M K University of Amsterdam
How Childhood Circumstances Moderate the
Long-Term Impact of Divorce on Father–Child
Relationships
Using retrospective survey data collected in the
Netherlands in 2012, the author examined how
childhood circumstances moderate the effect
of an early parental divorce on relationships
between fathers and adult children. Using adult
children’s reports about the frequency of con-
tact and the quality of the relationship, he found
strong negative effects of parental divorce.
These effects are moderated by 3 childhood
conditions. The more fathers were involved in
childrearing during marriage, the less negative
the divorce effect on father–child relationships.
Father’s resources also moderated the effect,
with a smaller divorce effect for more highly
educated fathers. Finally, high levels of inter-
parental conict reduce the impact of divorce as
well, generalizing the stress relief effect to a new
outcome. In general, the study shows that the
impact of divorce is heterogeneous; that child-
hood circumstances play an important role in
this; and that, under specic conditions, there is
virtually no negative effect of parental divorce.
Many studies have documented that parental
divorce has a negative impact on relationships
Department of Sociology, University of Amsterdam,
Nieuwe Achtergracht 166, Building REC B/C, Room
B6.08, 1018 WV Amsterdam, the Netherlands
This article was edited by Deborah Carr.
Key Words: divorce, education, fatherhood, gender roles,
intergenerational relationships, parent involvement.
between fathers and adult children. Adult chil-
dren of divorced parents have less frequent
contact with the father than children of mar-
ried parents, they report more conict with
the father, and they perceive the quality of the
relationship more often as poor (Albertini &
Garriga, 2011; Aquilino, 1994; Daatland, 2007;
de Graaf & Fokkema, 2007; Kalmijn, 2015;
Lin, 2008; Pezzin & Schone, 1999; Shapiro
& Cooney, 2007). Relationships with divorced
mothers are negatively affected as well, but
these effects are more modest than they are
for fathers (Aquilino, 1994; Daatland, 2007).
Negative effects have been found in Europe and
in the United States and hence are not unique
to a specic national context. In a more general
way, these ndings demonstrate that childhood
circumstances can have a lasting inuence on
personal relationships.
The impact of divorce on parent–child rela-
tionships is relevant for both children and
fathers. Because the socioeconomic resources
of the father are important for the life chances
of children, children may be harmed if they
have a weak relationship with their father during
the transition to adulthood. Children may also
suffer emotionally from having no contact with
their father (Amato, 1994; Fischer, 2004). The
problem is relevant for fathers as well. Research
shows that many fathers are unhappy about
the reduction in contact with their minor-age
children after divorce (Parkinson & Smyth,
2004; Spillman, Deschamps, & Crews, 2004).
When fathers are older, weakened ties to adult
Journal of Marriage and Family 77 (August 2015): 921–938 921
DOI:10.1111/jomf.12202
922 Journal of Marriage and Family
children may also lead to less emotional and
practical support in times when such support is
needed. Older divorced men without a partner
may become particularly vulnerable during old
age (Lin, 2008).
Although the long-term effects of divorce on
father–child relationships are well established
empirically, it is clear that there is heterogeneity
in these effects. Not all divorced fathers see the
ties to their children deteriorate and, if there are
negative effects, they vary in degree. As a result,
it is important to shift the focus from estimat-
ing the divorce effect itself to studying factors
that moderate this effect. Some studies have ana-
lyzed variables that affect father–child contact
in samples of divorced fathers, but these vari-
ables may be relevant for married fathers as well
such that it is unclear whether they moderate
the impact of divorce (Aquilino, 2006; Cook-
sey & Craig, 1998; de Graaf & Fokkema, 2007;
Swiss & Le Bourdais, 2009). Other studies have
focused on events occurring after the divorce,
such as repartnering and (second) family for-
mation, to see whether and how these changes
moderate the effect of divorce (Aquilino, 2006;
Clark & Kenney, 2010; Juby, Billette, Laplante,
& Le Bourdais, 2007). Few studies, however,
have looked at the moderating role of circum-
stances during marriage for the long-term effects
of divorce on father–child relationships.
One reason for this gap in the literature lies in
data limitations. Panel studies are generally not
long enough to examine the long-term impact
of parental divorce on father–child relationships.
Retrospective studies can be used for this pur-
pose, but the amount of information on the
parental home in such studies is often limited.
In this study, I analyzed a new nationally rep-
resentative survey that was designed to collect
extensive retrospective information on the bio-
logical father and mother, on their marriage,
and on the way the child was raised by his or
her parents when they were still together. Adult
children were the respondents in the survey,
and all reports on the father and mother were
obtained retrospectively from the adult children.
These data allowed me to examine three impor-
tant childhood circumstances that may moder-
ate the long-term effect of parental divorce on
father–child relationships.
First, I looked at the involvement of fathers
in childrearing during marriage. To what extent
is the involvement of fathers benecial for
the relationship in the long run, and is this
an especially “protective” factor for divorced
fathers? Research has shown that sharing roles
during marriage has positive effects on the
chances of joint custody and on visitation
arrangements after divorce (Juby, Le Bour-
dais, & Marcil-Gratton, 2005). Less is known
about the ramications of role sharing for
the father–child relationship in the long run.
Second, I looked at marital conict. To what
extent do conicts between the father and the
mother during marriage moderate the long-term
impact of divorce on father–child relationships?
Research has found that the effect of divorce
on child well-being is less negative when there
was much conict during marriage (Hanson,
1999). This so-called stress relief hypothesis
may be relevant for a broader set of outcomes
after divorce, including parent–child relations
(Yu, Pettit, Lansford, Dodge, & Bates, 2010).
Third, I looked at the resources of the father. Are
fathers with more resources—more education
and better jobs—better able to secure the ties
to their children when these are older? Previous
studies have shown that better educated nonres-
ident fathers have more frequent contact with
their minor-age children after divorce than less
well educated fathers (Cooksey & Craig, 1998;
Ryan, Kalil, & Ziol-Guest, 2008). If and to what
extent this translates into better relationships
with adult children is not well known.
I studied two outcomes: (a) the frequency
of contact with the father and (b) the per-
ceived quality of the relationship with the father.
These two indicators are often used in research
on adult intergenerational relationships. Contact
frequency is typically seen as a measure of inter-
generational support and solidarity (Silverstein
& Bengtson, 1997). For older parents, contact is
both a form of social support and a condition for
the exchange of other, more instrumental forms
of support. Contact frequency is related to the
quality of the relationship, but contact and qual-
ity are not equivalent. Frequent contact can be
driven by normative obligations (Gans & Silver-
stein, 2006), in which case it does not need to
coincide with a high-quality relationship. Fre-
quent contact can also go together with frequent
conict and lead to feelings of ambivalence
toward parents (Fingerman, Pitzer, Lefkowitz,
Birditt, & Mroczek, 2008). Finally, children may
have close ties with their parents without seeing
them often, for instance, because of geographic
constraints (Hank, 2007). Finally, research has
shown that the quality of the parent–child tie is
Impact of Divorce on Father–Child Relationships 923
the most important predictor of the well-being of
older parents, more important than contact and
support (Merz, Schuengel, & Schulze, 2009).
For all those reasons, contact and quality pro-
vide important complementary insights into the
nature of intergenerational relationships.
The new data were collected in the Nether-
lands, a country with a divorce rate that is simi-
lar to that in other Western European countries
but lower than in the United States (Organi-
sation for Economic Co-operation and Devel-
opment, 2008). Gender role attitudes in the
Netherlands have become considerably more
egalitarian in the past few decades and are
now more egalitarian than the European aver-
age (Kalmijn, 2003). Men’s share of house-
hold work is similar to that in other Western
European countries although somewhat lower
than in the United States (Knudsen & Waer-
ness, 2008). In 1998, policies about postdivorce
arrangements changed in the Netherlands, and
this has resulted in more frequent visits of chil-
dren to divorced fathers (Spruijt, 2006). Most
of the respondents whom I analyzed, however,
experienced the divorce of their parents before
these changes took place (83%).
B H
The goal of this study was to compare children
whose parents divorced during childhood (i.e.,
when the children were living at home) to chil-
dren whose parents remained married during
childhood. For both groups, I considered three
aspects of childhood circumstances: (a) father’s
involvement in childrearing, (b) interparental
conict, and (c) father’s resources. For chil-
dren of divorced parents, these circumstances
apply to the time the parents were still together
(predivorce circumstances). The rst question
was, “To what extent these do circumstances
affect the relationship between father and adult
child many years later, regardless of whether the
parents divorced (main effects).” The second
question was, “To what extent do these circum-
stances moderate the difference in father–child
relationships between married and divorced
fathers (interaction effects)?” For each of the
three circumstances I developed two hypotheses
(discussed below): one about the main effect
of childhood circumstances and one about the
interaction effect.
Although it is clear that fathers vary in terms
of their resources, the conicts they have with
the mother, and the degree to which they are
involved in the child’s life, there are other
sources of heterogeneity that apply specically
to divorced fathers. One source of heterogeneity
lies in events or conditions occurring after the
divorce, for instance, visiting arrangements,
alimony payments, repartnering, family for-
mation, and so on. Such factors can affect
postdivorce relationships between fathers and
children and, in doing so, mediate the impact of
childhood circumstances on adult intergenera-
tional relationships. These postdivorce factors
were not considered here given that they will
not lead to spurious interaction effects, although
they could in principle mediate the interaction
effects. Circumstances that have to do with the
divorce itself also can lead to heterogeneity.
One of these is the age at which children experi-
ence the divorce of their parents. When parents
divorce at an early age of the child, the long-term
effect of divorce on the father–child relationship
tends to be more negative (Aquilino, 2006;
Kalmijn, 2015). To address this issue, I included
the age at divorce in the models, and I examined
whether the interactions of parental divorce with
father involvement, conict, and resources are
affected when the age of the child at the time of
the divorce is taken into account.
Father Involvement
Studies of intergenerational solidarity have
traditionally argued that relationships between
parents and their adult children can be seen
as exchange relations (e.g., Kalmijn, 2014;
Silverstein, Conroy, Wang, Giarrusso, & Bengt-
son, 2002). Parents invest time, energy, and
money in children when they are at younger
ages. At a later age, when parents themselves
are in need of support, adult children recip-
rocate by giving social, emotional, and—in
some contexts—nancial support back to their
parents. The exchange process is often believed
to be subtle and implicit. Exchange also has
a long-term effect: The support that parents
provide to their children works as an investment
in the relationship that sets in motion a cycle of
exchange that strengthens the parent–child tie in
the long run. Longitudinal studies have revealed
indirect evidence for both extended and immedi-
ate reciprocity in intergenerational relationships
(Leopold & Raab, 2011; Silverstein et al., 2002).
In this study I rst examined the investment
perspective for all fathers, regardless of marital
924 Journal of Marriage and Family
status. To make the perspective testable, I
looked at variation in father’s involvement
in childrearing (Sayer, Bianchi, & Robinson,
2004). Although gender roles have become less
traditional, in contemporary society there are
both traditional fathers who spend little time
with their children and more egalitarian fathers
who spend much time with their children (Hook
& Wolfe, 2012; Roeters, van der Lippe, &
Kluwer, 2010; Yeung, Sandberg, Davis-Kean,
& Hofferth, 2001). On average, fathers spend
more time on what is called interactional care
than on physical care, even though contempo-
rary fathers also spend more time on physical
care than fathers in the past (Hook & Wolfe,
2012). According to the investment perspec-
tive discussed above, I expected the following
main effect for fathers: The more fathers were
involved in childrearing during marriage, the
stronger the relationship they have with their
adult children (Hypothesis 1A).
Do investments in a child work differently for
fathers who divorce and fathers who remain mar-
ried? As long as fathers are married to the mother
of their children, they benet from the invest-
ments that the mother made (Seery & Crowley,
2000). Children often see their parents together,
so the relationship of the adult child with the
father will continue independently of how much
fathers invested in the children at an earlier
age. Mothers may also actively manage the
father–child relationship, thus strengthening the
father–child tie at an adult age (Seery & Crow-
ley, 2000). Fathers who are divorced, in contrast,
rely more strongly on their own investments
(Kalmijn, 2007; Stephens, 1996). In other words,
the investments fathers made in the child will pay
off more if he is alone than when he is married
to the mother of his child. The effect of father
involvement on the father–child relationship will
therefore be stronger for divorced fathers than
for married fathers. This implies the following
moderator hypothesis: The more fathers were
involved in the child’s life during marriage, the
less negative the long-term effect of divorce on
father–child relationships (Hypothesis 1B).
Interparental Conict
Many authors have shown that conicts between
parents have negative effects on children’s
well-being (Amato & Cheadle, 2008; Buehler
& Gerard, 2002; Orbuch, Thornton, & Can-
cio, 2000; Schoppe-Sullivan, Schermerhorn,
& Cummings, 2007). Moreover, authors have
argued that marital conict and divorce interact:
The effect of divorce on child well-being is
less negative when there was much conict in
the marriage (Booth & Amato, 2001; Hanson,
1999; Jekielek, 1998). Under these conditions,
a divorce provides a relief from a tense and
difcult situation, and this can be positive for
the child.
Most research on marital conict has focused
on child well-being; fewer studies have exam-
ined parent–child relationships. What role will
marital conict play for the relationship between
the father and the child many years afterward? In
general, one would expect that conict between
parents will have a negative long-term effect
on the parent–child relationship. One reason
for this lies in spillover effects: When there is
much conict between parents, the relationship
between the parent and the child is also less
warm, less supportive, and more conict ridden
(Buehler & Gerard, 2002; Gerard, Krishnaku-
mar, & Buehler, 2006; Schoppe-Sullivan et al.,
2007). Because exposure to conict is unpleas-
ant for children, children will also try to limit
exposure to their parents’ conict. When chil-
dren are living at home, this is often not pos-
sible, but when they start living on their own
they can distance themselves from their par-
ents. This could lead to a gradual decline in
the parent–child relationship, a decline that may
not be turned around when the quality of the
parents’ marriage improves again. In sum, one
would expect that the more conict the parents
had when the child was young, the weaker the
relationship that fathers have with their adult
children (Hypothesis 2A).
When parents divorce, marital conict could
play a different role. A divorce between parents
with high levels of conict may lead to a stress
relief effect. According to Yu et al. (2010),
under conditions of high parental conict,
“children are freed from a dysfunctional family
environment and may genuinely welcome the
shift to a calmer single-parent family” (p. 283).
This effect could also lead to better parent–child
relationships in the long run. When high-conict
parents separate, children are able to see their
parents separately, so the conicts they had
during marriage will not be as visible. The child
can focus on his or her relationship with one
parent at a time, without being in the middle of
their problematic relationship. In other words,
parental conict will have a less negative effect
Impact of Divorce on Father–Child Relationships 925
on father–child relationships when the parents
separate than when they remain married. This
implies the following moderator hypothesis: The
more conict the parents had when the child was
young, the less negative the long-term effect of
divorce on father–child relationships (Hypoth-
esis 2B). In a previous test of this hypothesis
for 22-year-old children in the United States,
Yu et al. (2010) conrmed this interaction
effect for mother–child relationships but not for
father–child relationships.
Father’s Resources
Fathers also differ in the resources they have and,
because these resources are positively related
with their involvement in childrearing (Sayer,
Gauthier, & Furstenberg, 2004), it is impor-
tant to incorporate resources into the analysis
of childhood effects. With resources, I refer to
cultural and socioeconomic resources, as indi-
cated by father’s education and occupational
status. First, one has to consider the possi-
ble main effects of father’s resources. Research
has generally shown that better educated chil-
dren and parents have less frequent face-to-face
contact with each other, although there is no neg-
ative effect on the quality of the intergenera-
tional tie. The effects on contact frequency are
partly due to geographic constraints and in part
to a more individualistic orientation to family
relationships (Kalmijn, 2006). That the quality
of the intergenerational tie is not affected sug-
gests that family ties of higher status groups
are characterized by “intimacy at a distance”
(van Gaalen & Dykstra, 2006). On the basis
of these considerations, I proposed the follow-
ing hypothesis: The more resources that fathers
had during childhood, the less often fathers have
face-to-face contact with their adult children,
whereas there is no effect on the quality of the tie
(Hypothesis 3A).
To what extent would the resources of the
father moderate the impact of divorce? Sev-
eral recent studies have examined how father’s
(and mother’s) education moderate the impact
of divorce on child outcomes (Bernardi & Radl,
2014; Mandemakers & Kalmijn, 2014), but less
is known about how relationships between adult
children and their fathers are affected. One argu-
ment is based on nancial resources. A resource-
ful father may live in a more attractive home, he
may have more consumer goods to share, and he
may pay child support more often (Ryan et al.,
2008; Shehan, Berardo, Owens, & Berardo,
2002). These advantages may strengthen the tie
with the child in the short run and may there-
fore have a positive effect on the father–child
tie when the child is making the transition
to adulthood. Note that repartnering, which is
more common among high-earning divorced
men, may work against this effect (Stewart,
Manning, & Smock, 2003). A second argument
is that resourceful fathers may be more effective
in working out postdivorce agreements with the
mother. More highly educated fathers may also
be more aware of public knowledge about the
negative effects of divorce on children because
education is strongly associated with reading
habits (van de Werfhorst & Kraaykamp, 2001).
These arguments emphasize the cognitive and
cultural aspects of status rather than the nancial
aspects. In sum, my last moderator hypothesis
wasasfollows:The more resources the father
had when the child was young, the less negative
the long-term effect of divorce on father–child
relationships (Hypothesis 3B).
M
The data I used come from the survey Longi-
tudinal Internet Studies for the Social Sciences
(LISS), which is publicly available through
www.lissdata.nl. A unique feature of this survey
is that even though it uses Internet question-
naires, it is based on a representative probability
sample of households in the Dutch popula-
tion. The response rate at the household level
was 48%, which is about average for response
rates in the Netherlands and similar to regular
face-to-face interview surveys (Scherpenzeel,
2009). All household members 16 years of age
and older were asked to complete short online
questionnaires. Respondents were paid when
they completed a questionnaire. The Nether-
lands is in the top 10 countries with the highest
Internet penetration rates (World Telecom-
munication/ICT Indicators Database Online,
www.itu.int/pub/D-IND-WTID.OL). House-
holds without Internet (or without broadband)
received a broadband Internet connection.
I used a special biographical module that
was held in 2012 and that contained detailed
retrospective questions on family background
(n = 5,247). In 2013, the nonresponding mem-
bers of the panel were approached again and
asked to ll out the questionnaire (n = 319).
The module was presented only to household
926 Journal of Marriage and Family
members who were the head of the household
or the partner of the head (single persons were
included). Of the eligible household heads who
were in the original panel, 89% answered this
module. Respondents in the survey were the
(adult) children, and they reported on their
childhood and their parents. Research in the
Netherlands does not suggest that there are large
discrepancies between parent and child reports
on contact, and there is little evidence that con-
tact is underreported when people are more dis-
satised with their relationship (Mandemakers
& Dykstra, 2008).
I excluded respondents who were not living
with both their biological parents when they
were born (n = 104), respondents whose father
was no longer alive (n = 3,377), and respondents
whose biological parents stopped living together
before the respondent left home for reasons
other than separation or divorce (n = 44). I also
deleted data from respondents who lived with
their father after divorce (n = 48) in the main
analyses, but I checked how this affected my
ndings by estimating models that included this
group. Respondents who live with their father
after a divorce are probably a rather select group
but interesting in that their investment oppor-
tunities will not be interrupted by the divorce.
Divorces that occurred recently (0–9 years ago;
n = 15) were excluded because I was interested
in long-term effects. The number of respondents
in the nal analyses was 1,978. The median age
of all respondents (regardless of parental mari-
tal status) was 39, with 90% of the cases falling
between the ages of 24 and 56. The respondents
with divorced parents experienced the divorce in
the years 1969–2004 (when the oldest and most
recent 5% of the distribution was excluded). The
average divorce year was 1987.
Variables and Models
The main independent variable was parental
divorce. Children who grew up with both their
parents were contrasted to children whose
parents divorced when they were still living
at home. Repartnering was not included as
an additional variable because it is a potential
mediator, and my interest was in the overall
effects of divorce and childhood circumstances.
Of the 1,978 respondents, 183 had divorced
parents (9.3%). The median number of years
since the divorce was 26, and the median age
of the respondent at the time of divorce was 9.
In an additional analysis, I made a distinction
between divorces that took place when the child
was young (6 years or younger) and divorces
that took place at an older age (7 or older).
This categorization follows the study conducted
by Aquilino (2006), who found that the most
important differences are between these two age
groups.
The rst dependent variable was the fre-
quency of face-to-face contact between the
father and the adult child in the past 12 months.
Respondents could choose from a number of
frequency categories. I analyzed this outcome
with ordered logit models because this model
does not assume that distances between adja-
cent categories are the same across the range.
I used the following categories: 1 = “never”;
2 = “once”; 3 = “a few times”; 4 = “about
monthly”; and 5 = “weekly, a few times per
week, or daily.” I combined weekly, a few times
per week, and daily because all three can be
considered as quite frequent contact; the more
meaningful variation is in the lower end of the
distribution. The second dependent variable
was the perceived quality of the relationship
between the father and the child. This was mea-
sured on a 5-point scale: 1 = “poor,” 2 = “not
so good,” 3 = “reasonable,” 4 = “good,” and
5 = “very good.” This dependent variable was
also analyzed with an ordered logit model. The
direction of the dependent variables was such
that a positive effect means a better quality
relationship and more frequent contact.
Father involvement in the child’s life was
measured with ve questions about how
childrearing tasks were divided between the
father and the mother (see Table 1 for details
on all items and scales). The tasks reect the
interactional rather than the physical aspect of
childrearing. Interactional care is correlated
with the warmth and closeness in a relationship
(Hook & Wolfe, 2012). The scale is the stan-
dardized sum of the standardized items, and the
reliability is reasonable (𝛼 = .73). Higher values
reect higher levels of involvement. Parental
conict was measured with questions about
four types of conicts occurring between the
parents when the respondent grew up and, in the
case of a divorce, when the parents were still
together. The reliability of this scale is very good
(𝛼 = .87). Higher values reect higher levels of
conict. Two measures of paternal resources
were used: educational level and occupational
status, the two most important variables in
Impact of Divorce on Father–Child Relationships 927
research on social stratication and intergener-
ational reproduction. Occupational status was
measured by recoding 10 broad occupational
groups to the average International Standard
Socio-Economic Index score, as suggested by
Ganzeboom (2005). Education level was the
highest level of completed schooling by the
father, scaled to the approximate number of
years that it takes to complete the degree.
As shown in Table 1, nonresponse in the items
for the scales (“don’t know” and “not appli-
cable”) was not very high. Item nonresponse
was solved internally, which means that when
a respondent had a missing value on one item
in a scale, I used the scores that this respondent
had on the other items in the scale. To do this, I
rst regressed each item of a scale onto the other
items of a scale for a sample of respondents who
had valid scores on all items. To predict a miss-
ing value for a given item, I took the weighted
average of the other valid items for that individ-
ual, where the regression coefcients were used
as weights. In essence, this is a regression impu-
tation of missing items, which is a better strategy
than simply taking the unweighted averages of
the items.
As control variables, I used the respondent’s
age, sex, number of siblings, education level,
marital status, and degree of urbanization of
the place of residence (ranging from rural to
highly urbanized). These variables are known
to be related to intergenerational relationships.
Women generally have better relationships with
their parents than men, more highly educated
children have less frequent face-to-face contact,
married children see their parents less often, and
there tends to be better contact with extended
family members in small towns compared to
large cities (Grundy & Read, 2012; Kalmijn,
2006; Sarkisian & Gerstel, 2008; Schilling &
Wahl, 2002; Ward, Deane & Spitze, 2014).
Means and standard deviations of the control
variables are presented in Table 2.
In the ordered logit models, missing values
on all variables and scales were imputed using
multiple imputation with chained estimation (mi
impute in Stata). Interaction effects with divorce
were included in the estimation procedure. The
dependent variables were also included in the
procedure, although there were no missing val-
ues on these variables (von Hippel, 2007). The
number of imputations was 20, and estimates
for the combined data sets were obtained using
Rubin’s rules in the procedure mi estimate in
Stata. This procedure does not give t statistics
for the nal models.
Table 1. Descriptive Information on Items and Scales Measuring Childhood Circumstances: Adult Children Reporting About
Their Parents
Variable MSDMin Max N
Father’s resources
Father’s ISEI 47.81 18.88 17 82 1,856
Father’s years of schooling 10.55 3.25 6 17 1,709
Father involvement
a
Bringing child to school 2.29 1.03 1 5 1,792
Talk to child about personal matters 2.17 0.87 1 5 1,783
Going on outings with child 2.66 0.83 1 5 1,807
Going with child to family/friends 2.81 0.54 1 5 1,899
Scale of father involvement (z, 𝛼 = .74
b
)0.00 1.00 −2.53 3.72 1,919
Interparental conict
c
Fierce discussions between parents 1.94 0.72 1 3 1,808
Parents strongly blaming each other 1.76 0.77 1 3 1,760
Parents refusing to talk to each other 1.38 0.66 1 3 1,780
Quarrels between parents escalating 1.24 0.56 1 3 1,837
Scale of conict (z, 𝛼 = .87
b
)0.00 1.00 −0.98 2.65 1,874
Note. Data are based on the Longitudinal Internet Studies for the Social Sciences, 2012–2013 (Biographical Module).
Min = minimum; Max = maximum; ISEI = International Standard Socio-Economic Index.
a
Answers were “almost always by mother,” “more often by mother,” “more or less equal,” “more often by father,” and
“almost always by father.”
b
Reliability coefcient calculated before imputation. Scales constructed as mean of standardized
items. The resulting scale is standardized as well.
c
Answers were “never,” “incidentally,” and “frequently.”
928 Journal of Marriage and Family
Table 2. Means and Standard Deviations of Variables Used in the Analyses
Variable MSDMin Max N
Child is daughter .58 .49 .00 1.00 1,978
Child’s age 39.31 9.76 18.00 73.00 1,978
Child’s schooling 12.55 2.87 6.00 17.00 1,970
Child married 0.54 0.50 0.00 1.00 1,978
Degree of urbanization of municipality 3.13 1.29 1.00 5.00 1,961
Number of siblings 1.90 1.40 0.00 10.00 1,952
Father divorced .09 .29 .00 1.00 1,978
Contact with father (as coded in model) 4.09 1.09 1.00 5.00 1,978
Perceived quality of tie with father 3.88 1.03 1.00 5.00 1,978
Note. Data are based on the Longitudinal Internet Studies for the Social Sciences, 2012–2013 (Biographical Module). For
the scales of childhood measures, see Table 1. Min = minimum; Max = maximum.
Because the age of the sample varied, some
respondents needed to look back further in
time than others. It is possible that retrospective
reports are less accurate for older respondents. In
these cases, reports can be biased more strongly
by current circumstances. Previous research on
the quality of retrospective measures of parental
resources has shown that that there is both
systematic error in retrospective reports, which
leads to an overestimation of childhood effects
on current resources, and random error in retro-
spective reports, leading to an underestimation
of such effects (de Vries & de Graaf, 2008). The
sum of these two tendencies is uncertain and
will probably vary from case to case. It was dif-
cult to solve this issue with the present data. In
principle, it is possible to examine interactions
with age, but if there are age interactions they
may also reect cohort interactions. For this
reason, the age variable does not help solve the
problem of retrospective bias.
R
I started with a descriptive analysis of the dif-
ferences in father–child relations between chil-
dren of married and divorced families. The data
in Table 3 show that children with divorced
fathers had less frequent contact with their father
than children with married fathers. This is not a
new nding, but the magnitude of the difference
has not often been discussed. Among respon-
dents with divorced fathers, 26% never saw their
father; among respondents with married fathers,
this was exceptional (3%). It is not just the
frequency of contact that is affected: Approxi-
mately 23% of respondents with divorced fathers
evaluated the relationship with the father as
“poor,” as opposed to only 3% of respondents
with married fathers. Despite these differences,
I also noted that a substantial number of chil-
dren with divorced fathers had frequent and good
contact with their father. The distributions for
children with divorced fathers were quite hetero-
geneous. Note that differences between married
and divorced fathers can in part be due to possi-
ble confounding factors such as personality dif-
ferences, mental health problems, and cultural
orientations of the father (or the child). Given the
magnitude of the differences, it is unlikely that
the entire effect of parental divorce is spurious.
I also looked at the association between the
two indicators of the father–child relationship.
Quality and contact frequency were positively
related, but the correlation differed between
married and divorced fathers. The Pearson cor-
relation between contact and quality (treated as
linear variables) was .49 among married fathers
and .72 among divorced fathers (a signicant
difference, z = 4.75, p < .01). One interpretation
of this difference is that among married fathers,
contact is partly a by-product of having contact
with the mother and hence is less dependent on
the quality of the father–child tie. This would
be consistent with the reasoning behind the
moderator hypothesis about father involvement
(Hypothesis 1B). It is also possible that norma-
tive obligations play a weaker role in families
who experience a divorce, leading to a stronger
association between contact and quality in this
group.
Contact
Ordered logit models for contact are presented in
Table 4. When the coefcients are expressed in
Impact of Divorce on Father–Child Relationships 929
Table 3. Relationships With Fathers According to Adult Children
Relationship variable
Parents together when child was
living at home
Parents divorced when child was
living at home χ
2
test
Current face-to-face contact with father
Never (1)
a
3.326.2
Once (2) 1.68.7
A few times (3) 13.530.6
About monthly (4) 32.922.4
About weekly (5) 29.09.3
Several times a week (5) 16.00.6
Daily (5) 3.62.2
N 1,795 183 286.1
*
Current perceived quality of relationship with father
Poor (1)
a
2.823.0
Not so good (2) 3.610.9
Reasonable (3) 17.028.4
Good (4) 45.526.8
Very good (5) 31.010.9
N 1,795 183 211.8
*
Note. Data are based on the Longitudinal Internet Studies for the Social Sciences, 2012–2013 (Biographical Module).
a
Numbers in parentheses refer to the coding in the ordered logit model.
*
p < .05.
exponential form (i.e., e
b
) they can be interpreted
as odds ratios (ORs; Long & Cheng, 2004).
For example, the effect of years of schooling in
Table 4 (Model 1) is − .05; in exponential form,
this is .95. This means that for each year increase
in schooling, the odds of being in contact cat-
egory j > k rather than in contact category j ≤ k
decline by 5%. For example, the odds of hav-
ing at least weekly contact (rather than less than
weekly) decline by 5% for each year increase
in schooling, as do the odds of having at least
monthly contact (rather than less than monthly),
and so forth (Long & Cheng, 2004, p. 273).
Model 1 conrms the negative effect of
parental divorce on contact. In addition, contact
was more frequent when the child was younger,
when the child was less well educated, when
there were fewer siblings, and when he or she
lived in a less urbanized area. These effects are
broadly consistent with the literature.
Two of the childhood variables had signicant
main effects (Model 1). The more conict there
was between parents when the child was grow-
ing up, the less often the child had contact with
the father as an adult. This points to the long
arm of marital conict. There also was a sig-
nicant negative effect of father’s education on
current contact. Fathers with more education had
less frequent face-to-face contact with their adult
children. Father’s occupation did not affect con-
tact. These effects are broadly consistent with
Hypotheses 2A and 3A. In contrast to Hypothe-
sis 1A, father’s prior involvement in childrearing
did not affect current contact, but the effect
was marginally signicant and in the expected
direction.
To evaluate the magnitude of the main effects,
I calculated ORs. These indicate how much the
odds of being in contact category j > k rather
than in contact category j ≤ k are changed by
a1-SD change in the independent variable (all
interpreted here in the positive direction). Note
that the three childhood variables were stan-
dardized (M = 0, SD = 1). For conict, the OR
was 1.67, whereas for father’s education it was
smaller, 1.21.
In Model 2, I added interactions of child-
hood variables and parental divorce. In line
with expectations, there was a positive and
signicant interaction effect between parental
divorce and father involvement. The main effect
refers to married fathers and was almost zero
(b = 0.035). The sum of the main effect and
the interaction effect refers to divorced fathers
and was positive (b = 0.035 + 0.362 = 0.397).
In other words, when divorced fathers were
more involved in the child’s life when the
children were young, they were more likely
930 Journal of Marriage and Family
Table 4. Ordinal Logit Regression of More Frequent Father–Child Contact: Regression Coefcients and p Values
Model 1 Model 2 Model 3 Model 4
(N = 1,978) (N = 1,978) (N = 1,978) (N = 2,026)
Predictor Coeff. p Coeff. p Coeff. p Coeff. p
Daughter 0.146
†
.10 0.160
†
.07 0.160
†
.07 0.149
†
.09
Age −0.023
*
.00 −0.024
*
.00 −0.024
*
.00 −0.023
*
.00
Years of schooling −0.049
*
.00 −0.047
*
.00 −0.048
*
.00 −0.047
*
.00
Married −0.159
†
.09 −0.155 .10 −0.155 .10 −0.165
†
.08
Urbanization −0.125
*
.00 −0.132
*
.00 −0.133
*
.00 −0.139
*
.00
Number of siblings −0.151
*
.00 −0.156
*
.00 −0.157
*
.00 −0.153
*
.00
Father divorced −1.783
*
.00 −2.057
*
.00 −1.683
*
.00
Divorced age ≤ 6 −2.574
*
.00
Divorced age 7+−1.854
*
.00
Father involvement 0.087
†
.06 0.035 .49 0.035 .48 0.032 .52
× father divorced 0.362
*
.01 0.309
*
.03 0.505
*
.00
Parental conict −0.510
*
.00 −0.550
*
.00 −0.550
*
.00 −0.549
*
.00
× father divorced 0.339
*
.02 0.345
*
.02 0.290
*
.03
Father’s education −0.190
*
.01 −0.247
*
.00 −0.246
*
.00 −0.248
*
.00
× father divorced 0.608
*
.01 0.693
*
.00 0.513
*
.01
Father’s occupation −0.038 .59 −0.008 .91 −0.009 .90 0.006 .93
× father divorced −0.251 .37 −0.342 .23 −0.192 .40
Cutpoint 1 −5.712
*
.00 −5.758
*
.00 −5.812
*
.00 −5.715
*
.00
Cutpoint 2 −5.281
*
.00 −5.319
*
.00 −5.370
*
.00 −5.291
*
.00
Cutpoint 3 −3.766
*
.00 −3.785
*
.00 −3.831
*
.00 −3.769
*
.00
Cutpoint 4 −2.099
*
.00 −2.106
*
.00 −2.150
*
.00 −2.120
*
.00
Note. Data are based on the Longitudinal Internet Studies for the Social Sciences, 2012–2013 (Biographical Module).
“Weekly,” “several times a week,” and “daily” are combined. Missing values were imputed with multiple imputation.
†
p < .10.
*
p < .05.
to see their children when they were adult
and living on their own. I also observed a
signicant interaction of marital conict with
parental divorce. For children of married
fathers, there was a negative effect of parental
conict on contact frequency, but for children of
divorced fathers this effect was weaker, although
still negative (b =−0.550 + 0.339 =−0.211).
Finally, there was a positive interaction with
father’s education. For married fathers, the edu-
cational effect was negative; for divorced
fathers, the implied effect was positive
(b =−0.247 + 0.608 = 0.361). In other words,
more highly educated divorced fathers had more
frequent contact with their adult children than
divorced fathers who were less well educated.
In Model 3 in Table 4, I added the child’s
age at divorce to the model. It is clear that
experiencing a divorce at an early age has a
more negative long-term effect on the amount
of contact with the father. The effect of divorce
for children who experienced the divorce at age
6 or earlier was more negative than the effect
for children who experienced the divorce at an
older age. The difference between these two
coefcients was signicant (χ
2
= 5.58, p = .02).
The interaction effects of divorce and the three
childhood variables did not change when the
age at divorce was included. Note that Model 3
assumes that the interactions with involvement,
conict, and father’s resources were the same
for the two age groups. It is possible that there
were differences in this respect, but the sample
of divorced families was not large enough to test
such three-way interactions convincingly.
In Model 4, I added fathers who lived with
their children after divorce (n = 48) to the sam-
ple. In this somewhat larger sample, the inter-
action between father involvement and divorce
became stronger (Model 4). The other interac-
tion effects remained signicant, however. I will
come back to this nding when I discuss the
models for the quality of the relationship.
To interpret the interactions in light of my
hypotheses, I calculated how a 1-SD change in
the moderator variable changes the overall effect
Impact of Divorce on Father–Child Relationships 931
of parental divorce. To do this, I divided the
interaction effects with parental divorce by the
overall effect of parental divorce. The overall
effect is obtained from Model 1 in Table 4. A
1-SD increase in father involvement reduced the
average divorce effect on contact by 20% in
Model 2 (i.e., 0.362/1.783) and 28% in Model
4 (i.e., 0.505/1.783). For a 1-SD increase in con-
ict, the reduction in the divorce effect was 19%.
For a 1-SD increase in education, the reduc-
tion was 34%. These results are in line with all
three hypotheses (1B, 2B, 3B) and show that the
impact of parental divorce on father–child rela-
tionships was clearly less negative when fathers
were highly involved in childrearing, when they
had much education, and when there was much
conict between parents.
Perceived Quality
The models for the quality of the relationship
are presented in Table 5. There was a signi-
cant effect of parental divorce, in line with the
descriptive results. The control variables had
weaker effects on quality than on contact. The
relationship was perceived as better when the
respondent was younger, when there were fewer
siblings, and when the child was less educated.
Daughters had a somewhat better relationship,
on average, with the father than sons did.
Let us now look at the childhood variables in
Model 1. Marital conict had a signicant nega-
tive effect: The more conict parents had during
childhood, the lower the quality of the relation-
ship between the father and his adult child many
years later. Father’s education and occupation
did not have main effects on the quality of the
tie. In the models for contact, we saw no overall
effect of father involvement; here we did see a
signicant positive effect. In other words, the
more fathers were involved in the child’s life
during marriage, the better the relationship with
their adult children later. In sum, for the quality
of the tie, two hypotheses were conrmed (1A,
2A). When evaluating the magnitude of the
Table 5. Ordinal Logit Regression of Higher Quality Father–Child Relationship: Regression Coefcients and p Values
Model 1 Model 2 Model 3 Model 4
(N = 1,978) (N = 1,978) (N = 1,978) (N = 2,026)
Predictor Coeff. p Coeff. p Coeff. p Coeff. p
Daughter 0.246
*
.00 0.260
*
.00 0.261
*
.00 0.250
*
.00
Age −0.020
*
.00 −0.020
*
.00 −0.020
*
.00 −0.019
*
.00
Years of schooling −0.027 .10 −0.022 .17 −0.023 .16 −0.018 .27
Married 0.038 .69 0.045 .63 0.045 .63 0.025 .78
Urbanization 0.065
†
.06 0.061
†
.08 0.061
†
.08 0.056 .10
Number of siblings −0.082
*
.01 −0.087
*
.01 −0.088
*
.01 −0.086
*
.01
Father divorced −1.213
*
.00 −1.649
*
.00 −1.177
*
.00
Divorced age ≤ 6 −1.835
*
.00
Divorced age 7+−1.580
*
.00
Father involvement 0.270
*
.00 0.247
*
.00 0.247
*
.00 0.240
*
.00
× father divorced 0.161 .26 0.143 .32 0.336
*
.01
Parental conict −0.636
*
.00 −0.680
*
.00 −0.680
*
.00 −0.677
*
.00
× father divorced 0.425
*
.01 0.429
*
.01 0.382
*
.01
Father’s education −0.002 .97 −0.066 .34 −0.066 .34 −0.066 .34
× father divorced 0.854
*
.00 0.886
*
.00 0.518
*
.03
Father’s occupation 0.018 .77 0.034 .60 0.034 .60 0.036 .56
× father divorced −0.279 .35 −0.306 .31 −0.096 .69
Cutpoint 1 −4.501
*
.00 −4.502
*
.00 −4.515
*
.00 −4.315
*
.00
Cutpoint 2 −3.706
*
.00 −3.689
*
.00 −3.702
*
.00 −3.537
*
.00
Cutpoint 3 −2.137
*
.00 −2.096
*
.00 −2.108
*
.00 −1.984
*
.00
Cutpoint 4 0.041 .90 0.093 .77 0.082 .79 0.152 .62
Note. Data are based on the Longitudinal Internet Studies for the Social Sciences, 2012–2013 (Biographical Module).
Missing values were imputed with multiple imputation.
†
p < .10.
*
p < .05.
932 Journal of Marriage and Family
main effects (in Model 1), one can see that the
effect was stronger for conict (OR = 1.89) than
for father involvement (OR = 1.31).
To what extent do the childhood variables
moderate the negative effect of divorce on
the quality of the father–child tie? Model 2
shows that there were two signicant interaction
effects. First, there was a positive interaction
with marital conict. The main effect of mar-
ital conict was b =−0.680, which refers to
married fathers. The effect was weaker—less
negative—for divorced fathers (the implied
effect was b =−0.255). Hence, it appears that
marital conict is less detrimental in the long
run for father–child relationships when parents
divorced. There also was an interaction with
father’s education. For married fathers, there
was no positive effect of education on quality,
but for divorced fathers there was a positive
effect (the implied effect was b = 0.788). Hence,
highly educated divorced fathers were better
able to maintain high-quality relationships with
their children than divorced fathers who had
a lower level of education. Finally, there was
no signicant interaction with father involve-
ment, in contrast to what was found for contact.
Father’s occupation again had no main effect
and no interaction effect with divorce.
In Model 3, I added the child’s age at divorce
to the model. I noted that the father–child rela-
tionship was more negatively affected when the
child was younger when the parents’ divorce
occurred. The difference between the age
groups was not signicant, however (χ
2
= 0.64,
p = 0.42). The interactions between parental
divorce and the three childhood conditions did
not change between Models 2 and 3.
Model 4 estimated the interaction effects
again, now for a sample that included fathers
who lived with their children after divorce.
In this larger sample, the interaction between
divorce and father involvement became larger
and passed the signicance level. That the effect
was signicant in Model 4 suggests that if there
is a “payoff” of father’s involvement for later ties
to his children, this works in part through post-
divorce living arrangements. Previous research
has shown that sharing roles during marriage
has positive effects on the chances that the father
has (joint or sole) custody (Juby et al., 2005). In
other words, more involvement during marriage
may be related to more intensive father–child
relationships in the long run via postdivorce
living arrangements that were more favorable
to the father. Readers should keep in mind
that children who live with their father after a
divorce are a select group, and hence there may
also be alternative interpretations involved in
this nding.
To decide about the validity of the moderator
hypotheses, I evaluated how the effect of divorce
was modied by the three childhood circum-
stances by comparing the interaction effects to
the overall effect of divorce. A 1-SD increase in
father’s education had the strongest moderator
effect: It reduced the divorce effect by 70%
(i.e., 0.931/1.221). This is clear conrmation of
Hypothesis 3B. The effect was substantial for
conict as well, in line with Hypothesis 2B: Each
1-SD increase in conict reduced the divorce
effect by 35%. The moderator hypothesis about
father involvement (1B) received mixed support
because it was signicant only in the sample that
included children who lived with their father. In
this case, the moderator effect was 28%.
Closing the Gap?
Because most childhood circumstances reduce
the gap between married and divorced fathers,
it is important to examine whether the gap
“closes.” To evaluate this, in Figure 1 I present
graphs of the interaction effects. The ordered
logit models were used for the imputations,
and predictions refer to weekly contact and
very good quality relationships (expressed in
proportions). The observed ranges of father
involvement, father’s education, and marital
conict were used to illustrate the interaction
effects. The horizontal axes contain the stan-
dardized scores for the moderator variables
(M = 0, SD = 1). With vertical lines, I show
where the 5% highest and lowest values in
the distribution are. Father’s education did not
have extreme values, and there were also no
extreme values on the lower side of the conict
variable. In these cases, no vertical line was
drawn. Predictions were made while holding all
other variables constant at the mean. Hence, the
proportions reect the differences between
married and divorced fathers when these
groups would be similar in all other measured
respects.
One can see that there was only a small gap
left in contact and quality when there was much
interparental conict. Note that the decline in
this gap was largely caused by the fact that it
was parental conict among married fathers that
Impact of Divorce on Father–Child Relationships 933
F . A F–C R P D C C.
0
0.1
0.2
0.3
0.4
0.5
0.6
-2.7
-2.2
-1.7
-1.2
-0.7
-0.2
0.3
0.8
1.3
1.8
2.3
2.8
3.3
Proportion weekly contact
Father involvement
Married fathers Divorced fathers
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
-0.9
-0.6
-0.3
0
0.3
0.6
0.9
1.2
1.5
1.8
2.1
2.4
2.7
Proportion weekly contact
Marital conflict
Married fathers Divorced fathers
0
0.1
0.2
0.3
0.4
0.5
0.6
-1.4
-1.1
-0.8
-0.5
-0.2
0.1
0.4
0.7
1
1.3
1.6
1.9
Proportion weekly contact
Father’s education
Married fathers Divorced fathers
0
0.1
0.2
0.3
0.4
0.5
-2.7
-2.2
-1.7
-1.2
-0.7
-0.2
0.3
0.8
1.3
1.8
2.3
2.8
3.3
Proportion good relationship
Father involvement
Married fathers Divorced fathers
0
0.1
0.2
0.3
0.4
0.5
-0.9
-0.6
-0.3
0
0.3
0.6
0.9
1.2
1.5
1.8
2.1
2.4
2.7
Proportion good relationship
Marital conflict
Married fathers Divorced fathers
0
0.1
0.2
0.3
0.4
0.5
-1.4
-1.1
-0.8
-0.5
-0.2
0.1
0.4
0.7
1
1.3
1.6
1.9
Proportion good relationship
Father’s education
Married fathers
Divorced fathers
Note. Predicted proportions are evaluated at the mean of the other variables in the model.
934 Journal of Marriage and Family
leads to poor father–child relationships. The gap
in contact with the child also declined when
there was much father involvement, but even
at high levels of involvement there still was a
gap left. Finally, for highly educated fathers,
there was only a small difference in the qual-
ity of the tie to the child and the amount of
contact between married and divorced fathers.
These calculations suggest that the gap was
less problematic under specic childhood cir-
cumstances. At the same time, however, the
gap did not go away entirely, so additional
explanations are still needed to understand the
differences.
D
This study shows that childhood circumstances
have a lasting effect on father–child relation-
ships. Moreover, childhood circumstances mod-
ify the impact of an early parental divorce on
relationships between fathers and their adult
children.
First, the analyses point to the importance of
role sharing during marriage. When fathers were
more involved in the child’s life—regardless of
marital status—they had a better quality relation-
ship with their adult children. More important,
this effect was greater for divorced fathers than
for married fathers. The interpretation of this
interaction is that for fathers who are divorced,
a low level of prior investments in the child can
lead to a downward spiral in the relationship.
Under those conditions, postdivorce arrange-
ments will probably be less benecial for the
father, and maintaining ties to his children with-
out a female kinkeeper may be more difcult. In
a sense, divorced fathers rely more on their own
investments, hence the stronger effect of father
involvement in childrearing on father–child
relationships.
This nding is important because it implies
that when fathers are highly involved during
marriage, they experience less negative conse-
quences of a divorce, at least in terms of the
relationships they have with their children. In
a more general sense, this also suggests that
the increasingly egalitarian gender roles in mar-
riage will eventually reduce the gender gap
in the (social) consequences of divorce. The
nding that involvement during marriage plays
an important role echoes ndings from past
research that have shown that role sharing leads
to better custody and visiting arrangements for
fathers immediately after divorce (Juby et al.,
2005). It is also consistent with cross-national
research in Europe that suggests that the effect
of divorce on adult father–child relations is less
negative in more gender-egalitarian countries
(Kalmijn, 2008). Future research could exam-
ine the extent to which the effect of role sharing
during marriage on the long-term father–child
relationship is mediated by more favorable cir-
cumstances immediately after divorce.
Second, the analyses showed that the amount
of conict between parents is an important
factor. When main effects were compared using
standardized variables, interparental conict
turned out to be the most relevant for the future
relationship between fathers and children. Some
studies have shown that marital conict has
negative spillover effects: Young children have
poorer relationships with their parents when
parents frequently have conict. The present
analyses show that there is a long-term effect of
marital conict as well. However, when parents
are divorced, marital conict had a weaker
effect on adult father–child relationships. This
nding echoes the so-called stress relief effect
in research on child well-being (Hanson, 1999).
In the present study, the stress relief effect
generalized to the quality of the father–child
relationship (Yu et al., 2010). Under conditions
of high interparental conict children are better
able to interact with their father if they are not
in the middle of these conicts, as is more likely
to be the case if parents remain married. When
parents are married and have high levels of
conict, children may nd it unpleasant to visit
their parental home and gradually become more
distant from their parents.
Third, I examined the extent to which the
resources of the father moderate the divorce
effect. I expected that high-resource fathers
would be better able to maintain a good rela-
tionship with their children after divorce than
low-resource fathers. I found positive evi-
dence for this hypothesis, but only for father’s
education, not for father’s occupation. The
more education the father had, the less nega-
tive the effect of divorce on the quality of the
father–child tie and on the frequency of contact.
It is important to note that this nding was
obtained while controlling for marital conict
and father involvement. Hence, the interpreta-
tion must be sought in arguments not related
to conict and gender roles. It is possible that
high-status fathers are better able to provide a
Impact of Divorce on Father–Child Relationships 935
more comfortable home environment for the
child and are more likely to pay child sup-
port. Because only the father’s education, and
not father’s occupation, yielded a signicant
interaction effect, this explanation seems less
plausible. Occupational status is more closely
linked to material well-being and income than
education. High-status fathers may also be
more aware of the possible negative effects
of divorce on children and be more skilled in
negotiating good arrangements after divorce.
That education is a more important moderating
variable than occupation suggests that cognitive
skills in handling the divorce and its aftermath
may indeed be more relevant. My nding also
suggests that relatively enduring resources of
the father (e.g., education) are more important
in this case than the resources he happens to
have during childhood.
In contrast to previous studies, the present
analysis is based on nationally representative ret-
rospective survey data. One important advantage
of the retrospective design is that it was relatively
easy to examine long-term effects of divorce
and childhood circumstances. The average time
between the divorce and the measurement of the
relationship was 26 years. Most previous stud-
ies have been prospective. Such studies tend to
be short, such that consequences of divorce are
examined for adolescents or young adults. A
disadvantage of the retrospective design is that
reports about the past can be biased. The extent
to which this is true is difcult to answer conclu-
sively with the present data. Past methodologi-
cal studies have shown that retrospective reports
on childhood circumstances have both random
and systematic measurement error and that these
two sources of error work against each other (de
Vries & de Graaf, 2008). As a result, the direc-
tion of bias in the effects of childhood variables
is not necessarily upward.
Finally, I need to offer a caveat about the
causal nature of the effects. It is clear that other
characteristics of fathers and children may be
related to father–child relationships and to the
risk of divorce. Examples are personality differ-
ences, differences in mental health, or different
cultural orientations. If such factors play a role,
then the main divorce effects can be biased in
an upward fashion. The magnitude of the main
effects is such that it is hard to believe that
the entire effect would be spurious. These pos-
sible confounders could also bias the modera-
tor effects, although it requires more complex
theorizing to evaluate the direction in which this
would go. To address these concerns, either more
complex statistical methods based on instrumen-
tal variables (Kim, 2011) or better theoretical
arguments about what these unmeasured fac-
tors could be, in combination with new data that
measures these factors, are needed.
R
Albertini, M., & Garriga, A. (2011). The effect of
divorce on parent–child contacts: Evidence on two
declining effect hypotheses. European Societies,
13, 257–278. doi:10.1080/14616696.2010.483002
Amato, P. R. (1994). Father–child relations, mother–
child relations, and offspring psychological well-
being in early adulthood. Journal of Marriage and
the Family, 56, 1031–1042.
Amato, P. R., & Cheadle, J. E. (2008). Parental
divorce, marital conict and children’s behavior
problems. Social Forces, 86, 1139–1161.
Aquilino, W. S. (1994). Impact of family disruption
on young adults’ relationships with parents. Jour-
nal of Marriage and the Family, 56, 295–313.
doi:10.2307/353602
Aquilino, W. S. (2006). Noncustodial father–child
relationship from adolescence into young adult-
hood. Journal of Marriage and Family, 68,
929–946. doi:10.1111/j.1741-3737.2006.00305.x
Bernardi, F., & Radl, J. (2014). The long-term con-
sequences of parental divorce for children’s edu-
cational attainment. Demographic Research, 30,
1653–1680.
Booth, A., & Amato, P. R. (2001). Parental predivorce
relations and offspring postdivorce well-being.
Journal of Marriage and Family, 63, 197–212.
doi:10.1111/j.1741-3737.2001.00197.x
Buehler, C., & Gerard, J. M. (2002). Marital con-
ict, ineffective parenting, and children’s and
adolescents’ maladjustment. Journal of Marriage
and Family, 64, 78–92. doi:10.1111/j.1741-3737.
2002.00078.x
Clark, S., & Kenney, C. (2010). Is the United States
experiencing a “matrilineal tilt?”: Gender, family
structures and nancial transfers to adult children.
Social Forces, 88, 1753–1776.
Cooksey, E. N., & Craig, P. H. (1998). Parenting
from a distance: The effects of paternal charac-
teristics on contact between nonresidential fathers
and their children. Demography, 35, 187–200.
doi:10.2307/3004051
Daatland, S. O. (2007). Marital history and inter-
generational solidarity: The impact of divorce and
unmarried cohabitation. Journal of Social Issues,
63, 809–825.
de Graaf, P. M., & Fokkema, T. (2007). Contacts
between divorced and non-divorced parents
and their adult children in the Netherlands: An
936 Journal of Marriage and Family
investment perspective. European Sociological
Review, 23, 263–277. doi:10.1093/esr/jcl032
de Vries, J., & de Graaf, P. M. (2008). Is the
intergenerational transmission of high cultural
activities biased by the retrospective measure-
ment of parental high cultural activities? Social
Indicators Research, 85, 311–327. doi:10.1007/
s11205-007-9096-4
Fingerman, K. L., Pitzer, L., Lefkowitz, E. S., Birditt,
K. S., & Mroczek, D. (2008). Ambivalent rela-
tionship qualities between adults and their parents:
Implications for the well-being of both parties.
Journals of Gerontology Series B: Psychological
Sciences and Social Sciences, 63, 362–371.
Fischer, T. (2004). Parental divorce, conict, and
resources. Amsterdam: Thesis.
Gans, D., & Silverstein, M. (2006). Norms of lial
responsibility for aging parents across time and
generations. Journal of Marriage and Family, 68,
961–976. doi:10.1111/j.1741-3737.2006.00307.x
Ganzeboom, H. B. G. (2005). On the cost of being
crude: A comparison of detailed and coarse occu-
pational coding. In J. H. P. Hoffmeyer-Zlotnik
(Ed.), Methodological aspects of cross-national
research (pp. 241–258). Mannheim, Germany:
ZUMA.
Gerard, J. M., Krishnakumar, A., & Buehler, C.
(2006). Marital conict, parent–child relations, and
youth maladjustment: A longitudinal investigation
of spillover effects. Journal of Family Issues, 27,
951–975. doi:10.1177/0192513X05286020
Grundy, E., & Read, S. (2012). Social contacts and
receipt of help among older people in England: Are
there benets of having more children? Journals
of Gerontology Series B: Psychological Sciences
and Social Sciences, 67, 742–754. doi:10.1093/
geronb/gbs082
Hank, K. (2007). Proximity and contacts between
older parents and their children: A European com-
parison. Journal of Marriage and Family, 69,
157–173. doi:10.1177/0192513X08322627
Hanson, T. L. (1999). Does parental conict explain
why divorce is negatively associated with child
welfare? Social Forces, 77, 1283–1315.
Hook, J. L., & Wolfe, C. M. (2012). New fathers?
Residential fathers’ time with children in four
countries. Journal of Family Issues, 33, 415–450.
doi:10.1177/0192513X11425779
Jekielek, S. M. (1998). Parental conict, marital
disruption and children’s emotional well-being.
Social Forces, 76, 905–935.
Juby, H., Billette, J. M., Laplante, B., & Le Bourdais,
C. (2007). Nonresident fathers and children: Par-
ents’ new unions and frequency of contact. Jour-
nal of Family Issues, 28, 1220–1245. doi:10.1177/
0192513X07302103
Juby, H., Le Bourdais, C., & Marcil-Gratton, N.
(2005). Sharing roles, sharing custody? Couples’
characteristics and children’s living arrangements
at separation. Journal of Marriage and Family, 67,
157–172. doi:10.1111/j.0022-2445.2005.00012.x
Kalmijn, M. (2003). Country differences in sex-role
attitudes: Cultural and economic explanations. In
W. Arts, J. Hagenaars, & L. Halman (Eds.), The
cultural diversity of European unity: Findings,
explanations and reections from the European
Values Study (pp. 311–337). Leiden, the Nether-
lands: Brill.
Kalmijn, M. (2006). Educational inequality and fam-
ily relationships: Inuences on contact and prox-
imity. European Sociological Review, 22, 1–16.
doi:10.1093/esr/jci036
Kalmijn, M. (2007). Gender differences in the
effects of divorce, widowhood and remarriage
on intergenerational support: Does marriage
protect fathers? Social Forces, 85, 1079–1104.
doi:10.1353/sof.2007.0043
Kalmijn, M. (2008). The effects of separation and
divorce on parent–child relationships in ten Euro-
pean countries. In C. Saraceno (Ed.), Families,
ageing and social policy: Intergenerational soli-
darity in European welfare states (pp. 170–193).
Cheltenham: Edward Elgar. doi:10.4337/
9781848445147
Kalmijn, M. (2014). Adult intergenerational rela-
tionships. In J. Treas, J. Scott, & M. Richards
(Eds.), The Wiley-Blackwell companion to the soci-
ology of families (pp. 385–403). Hoboken, NJ:
Wiley-Blackwell.
Kalmijn, M. (2015). Relationships between fathers
and adult children: The cumulative effects of
divorce and repartnering. Journal of Family Issues,
36, 737–759. doi:10.1177/0192513X13495398
Kim, H. S. (2011). Consequences of parental divorce
for child development. American Sociological
Review, 76, 487–511.
Knudsen, K., & Wærness, K. (2008). National context
and spouses’ housework in 34 countries. European
Sociological Review, 24, 97–113. doi:10.1093/
esr/jcm037
Leopold, T., & Raab, M. (2011). Short-term reci-
procity in late parent–child relationships. Jour-
nal of Marriage and Family, 73, 105–119. doi:10.
1111/j.1741-3737.2010.00792.x
Lin, I. F. (2008). Consequences of parental divorce
for adult children’s support of their frail parents.
Journal of Marriage and Family, 70, 113–128.
Long, J. S., & Cheng, S. (2004). Regression mod-
els for categorical outcomes. In M. Hardy &
A. Bryman (Eds.), Handbook of data analysis
(pp. 259–284). London: Sage.
Mandemakers, J. J., & Dykstra, P. A. (2008). Dis-
crepancies in parent’s and adult child’s reports
of support and contact. Journal of Marriage and
Family, 70, 495–506. doi:10.1111/j.1741-3737.
2008.00496.x
Mandemakers, J. J., & Kalmijn, M. (2014). Do
mother’s and father’s education condition the
Impact of Divorce on Father–Child Relationships 937
impact of parental divorce on child well-being?
Social Science Research, 44, 187–199. doi:10.
1016/j.ssresearch.2013.12.003
Merz, E. M., Schuengel, C., & Schulze, H. J.
(2009). Intergenerational relations across 4 years:
Well-being is affected by quality, not by sup-
port exchange. The Gerontologist, 49, 536–548.
doi:10.1093/geront/gnp043
Orbuch, T. L., Thornton, A., & Cancio, J. (2000). The
impact of marital quality, divorce and remarriage
on the relationships between parents and their chil-
dren. Marriage & Family Review, 29, 221–246.
doi:10.1300/J002v29n04_01
Organisation for Economic Co-operation and Devel-
opment. (2008). SF8: Marriage and divorce rates.
Retrieved from www.oecd.org/els/social/family/
database
Parkinson, P., & Smyth, B. (2004). Satisfaction and
dissatisfaction with father–child contact arrange-
ments in Australia. Child and Family Law Quar-
terly, 16, 289–304.
Pezzin, L. E., & Schone, B. S. (1999). Parental mar-
ital disruption and intergenerational transfers: An
analysis of lone elderly parents and their children.
Demography, 36, 287–297. doi:10.2307/2648053
Roeters, A., van der Lippe, T., & Kluwer, E. S. (2010).
Work characteristics and parent–child relationship
quality: The mediating role of temporal involve-
ment. Journal of Marriage and Family, 72, 1317–
1328. doi:10.1111/j.1741-3737.2010.00767.x
Ryan, R. M., Kalil, A., & Ziol-Guest, K. M. (2008).
Longitudinal patterns of nonresident fathers’
involvement: The role of resources and relations.
Journal of Marriage and Family, 70, 962–977.
Sarkisian, N., & Gerstel, N. (2008). Till marriage
do us part: Adult children’s relationships with
their parents. Journal of Marriage and Family, 70,
360–376. doi:10.1111/j.1741-3737.2008.00487.x
Sayer, L. C., Bianchi, S. M., & Robinson, J. P. (2004).
Are parents investing less in children? Trends in
mother’s and father’s time with children. Ameri-
can Journal of Sociology, 110, 1–43. doi:10.1086/
386270
Sayer, L. C., Gauthier, A. H., & Furstenberg, F.
F. (2004). Educational differences in parents’
time with children: Cross-national variations.
Journal of Marriage and Family, 66, 1152–1169.
doi:10.1111/j.0022-2445.2004.00084.x
Scherpenzeel, A. (2009). Start of the LISS panel: Sam-
ple and recruitment of a probability-based Internet
panel. Tilburg, the Netherlands: CentERdata.
Schilling, O., & Wahl, H. W. (2002). Family networks
and life satisfaction of older adults in rural and
urban regions. Kölner Zeitschrift für Soziologie
und Sozialpsychologie, 54, 304–317. doi:10.1007/
s11577-002-0041-x
Schoppe-Sullivan, S. J., Schermerhorn, A. C., &
Cummings, M. E. (2007). Marital conict and
children’s adjustment: Evaluation of the parenting
process model. Journal of Marriage and Family,
69, 1118–1134. doi:10.1111/j.1741-3737.2007.
00436.x
Seery, B. L., & Crowley, M. S. (2000). Women’s
emotion work in the family relationship manage-
ment and the process of building father–child rela-
tionships. Journal of Family Issues, 21, 100–127.
doi:10.1177/019251300021001005
Shapiro, A., & Cooney, T. M. (2007). Divorce and
intergenerational relations across the life course.
Advances in Life Course Research, 12, 191–219.
Shehan, C. L., Berardo, F. M., Owens, E., & Berardo,
D. H. (2002). Alimony: An anomaly in family
social science. Family Relations, 51, 308–316.
doi:10.1111/j.1741-3729.2002.00308.x.
Silverstein, M., & Bengtson, V. L. (1997). Inter-
generational solidarity and the structure of adult
child–parent relationships in American families.
American Journal of Sociology, 103, 429–460.
doi:10.1086/231213
Silverstein, M., Conroy, S. J., Wang, H., Giarrusso,
R., & Bengtson, V. L. (2002). Reciprocity in
parent–child relations over the adult life course.
Journals of Gerontology Series B: Psychological
Sciences and Social Sciences, 57B, 3–13.
Spillman, J. A., Deschamps, H. S., & Crews, J.
A. (2004). Perspectives on nonresidential pater-
nal involvement and grief: A literature review. The
Family Journal, 12, 263–270.
Spruijt, A. P. (2006). Omgang met scheidingskinderen
en ouderlijke ruzies [Interaction with children of
divorce and parental conict]. Tijdschrift voor
Familie en Jeugdrecht, 31, 41–47.
Stephens, L. S. (1996). Will Johnny see Daddy this
week? An empirical test of three theoretical per-
spectives of postdivorce contact. Journal of Fam-
ily Issues, 17, 466–494. doi:10.1177/01925139601
7004003
Stewart, S. D., Manning, W. D., & Smock, P. J. (2003).
Union formation among men in the US: Does
having prior children matter? Journal of Marriage
and Family, 65, 90–104. doi:10.1111/j.1741-3737.
2003.00090.x
Swiss, L., & Le Bourdais, C. (2009). Father–child
contact after separation: The inuence of liv-
ing arrangements. Journal of Family Issues, 30,
623–652. doi:10.1177/0192513X08331023
van de Werfhorst, H. G., & Kraaykamp, G. (2001).
Four eld-related educational resources and their
impact on labor, consumption, and sociopolitical
orientation. Sociology of Education, 74, 296–317.
doi:10.2307/2673137
van Gaalen, R. I., & Dykstra, P. A. (2006). Soli-
darity and conicts between adult children and
parents: A latent class analysis. Journal of Mar-
riage and Family, 68, 947–960. doi:10.1111/j.
1741-3737.2006.00306.x
von Hippel, P. T. (2007). Regression with missing
Ys: An improved strategy for analyzing multiply
938 Journal of Marriage and Family
imputed data. Sociological Methodology, 37,
83–117. doi:10.1111/j.1467-9531.2007.00180.x
Ward, R., Deane, G., & Spitze, G. (2014). Life-course
changes and parent–adult child contact. Research
on Aging, 36, 568–602. doi:10.1177/01640275
13510325
Yeung, W. J., Sandberg, J. F., Davis-Kean, P. E.,
& Hofferth, S. L. (2001). Children’s time with
fathers in intact families. Journal of Marriage and
Family, 63, 136–154. doi:10.1111/j.1741-3737.
2001.00136.x
Yu, T. Y., Pettit, G. S., Lansford, J. E., Dodge, K. A., &
Bates, J. E. (2010). The interactive effects of mar-
ital conict and divorce on parent–adult children’s
relationships. Journal of Marriage and Family, 72,
282–292. doi:10.1111/j.1741-3737.2010.00699.x
