged eﬀect of family dissolution on intergenerational
support. Future research may explore the eﬀect of more complex
partnership trajectories on intergenerational support. However, this
cannot be done – as we have pointed out – without losing a considerable
number of observations. This shortcoming may be overcome in the
future as the SHARE database increases in size and more respondents
are added.
Acknowledgement
The writing of this article was ﬁnancially supported by the Czech
science foundation (project num. 13-34958S); ﬁnal revisions received
ﬁnancial support also from the Grant Agency of Masaryk University
(project num. MUNI/E/1313/2017).
Appendix A. Supplementary data
Supplementary material related to this article can be found, in the
online version, at doi:https://doi.org/10.1016/j.alcr.2018.08.003.
References
Aassve, A., Arpino, B., & Goisis, A. (2012a). Grandparenting and mothers’ labour force
participation: A comparative analysis using the generations and gender survey.
Demographic Research, 27, 53.
Aassve, A., Meroni, E., & Pronzato, C. (2012b). Grandparenting and childbearing in the
extended family. European Journal of Population/Revue européenne de Démographie,
28(4), 499–518.
Ahrons, C. R. (2004). We’re still family: What grown children have to say about their parents’
divorce. Harper Collins.
Albertini, M., & Garriga, A. (2011). The eﬀect of divorce on parent–Child contacts:
Evidence on two declining eﬀect hypotheses. European Societies, 13(2), 257–278.
Amato, P. R. (1993). Children’s adjustment to divorce: Theories, hypotheses, and empirical
support. Journal of Marriage and the Family, 55(1), 23–38.
Amato, P. R. (2000). The consequences of divorce for adults and children. Journal of
Marriage and the Family, 62(4), 1269–1287.
Amato, P. R., & Booth, A. (1991). The Consequences of Divorce for Attitudes Toward
Divorce and Gender Roles. Journal of Family Issues, 12(3), 306–322.
Amato, P. R., & Booth, A. (1996). A prospective study of divorce and parent-child relationships.
Journal of Marriage and the Family, 356–365.
Amato, P. R., & Hohmann-Marriott, B. (2007). A comparison of high‐and low‐distress
marriages that end in divorce. Journal of Marriage and the Family, 69(3), 621–638.
Amato, P. R., Loomis, L. S., & Booth, A. (1995). Parental divorce, marital conﬂict, and
oﬀspring well-being during early adulthood. Social Forces, 73(3), 895–915.
Aquilino, W. S. (1994). Impact of childhood family disruption on young adults’ relationships
with parents. Journal of Marriage and the Family, 295–313.
Barnett, M. A., Scaramella, L. V., Neppl, T. K., Ontai, L., & Conger, R. D. (2010).
Intergenerational relationship quality, gender, and grandparent involvement. Family
Relations, 59(1), 28–44.
Baydar, N., & Brooks-Gunn, J. (1998). Proﬁles of grandmothers who help care for their
grandchildren in the United States. Family Relations, 47(4), 385–393.
Bengtson, V. L. (2001). Beyond the nuclear family: The increasing importance of multigenerational
bonds. Journal of Marriage and the Family, 63, 1–16.
Bengtson, V. L., Biblarz, T. J., & Roberts, R. E. L. (2002). How families still matter: A
longitudinal study of youth in two generations. New York: Cambridge University Press.
Bolin, K., Lindgren, B., & Lundborg, P. (2008). Informal and formal care among single‐living
elderly in Europe. Health Economics, 17(3), 393–409.
Booth, A., & Amato, P. R. (2001). Parental predivorce relations and oﬀspring postdivorce
well-being. Journal of Marriage and the Family, 63(1), 197–212.
Bracke, P., Christiaens, W., & Wauterickx, N. (2008). The pivotal role of women in informal
care. Journal of Family Issues, 29(10), 1348–1378.
Brewster, K. L., & Rindfuss, R. R. (2000). Fertility and women’s employment in industrialized
nations. Annual Review of Sociology, 26, 271–296.
Brown, S. L., & Lin, I. F. (2012). The gray divorce revolution: Rising divorce among
middle-aged and older adults, 1990–2010. The Journals of Gerontology Series B,
Psychological Sciences and Social Sciences, 67(6), 731–741.
Bulcroft, K. A., & Bulcroft, R. A. (1991). The timing of divorce: Eﬀects on parent-child
relationships in later life. Research on Aging, 13(2), 226–243.
Cain, B. S. (1989). Parental divorce during the college years. Psychiatry, 52(2), 135–146.
Chan, T. W., & Ermisch, J. (2015). Proximity of couples to parents: inﬂuences of gender,
labour market and family. Demography, 52(2), 379–399.
Chen, F., Liu, G., & Mair, C. A. (2011). Intergenerational ties in context: Grandparents
caring for grandchildren in China. Social Forces, 90(2), 571–594.
Cherlin, A., & Furstenberg, F. A. (1985). Styles and strategies of grandparenting. In L.
Bengston, & J. Robertson (Eds.). Grandparenthood (pp. 97–116). Beverly Hills, CA:
Sage.
Connidis, I. A. (2010). Family ties and aging. Pine Forge Press.
Cooney, T. M. (1994). Young adults’ relations with parents: The inﬂuence of recent
parental divorce. Journal of Marriage and the Family, 45–56.
Cooney, T. M., Hutchinson, M. K., & Leather, D. M. (1995). Surviving the breakup?
Predictors of parent-adult child relations after parental divorce. Family Relations,
153–161.
Crosnoe, R., & Elder, G. H., Jr (2004). Family dynamics, supportive relationships, and
educational resilience during adolescence. Journal of Family Issues, 25(5), 571–602.
Cunningham, M., & Thornton, A. (2006). The Inﬂuences of Parents’ and Oﬀsprings’
Experience with Cohabitation, Marriage, and Divorce on Attitudes Toward Divorce in
Young Adulthood. Journal of Divorce & Remarriage, 44(1–2), 119–144.
Daatland, S. O. (2007). Marital history and intergenerational solidarity: The impact of
divorce and unmarried cohabitation. The Journal of Social Issues, 63(4), 809–825.
Danielsbacka, M., & Tanskanen, A. O. (2016). Grandfather involvement in Finland: Impact of
divorce, remarriage, and widowhood. Grandfathers. UK: Palgrave Macmillan183–197.
Del Boca, D. (2002). The Eﬀect of child care on participation and fertility. Journal of
Population Economics, 15(3), 549–573.
Devine, M., & Earle, T. (2011). Grandparenting: Roles and responsibilities and its implications
for kinship care policies. Vulnerable Children and Youth Studies, 6(2),
124–133.
Dronkers, J., & Härkönen, J. (2008). The intergenerational transmission of divorce in
cross-national perspective: Results from the Fertility and Families Surveys. Population
Studies, 62(3), 273–288.
Dronkers, J., Kalmijn, M., & Wagner, M. (2006). Causes and consequences of divorce:
Cross-national and cohort diﬀerences, an introduction to this special issue. European
Sociological Review, 22(5), 479–481.
Dykstra, P. A., & Fokkema, T. (2011). Relationships between parents and their adult
children: A West European typology of late-life families. Ageing and Society, 31(04),
545–569.
Z. Žilinčíková, M. Kreidl Advances in Life Course Research 38 (2018) 61–71
70
ESS Round 2: European Social Survey Round 2 Data (2004). Data ﬁle edition 3.4.
Norwegian Social Science Data Services, Norway – Data archive and distributor of ESS
data. .
ESS Round 3: European Social Survey Round 3 Data (2006). Data ﬁle edition 3.5.
Norwegian Social Science Data Services, Norway – Data archive and distributor of ESS
data. .
ESS Round 5: European Social Survey Round 5 Data (2010). Data ﬁle edition 3.2.
Norwegian Social Science Data Services, Norway – Data archive and distributor of ESS
data.
Eurostat (2013). Crude divorce rate 1970 to 2011 [statistics]. Available from Eurostat da-
tabase..
Fuller-Thomson, E., & Minkler, M. (2001). American grandparents providing extensive
child care to their grandchildren: Prevalence and proﬁle. The Gerontologist, 41,
201–209. https://doi.org/10.1093/geront/41.2.201.
Gähler, M., & Palmtag, E. L. (2015). Parental Divorce, Psychological Well-Being and
Educational Attainment: Changed Experience, Unchanged Eﬀect Among Swedes Born
1892–1991. Social Indicators Research, 123(2), 601–623.
Geurts, T., Van Tilburg, T., Poortman, A. R., & Dykstra, P. A. (2015). Child care by
grandparents: Changes between 1992 and 2006. Ageing and Society, 35(06),
1318–1334.
Goode, W. J. (1993). World changes in divorce patterns. Yale University Press.
Gray, A. (2005). The changing availability of grandparents as carers and its implications
for childcare policy in the UK. Journal of Social Policy, 34(4), 557–577.
Hammond, R. J., & Muller, G. O. (1992). The late-life divorced: Another look. Journal of
Divorce & Remarriage, 17(3-4), 135–150.
Hank, K. (2007). Proximity and contacts between older parents and their children: A
European comparison. Journal of Marriage and the Family, 69(1), 157–173.
Hank, K., & Buber, I. (2009). Grandparents caring for their grandchildren: Findings from
the 2004 Survey of Health, Ageing, and Retirement in Europe. Journal of Family
Issues, 30, 53–73.
Hank, K., & Kreyenfeld, M. (2003). A multilevel analysis of child care and women’s fertility
decisions in West Germany. Journal of Marriage and the Family, 65(3), 584–596.
Hanson, T. L. (1999). Does parental conﬂict explain why divorce is negatively associated
with child welfare? Social Forces, 77(4), 1283–1316.
Härkönen, J., & Dronkers, J. (2006). Stability and change in the educational gradient of
divorce. A comparison of seventeen countries. European Sociological Review, 22(5),
501–517.
Hodgson, L. G. (1998). Grandparents and older grandchildren. In M. E. Szinovacz (Ed.).
Handbook of grandparenthood. (pp. 171–183). Westport, Conn: Greenwood Press.
Hughes, M. E., Waite, L. J., LaPierre, T. A., & Luo, Y. (2007). All in the family: The impact
of caring for grandchildren on grandparents’ health. The Journals of Gerontology Series
B, Psychological Sciences and Social Sciences, 62(2), 108–119.
Igel, C., & Szydlik, M. (2011). Grandchild care and welfare state arrangements in Europe.
Journal of European Social Policy, 21(3), 210–224.
Jappens, M., & Van Bavel, J. (2012). Regional family cultures and child care by grandparents
in Europe. Demographic Research, 27, 85–120.
Jekielek, S. M. (1998). Parental conﬂict, marital disruption and children’s emotional wellbeing.
Social Forces, 76(3), 905–936. https://doi.org/10.1093/sf/76.3.905.
Kalmijn, M. (2008). The eﬀects of separation and divorce on parent-child relationships in
ten European countries. In C. Saraceno (Ed.). Families, ageing and social policy (pp.
170–193). Cheltenham, England: Edward Elgar Publishing Ltd.
Kalmijn, M. (2010). Country diﬀerences in the eﬀects of divorce on well-being: The role of
norms, support, and selectivity. European Sociological Review, 26(4), 475–490.
Kalmijn, M., & Uunk, W. (2007). Regional value diﬀerences in Europe and the social
consequences of divorce: A test of the stigmatization hypothesis. Social Science
Research, 36(2), 447–468.
Kaptijn, R., Thomese, F., Tilburg, T. G., & Liefbroer, A. C. (2010). How grandparents
matter: Support for the cooperative breeding hypothesis in a contemporary Dutch
population. Human Nature, 21(4), 393–405.
Kennedy, S., & Ruggles, S. (2014). Breaking up is hard to count: The rise of divorce in the
United States, 1980–2010. Demography, 51(2), 587–598.
King, V. (2003). The Legacy of a Grandparent’s Divorce: Consequences for Ties Between
Grandparents and Grandchildren. Journal of Marriage and the Family, 65, 170–183.
King, V., & Elder, G. H. (1995). American children view their grandparents: Linked lives
across three rural generations. Journal of Marriage and the Family, 57, 165–178.
Knudsen, K. (2012). European grandparents’ solicitude: Why older men can be relatively
good grandfathers. Acta Sociologica, 55(3), 231–250.
Ko, P.-C., & Hank, K. (2013). Grandparents Caring for Grandchildren in China and Korea:
Findings From CHARLS and KLoSA. The Journals of Gerontology Series B, Psychological
Sciences and Social Sciences, 69(4), 646–651. https://doi.org/10.1093/geronb/
gbt129.
Kohli, M., Künemund, H., & Lüdicke, J. (2005). Family structure, proximity and contact.
Health, ageing and retirement in Europe-First Results from the Survey of Health, Ageing
and Retirement in Europe, 164–170.
Kreidl, M., Štípková, M., & Hubatková, B. (2017). Parental separation and children’s
education in a comparative perspective: Does the burden disappear when separation
is more common? Demographic Research, 36, 73–110.
Lakomý, M., & Kreidl, M. (2015). Full-time versus part-time employment: Does it inﬂuence
frequency of grandparental childcare? European Journal of Ageing, 12, 321–331.
Lee, J., & Bauer, J. W. (2010). Proﬁles of grandmothers providing child care to their
grandchildren in South Korea. Journal of Comparative Family Studies, 41(3), 455–475.
Lin, I. F. (2008). Consequences of parental divorce for adult children’s support of their
frail parents. Journal of Marriage and the Family, 70(1), 113–128.
Luo, Y., LaPierre, T. A., Hughes, M. E., & Waite, L. J. (2012). Grandparents providing care
to grandchildren: A population-based study of continuity and change. Journal of
Family Issues, 33, 1143–1167.
Manning, W. D., & Brown, S. L. (2011). The demography of unions among older americans,
1980 – Present: A family change approach. In R. A. Settersten, & J. L. Angle
(Eds.). Handbook of sociology of aging (pp. 193–212). New York: Springer.
Monserud, M. A. (2008). Intergenerational relationships and aﬀectual solidarity between
grandparents and young adults. Journal of Marriage and the Family, 70(1), 182–195.
Mood, C. (2010). Logistic regression: Why we cannot do what we think we can do, and
what we can do about it. European Sociological Review, 26(1), 67–82.
Morrison, D., & Cherlin, A. (1995). The divorce process and young children’s well-being:
A prospective analysis. Journal of Marriage and the Family, 57(3), 800–812.
Mueller, M. M., & Elder, G. H. (2003). Family contingencies across the generations:
Grandparent‐grandchild relationships in holistic perspective. Journal of Marriage and
the Family, 65(2), 404–417.
Musil, C. M., Warner, C. B., Zauszniewski, J. A., Jeanblanc, A. B., & Kercher, K. (2006).
Grandmothers, caregiving, and family functioning. Journals of Gerontology - Series B
Psychological Sciences and Social Sciences, 61(2), 89–98.
Pruchno, R. A., & McKenney, D. (2002). Psychological well-being of black and white
grandmothers raising grandchildren examination of a two-factor model. The Journals
of Gerontology Series B, Psychological Sciences and Social Sciences, 57(5), 444–452.
Reed, K., Lucier-Greer, M., & Parker, T. S. (2016). Exploring parental divorce among
emerging adult women: The roles of support networks and family relationships.
Professional Psychology, Research and Practice, 47(3), 231.
Reher, D. S. (1998). Family ties in Western Europe: Persistent contrasts. Population and
Development Review, 203–234.
Riggio, H. R. (2004). Parental marital conﬂict and divorce, parent‐child relationships,
social support, and relationship anxiety in young adulthood. Personal Relationships,
11(1), 99–114. https://doi.org/10.1111/j.1475-6811.2004.00073.x.
Rossi, A. S., & Rossi, P. H. (1990). Of human bonding. Parent-child relations across the life
course. New York: Aldine de Gruyter.
Schenk, N., & Dykstra, P. A. (2012). Continuity and change in intergenerational family
relationships: An examination of shifts in relationship type over a three-year period.
Advances in Life Course Research, 17, 121–132.
Seltzer, J. A. (1991). Relationships between fathers and children who live apart: The
father’s role after separation. Journal of Marriage and the Family, 53(1), 79–101.
Sigle-Rushton, W., Hobcraft, J., & Kiernan, K. (2005). Parental divorce and subsequent
disadvantage: A cross-cohort comparison. Demography, 42(3), 427–446.
Silverstein, M., & Marenco, A. (2001). How Americans enact the grandparent role across
the family life course. Journal of Family Issues, 22, 493–522.
Spijker, J. (2012). Divorce atlas. Centre d’Estudis Demogràﬁcs/Universitat Autònoma de
Barcelona (Spain)https://divorceatlas.wordpress.com.
Suitor, J. J., Sechrist, J., Gilligan, M., & Pillemer, K. (2011). Intergenerational relations in
later-life families. Handbook of sociology of aging. New York: Springer161–178.
Sun, Y. (2001). Family environment and adolescents’ well-being before and after parents’
marital disruption: A longitudinal analysis. Journal of Marriage and the Family, 63,
697–713.
Thompson, R. A., & Amato, P. R. (1999). The postdivorce family: Children, parenting, and
society. Sage Publications.
Uhlenberg, P. (2004). Relationships. Annual review of gerontology and geriatrics, volume
24, 2004. Intergenerational Relations Across Time and Place, 23, 77.
Uhlenberg, P., & Hammill, B. (1998). Frequency of grandparent contact with grandchild
sets: Six factors that make a diﬀerence. The Gerontologist, 38, 276–285. https://doi.
org/10.1093/geront/38.3.276.
Vandell, D. L., McCartney, K., Owen, M. T., Booth, C., & Clarke-Stewart, A. (2003).
Variations in child care by grandparents during the ﬁrst three years. Journal of
Marriage and the Family, 65, 375–381. https://doi.org/10.1111/j.1741-3737.2003.
00375.x.
Wagner, M., & Weiß, B. (2006). On the variation of divorce risks in Europe: Findings from
a meta-analysis of European longitudinal studies. European Sociological Review, 22(5),
483–500.
Westphal, S. K., Poortman, A. R., & Van der Lippe, T. (2015). What about the grandparents?
Children’s postdivorce residence arrangements and contact with grandparents.
Journal of Marriage and the Family, 77(2), 424–440.
Whitbeck, L. B., Hoyt, D. R., & Huck, S. M. (1993). Family relationship history, contemporary
parent-grandparent relationship quality, and the grandparent-grandchild
relationship. Journal of Marriage and the Family, 1025–1035.
White, L. (1992). The eﬀect of parental divorce and remarriage on parental support for
adult children. Journal of Family Issues, 13(2), 234–250.
Wolﬁnger, N. H. (1999). Trends in the intergenerational transmission of divorce.
Demography, 36(3), 415–420.
Wu, Z., & Penning, M. J. (1997). Marital instability after midlife. Journal of Family Issues,
18(5), 459–478.
Yahirun, J. J., & Hamplová, D. (2014). Children’s union status and contact with mothers:
A cross-national study. Demographic Research, 30, 1413–1444.
Z. Žilinčíková, M. Kreidl Advances in Life Course Research 38 (2018) 61–71
71