‘Silver splits’ in Europe: The role of grandchildren and other correlates

DEMOGRAPHIC RESEARCH

VOLUME 46, ARTICLE 21, PAGES 619652

PUBLISHED 1 APRIL 2022

https://www.demographic-research.org/Volumes/Vol46/21/

DOI: 10.4054/DemRes.202

Research

Article

‘Silver splits’ in Europe: The role of

grandchildren and other correlates

Giammarco Alderotti

Cecilia Tomassini

Daniele Vignoli

This open-access work is published under the terms of the Creative Commons

Attribution 3.0 Germany (CC BY 3.0 DE), which permits use, reproduction,

and distribution in any medium, provided the original author(s) and source

are given credit.

See

https://creativecommons.org/licenses/by/3.0/de/legalcode

Contents

Introduction

620

Background

622

2.1

Emptying, then refilling, the nest: The role of children and grandchildren

622

2.2

Other factors (potentially) relate

d to late union dissolutions

624

Data and methods

627

3.1

Sample selection

627

3.2

Variables and imputations

628

3.3

Modelling

631

Results

631

4.1

Descriptive findings: Country differences in late union dissolution

631

4.2

The role of children and grandchildren for silver splits

632

4.3

The other correlates of silver splits

4.4

Robustness checks and further analyses

Conclusions

Acknowledg

ments

References

Appendix

Demographic Research: Volume 46, Article 21

Research Article

https://www.demographic-research.org 619

‘Silver splits’ in Europe:

The role of grandchildren and other correlates

Giammarco Alderotti

Cecilia Tomassini

Daniele Vignoli

Abstract

BACKGROUND

‘Silver splits’ – the union dissolutions after the age of 50 – have received growing

attention in both the press and nonacademic discourse. Nonetheless, while there is a vast

amount of research on the sociodemographic, health-related, and economic consequences

of late union dissolution, no studies have yet (to the best of our knowledge) analysed the

correlates of silver splits in Europe.

OBJECTIVE

This paper aims to document the correlates of union dissolution in later life in Europe,

with a specific focus on the role played by grandchildren.

METHODS

We used data from the Survey of Health, Ageing and Retirement in Europe (SHARE)

and employed logistic regression to model the probability of experiencing union

dissolution after the age of 50.

RESULTS

Our results show that (1) having grandchildren is related to a lower probability of

experiencing a silver split, (2) the other correlates of silver splits generally do not differ

from the classical correlates of union dissolution early in life, and (3) the European

correlates of silver splits accord with those found in the literature for North America.

CONTRIBUTION

This study sheds light on an increasingly relevant new family process occurring later in

life (silver splits), thereby filling a clear gap in the European literature. Among the

correlates of silver splits, the role of grandchildren appears crucial. They serve to ‘refill

Università degli Studi di Firenze, Italy. Email: [email protected].

Università degli Studi del Molise, Italy. Email: [email protected].

Università degli Studi di Firenze, Italy. Email: [email protected].

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

620 https://www.demographic-research.org

the nest’ once a couple’s children have left, thereby inhibiting silver splits as grandparents

assume new responsibilities in the family and society.

1. Introduction

De Shane and Brown-Wilson’s (1982) paper “Divorce in Late Life: A Call for Research”

emphasises the near absence of late divorces from academic research, with only a handful

of papers considering age as a control variable. A possible explanation for this research

gap could be the small scale of the phenomenon at the time. Nevertheless, the authors

stress both that the number of late divorces could increase in the future and that the

antecedents and consequences of divorce at older ages could significantly impact later

life stages, thus making the subject a stimulating new topic for gerontological literature

on the family life course (De Shane and Brown-Wilson 1982). In their call for research,

they suggest some theoretical and operational issues. First, they wonder how society’s

increasing acceptance of divorce might impact the incoming cohorts of older people,

who, until then, had been more averse to divorce. Second, they stress how women’s

widespread entrance into the job market had offered them interests and activities outside

the home, suggesting a possible increase in divorce even later in life. Finally, they

emphasise the importance of later life transitions, such as the ‘empty nest phase’ (for

women) and retirement (for men) as potentially important correlates for late divorce. In

a study of 121 couples aged 60 and over filing for divorce, Weingarten (1988) – in support

of De Shane and Brown-Wilson’s (1982) hypothesis – notes that having children was one

of the most important aspects influencing the decision to divorce after age 60. De Shane

and Brown-Wilson (1982) note that the consequences of late divorce could affect the

lives of older people in several important ways. It may generate a decline in support

received from the couple’s social network and reduced contact with children and

grandchildren – especially for divorced men. In addition, it could carry such

psychological consequences as increasing the likelihood of divorce in subsequent

generations of children and entail serious economic costs – especially for women.

In line with De Shane and Brown-Wilson’s (1982) prediction, divorces after the age

of 50 increased in the following decades in several wealthy countries, such as the United

States (Brown and Lin 2012; Kennedy and Ruggles 2014), Canada (Wu and Penning

1997), the United Kingdom (ONS 2017), and France (Solaz 2021), while concurrently

levelling in younger age groups. This increase has been ascribed to the ageing of the most

divorce-prone cohort, namely those born in the baby boom (Cohen 2019; Crowley 2019),

and to the elevated presence of older people in second or higher-order marriages – both

of which are at greater risk of divorce (Brown and Lin 2012; Crowley 2019). More recent

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 621

cohorts show later ages at marriage and higher levels of cohabitation, which have

contributed to flattening divorce rates (Raley and Sweeney 2020; Rotz 2016). However,

it should be noted that it may be too early to detect whether this association persists later

in life. In spite of increasing divorce rates, the correlates of so-called ‘grey divorces’

(marital dissolutions after the age of 50) – or more generally, ‘silver splits’

(the union

dissolutions after the age of 50) – have only been described in a limited number of studies,

despite the wide consensus that silver splits may have potentially relevant consequences

for both men and women. Moreover, the few studies which have examined silver splits

are largely limited to a US context (Brown and Lin 2012; Karraker and Latham 2015;

Lin et al. 2018). Surprisingly, the importance of the role of children and grandchildren –

originally emphasised by De Shane and Brown-Wilson (1982) and Weingarten (1988) –

has received scant attention, except for a recent paper based in the United States (Brown,

Lin, and Mellencamp 2021). In fact, following a life course approach, experiences at one

stage of life will impact later stages (Bernardi, Huinink, and Settersten 2019; Elder 1994),

suggesting that becoming parents and grandparents may have consequences on late-life

family transitions (Esterberg, Moen, and Dempster‐McCain 1994), including

experiencing a silver split. While the negative relation between childbearing and union

dissolution for couples in a reproductive age has been an enduring finding, comparatively

nothing is known about how they relate to late union dissolutions. Moreover, having

grandchildren – a distinct feature of later life – represents the continuation of the family

lineage and may improve marital stability at later ages, thus operating as a new shared

project for grandparents (Bair 2007; Berger and Kellner 1964; Brown, Lin, and

Mellencamp 2021).

Little is known about European silver splits. Indeed, information on their correlates

may be gathered from only longitudinal surveys, and few European studies have

combined a longitudinal design with a sample size large enough to allow such an

infrequent event to be studied. Surveys with retrospective questions are unable to provide

information on the previous partners of divorced or separated individuals, and the

potential determinants are collected at the time of the interview, thus hindering the

identification of accurate measures of the phenomenon’s causes (Uhlenberg, Cooney, and

Boyd 1990; Hoem and Kreyenfeld 2006). While data from population registers may be

used, they lack information on several correlates (e.g., physical and mental health, and

relations with children and grandchildren) that have been recognised as important for

understanding silver splits. Survey data also extend the possibility to focus not only on

grey divorces but also (more generally) late union dissolution, thus acknowledging the

increasing importance of cohabitation later in life. Additionally, while longitudinal

surveys have surely improved our understanding of the union dissolution process,

methodological problems arise when studying the antecedents or consequences of union

In this paper, we use the terms ‘silver split’ and ‘late union dissolution’ interchangeably.

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

622 https://www.demographic-research.org

dissolution due to attrition since union dissolution itself may be responsible for losing

participants before follow-up stages.

This paper aims to (1) detect the role of children and grandchildren – who are a

distinctive feature of later life – in facilitating or inhibiting union dissolution and (2)

compare European correlates of late union dissolution with those from previous research

in North America.

2. Background

Studies published in recent decades have improved our understanding of the correlates

of divorce. This improvement is, to a great extent, thanks to the availability of large-scale

longitudinal surveys that allow researchers to control for spurious associations and

reverse causation, which could have affected previous research based on cross-sectional

data (Glenn and Supancic 1984). Although longitudinal studies have certainly helped

identify the antecedents and consequences of divorce in general, they have rarely been

used to study grey divorce despite growing attention in the press and nonacademic

discourses. Only in the last ten years has gerontological and family research begun to

address divorce and repartnering in later life. Below, we briefly review the factors

associated with union dissolution at all ages. This review offers input for selecting

correlates of silver splits for our empirical analysis. Among these factors, we have

devoted particular consideration to having children and grandchildren and the role they

play in silver splits between different cohorts of older people.

2.1 Emptying, then refilling, the nest: The role of children and grandchildren

The various sociodemographic perspectives that explain the association between

childbearing and union dissolution all agree upon the fact that having children relates to

more stable unions, as children are a ‘union-specific capital’ that improves union stability

(e.g., Becker, Landes, and Michael 1977). Such a relationship could also be due to

selection mechanisms as family-oriented individuals are both more likely to have

children and less likely to end a union (Lesthaeghe and Moors 2000). An extensive body

of literature shows that having children (especially when they are young) usually

discourages divorce (see Lyngstad and Jalovaara 2010, for a review). However, scant

evidence is available about late union dissolutions. Weingarten (1988) conducts one of

the first studies on grey divorces by employing qualitative research methods. The author

finds that, among both men and women, having a relationship with children was one of

the most important factors to consider when deciding on whether to divorce after the age

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 623

of 60. Children can be important sources of support for older parents through their

provision of functional, emotional, and other forms of assistance (Brines and Joyner

1999; De Jong Gierveld, van Tilburg, and Dykstra 2016). This may contribute to the

quality of marital unions and thus positively influence their stability. However, as parents

(and children) age, “older children tend to be detrimental to marital stability due to

strained relationships associated with family conflicts, inheritance concerns and other

issues” (Wu and Penning 2018: 4), and such conflicts may jeopardise union stability.

Furthermore, older parents may postpone their silver split until the ‘empty nest’ phase

for the sake of their children since they are no longer responsible for supporting

dependent children (Bair 2007; Hiedemann, Suhomlinova, and O’Rand 1998). However,

other studies have found neither evidence for such a pattern nor an association between

the empty nest phase and grey divorce (Lin et al. 2018).

Less attention has been devoted to the role of grandchildren in shaping the risk of

silver splits. Grandchildren play a central role in later life, even if quantitative data from

longitudinal surveys on the importance of being a grandparent (and its associations with

other demographic events) remain scarce (Hank et al. 2018). Grandparenthood and

grandparental childcare (when not particularly intense) are found to be positively

associated with grandparents’ subjective well-being (Arpino, Bordone, and Balbo 2018).

The birth of the first grandchild is usually associated with feelings of youthfulness for the

grandparents, providing new meaning to life (Cunningham-Burley 1986). Conversely,

the loss of contact with grandchildren (e.g., after separation or divorce) is found to be

associated with reduced psychological well-being given the importance of such ties for

older people (Drew and Silverstein 2007; Drew and Smith 2002). Through examining US

data, Brown, Lin, and Mellencamp (2021) find that becoming a biological grandparent

lowers the likelihood of grey divorce compared to those not experiencing

grandparenthood. However, the relationship between grandparenthood and other late-life

transitions may be context-dependent because of country-specific forms of

intergenerational support, leaving the parental home at different ages, and other social

norms regarding intergenerational ties that differ across birth cohorts (e.g., Aassve,

Arpino, and Billari 2013; Tomassini et al. 2004). Numerous studies (e.g., Uhlenberg and

Hamill 1998) examine the consequences of divorce on grandparent–grandchild relations,

showing how divorced grandparents have less contact with their grandchildren compared

to their married counterparts. King (2003) finds that many aspects of grandparenting were

negatively associated with having experienced a grey divorce; for example, divorced

grandparents were less likely to agree that a valuable part of grandparenthood is the

involvement of grandchildren in their lives. Hence, grandchildren may act as an inhibitor

to silver splits since grandparents assume new responsibilities and, in a certain way, ‘refill

the nest’ after their children’s departure. To conclude, in light of the literature’s findings

on the positive effects of children and grandchildren on individual well-being and union

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

624 https://www.demographic-research.org

stability, we expect having children and grandchildren to be negatively associated to

silver splits.

2.2 Other factors (potentially) related to late union dissolutions

The majority of research on union dissolution has focused on young adults (see Lyngstad

and Jalovaara 2010 for a review). Therefore, when necessary, we cite a number of studies

related to young adults’ union dissolution in order to integrate the scant knowledge on

silver splits, bearing in mind that these results are not directly transferable to union

dissolution after the age of 50.

Birth cohort. Individuals belonging to different birth cohorts – or, similarly, to

different union cohorts – tend to have different values and thus bring different

expectations to their unions, possibly translating (for example) into a higher risk of union

dissolution among younger cohorts (Lyngstad and Jalovaara 2010). The social

acceptance of union dissolution can also differ by birth cohort (e.g., the baby boom

cohort) (see Cohen 2019), which we assume might especially be the case when

considering dissolutions after the age of 50. Brown and Wright (2019) find a sharp

increase in the acceptance of divorce among the baby boom cohorts. In line with the

available evidence on the relationship between birth cohort and union dissolution, we

expect the baby boom cohort to show the highest risk of late union dissolution.

Partnership history. Studies on grey divorce consistently find that divorce rates

decline as marital duration increases (Brown and Lin 2012; Wu and Pennig 1997). The

characteristics of the union (e.g., cohabitation versus marriage, first marriage versus

higher-order marriages) may also contribute to the risk of silver splits. Brown and Lin

(2012) find that the divorce rate was 2.5 times higher for remarriages than first marriages.

Based on Canadian data, Wu and Penning (2018) show that, although nonmarried

cohabiting couples aged 45 and over had on average a ten-year-long union, they still had

a higher risk of dissolution compared to married couples. This finding suggests that,

despite appearances, cohabitation in later life tends not to be as stable as marriage, even

if having biological children (rather than step-children) reduces such an association. Wu

and Penning (2018) address the impact of union and family biography (i.e., marital and

fertility history) on union dissolution in later life, highlighting how their effect on the risk

of grey divorce differs by sex. They stress that “short- and long-term transitions, in turn,

must be addressed within the context of individuals’ cohort experiences as well as their

location within the social structure as indexed by age, gender, and other factors” (Wu and

Penning 2018:3). Union duration is usually included among the main control variables

when studying the determinants of union dissolutions earlier in life due to its strong

relation to the risk of divorce (Jalovaara 2002; Kulu and Boyle 2010). The literature

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 625

shows that the risk of divorce is low in the first months of marriage, after which it rises,

reaches a maximum, and thereafter declines (Kulu 2014). In our study, we expect that

longer-lasting unions are also less likely to dissolve above the age of 50, in line with

previous findings on the generally negative effect of union duration on the risk of union

dissolution.

Educational level. Education may be salient to shaping the risk of silver splits, both

as a proxy for socioeconomic status and through its correlation with earning potential and

labour market activity. Studying grey divorces in Canada, Wu and Penning (1997) find

education to have a positive effect on late divorce among both men and women, while

US-based studies show that educational level has only a limited effect on the probability

of divorce (Brown and Lin 2012; Lin et al. 2018). The majority of studies about (not only

grey) divorces in the United States and in Scandinavian countries report a negative effect

of both spouses’ educational attainments on the risk of divorce (Hoem 1997; Jalovaara

2001; Martin 2006; Ono 1999; Pezzin and Schone 1999), while evidence from the rest of

Europe has been mixed (e.g., Poortman and Kalmijn 2002 [Netherlands]; De Rose 1992,

and Vignoli and Ferro 2009 [Italy]; Blossfeld et al. 1995 [various European countries]).

Generally speaking, the positive educational gradient weakens over time and even turns

negative as divorce diffused and became socially institutionalised (Matysiak, Styrk, and

Vignoli 2014). However, our analyses do not include the recent cohorts who might have

experienced the reverse in the educational gradient of divorce. Accordingly, we expect

to find a positive relationship between high educational level and the risk of a silver split.

Employment condition. How economic factors operate for retired older adults or

those with regular incomes is unclear, and related evidence is scarce. In their study about

divorce after the age of 50 in the United States, Brown and Lin (2012) find that

unemployed and full-time workers are more likely to divorce than those outside of the

labour force and that economic factors figure more prominently in women’s divorce

experiences. In another US-based study, Lin et al. (2018) find that the wife’s or husband’s

retirement is unrelated to grey divorce probability. Studies about divorce at younger ages

show that employment and earnings are negatively related to the risk of divorce (Amato

2010), while others also suggest that the effect of employment and income is ambiguous

among women. When the wife is employed, she increases the family’s total resources,

thus possibly benefiting marital stability (through the ‘income effect’). The wife’s greater

resources might also have a divorce-promoting effect, known as the ‘independence

effect.’ Indeed, rising female employment makes divorce a viable option since

employment provides women with the economic capacity to support themselves outside

of marriage (Bukodi and Robert 2003; Chan and Halpin 2002; Svarer and Verner 2006;

Vignoli et al. 2018). Finally, the role of perceived financial situations is gaining

importance in the literature as a factor associated with late divorce (e.g., Canham et al.

2014). To conclude, based on the available evidence about the relationship between

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

626 https://www.demographic-research.org

employment and (late) union dissolution, we cautiously expect female employment to be

positively related to silver splits (due to the independence effect) while expecting that

male employment may be negatively related to silver splits (in line with traditional gender

roles). However, we do not to make any specific predictions regarding the overall

association between employment and silver splits as investigating the gender-specific

relationship goes beyond the scope of this paper.

Tenure. One’s economic condition is not entirely dependent on employment or

income. Among the various types of assets, housing is often the most significant in the

majority of Western countries. Indeed, homes appear to be the most important

bequeathable token of wealth across the continent – especially for older Europeans

(Angelini, Laferrère, and Weber 2013). For the aged, a housing property provides a

financial buffer against such contingencies as ill health or economic hardships and offers

a nest egg for later life (Gaymu 2003). Despite between-country differences in terms of

state welfare protection, from a strictly economic point of view, exclusion from

homeownership translates into the absence of the most important (and safeguarding) asset

in old age (Vignoli, Tanturri, and Acciai 2016). Because economic and financial stress

may seep into private lives, we surmise that homeownership is negatively associated to

the risk of silver splits.

Health. Research on health and union dissolution mostly focuses on the health-

related consequences of divorce for spouses and their children (Lyngstad and Jalovaara

2010; Tosi and Van den Broek 2020). In fact, health may be prominent in shaping the

risk of silver splits due in large part to the fact that health problems increase with age.

Physical illness, as a stressor on the marital union, may increase divorce risk by reducing

marital quality (Daniel et al. 2009; Yorgason, Booth, and Johnson 2008). Research into

health as a determinant of grey divorces confirms that worsening health deteriorates

marital quality and increases the likelihood of divorce (Booth and Johnson 1994), and

that differences in spouses’ health statuses increase divorce risk (Wilson and Waddoups

2002). Examining a selected sample of couples who were physically healthy at the

beginning of the study, Karraker and Latham (2015) find that only the wives’ illness

onsets are associated with an elevated risk of late union dissolution. Similar studies

focusing on the risk of divorce at younger ages find that individuals with high levels of

psychological well-being are less likely to divorce (Mastekaasa 1994) and that married

persons reporting health complaints or chronic illnesses have an increased divorce risk

(Joung et al. 1998). However, other studies find no association between these factors

(Charles and Stephens 2004). In our study, we explore the role of health as an antecedent

of silver splits in terms of both physical and mental health. Based on previous findings,

we expect that the risk of silver splits is higher for individuals with health problems

compared to their healthy counterparts.

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 627

3. Data and methods

3.1 Sample selection

We used data from the Survey of Health, Ageing and Retirement in Europe (SHARE), a

multidomain longitudinal study that collects detailed information on adults aged 50 and

over and their current partner (if cohabiting), regardless of age. We used waves 1 (2004–

2005) through 7 (2017), with the exception of wave 3, which we excluded due to its

collection of retrospective information and lack of most current sociodemographic and

health variables. Therefore, our analysis is based on six time points. In order to observe

late union dissolutions, we dropped countries that participated in only one wave. We also

discarded countries in which the number of union dissolutions observed throughout the

study period was too small (i.e., fewer than 10). Our sample included respondents from

14 European countries: Austria, Belgium, the Czech Republic, Denmark, Estonia, France,

Germany, Hungary, Italy, the Netherlands, Portugal, Spain, Sweden, and Switzerland.

Differently from previous studies (e.g., Weingarten 1988), we studied union dissolution

rather than divorce in the strict sense of the word to avoid underestimating the

phenomenon. To this end, we restricted our sample to individuals who were (1) married

or in a registered cohabitation or (2) in an informal cohabiting relationship. We included

individuals who were ‘living with a partner’ but not in a formal relationship (i.e., marriage

or registered cohabitation) only if they were assigned a ‘couple ID.’

However, the share

of individuals in registered or informal cohabitations in our dataset was negligible (less

than 2% and 5%, respectively) in terms of the share of married cohabitations

(approximately 93% of the sample). Accordingly, we did not distinguish between

marriages and cohabitations. Married or cohabiting respondents in one wave who

subsequently reported being single or not being in a cohabiting union (i.e., without a

couple ID) in the following wave were considered to have experienced a union

dissolution. All individuals who were not at risk of experiencing union dissolution, such

as older people living without a coresident partner, were excluded from the risk set

(roughly 24% of observations).

The initial sample consisted of 72,032 eligible individuals. To observe union

dissolutions across waves, the dataset included only those individuals interviewed at least

twice between waves 1 and 7. This resulted in the loss of 18,721 individuals (25.9%) by

the follow-up stage. The models included a wide set of control variables, including all

factors that were found to be associated with sample attrition. Using as much information

as possible about selection on available covariates in the data reduces the amount of its

In SHARE, respondents are assigned a ‘couple ID’ if they have a coresiding partner. Accordingly, all

individuals in a union involving cohabitation have a couple ID (including the respondent’s partner, regardless

of age). For further information about SHARE surveying characteristics, see Alcser et al. (2005).

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

628 https://www.demographic-research.org

residual and unexplained variation due to attrition, likely reducing bias due to selection

on observables (Alderman et al. 2001).

Our final sample consisted of 53,311 individuals (26,001 men and 27,310 women)

– of whom 14,198 entered at wave 1; 5,474 at wave 2; 19,804 at wave 4; 11,310 at wave

5; and 2,525 at wave 6 – and all of whom were present in at least one subsequent wave.

Among these, 17,392 individuals who responded to at least two waves did not remain

under observation until the last wave (approximately 32% of the final sample). However,

among the latter, 3,448 left the survey due to their deaths (censored), and 135 left after

having experienced union dissolution (i.e., 18% of our silver splits), while the remaining

13,809 were lost due to attrition. Of this last group, we were able to recover information

about 1,831 respondents because their partners were still under observation (see Figure

A-1 in the Appendix for a flow chart of the sample selection process). We provide an

analysis and discussion of the characteristics associated with attrition in the Appendix

(see paragraph ‘Sample attrition’).

3.2 Variables and imputations

The dependent variable was the experience of divorce or union dissolution for those who

were married or in a relationship (n = 745). Explanatory variables were measured at the

last (observed) wave preceding union dissolution. In some cases (roughly 20% of those

experiencing union dissolutions), one or more explanatory variables were missing at the

wave before union dissolution. To counter this, we exploited the longitudinal nature of

data and recovered missing information from previous waves. If the respondent reported

missing information at wave t–1, we fixed covariates at the previous wave (t–2) or at the

closest wave with nonmissing information.

We employed two main explanatory

variables to study the influence of children and grandchildren. The first focused on the

number of grandchildren and distinguished between childless individuals, those who

have children but no grandchildren, those who have one or two grandchildren, and those

who have three or more grandchildren. The second captured the dimension of the

intensity of family ties by exploiting the information on the number of grandchildren and

the frequency of contact with their children. We therefore assumed that individuals in

regular contact with their children would consequently have regular contact with their

grandchildren. This variable distinguished between childless individuals, those who have

children but no grandchildren, those who have grandchildren but rare contact with their

children, and those who have grandchildren and have frequent contact with their

Among those who experienced a silver split, only 20% of individuals had missing values to one or more

variables at t–1. Among the latter, we imputed roughly 80% from wave t–2. For the rest, missing information

was imputed from waves t–3, t–4, or t–5.

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 629

children.

The set of explanatory variables also included gender, birth cohort (born pre-

1945, between 1945–1955, and post-1955), education level (primary, secondary, tertiary

education), employment status (retired, working, other), union duration (measured as a

continuous variable), previous divorce experiences

(has never divorced versus has

already divorced at least once), homeownership (yes or no), perceived financial distress

(household makes ends meet with great difficulty, with some difficulty, fairly easily,

easily), number of limitations in daily activities (scored from 0 to 6), depression level (a

0-to-12 scale based on the EURO-D depression scale, where 0 is ‘not depressed’ and 12

is ‘very depressed’) (see Prince et al. 1999), country of residence, and the wave in which

the respondent entered the observation. Descriptive statistics are reported in Table 1.

Unfortunately, two of our dataset’s variables were characterised by a non-negligible

number of missing values: union duration (approximately 10% of the sample) and

homeownership (roughly 3%). Importantly, missing information about union duration

showed no specific pattern by gender, country, birth cohort, or socioeconomic status,

whereas it was slightly more frequent in more recent waves. As union duration and

homeownership are key variables in our analysis, eliminating such a large share of

respondents would have remarkably reduced the final number of observations.

Consequently, we decided to retain them in the sample after having imputed the missing

information – which we achieved through multiple imputations by chained equations

(MICE) using STATA (see Lee and Carlin 2010). This technique allows each variable to

be imputed using its own conditional distribution and specifying different models.

Accordingly, we imputed union duration (a continuous variable) using a linear regression

model and a logistic regression for homeownership (a dummy variable). Multiple

imputation estimates several values for each missing data point, thus introducing the

uncertainty associated with the missing data into the model. We then used these values

in the analysis and combined the results following Rubin’s (1987) rule.

The variable about the frequency of contact is available in SHARE with the following categories: ‘coresiding

child(ren),’ ‘daily,’ ‘several times a week,’ ‘about once a week,’ and ‘rarely.’ Contact is considered either

personally, by phone, mail, email, or any other electronic mean during the previous 12 months. We used this

information for the dichotomisation made between those with grandchildren and report having coresiding

children or daily contact with them (‘frequent contact,’ about 60% of the sample), and those who have

grandchildren and report having contact with children several times a week or less (‘weak contact,’ about 40%

of the sample). Nevertheless, given the relatively small number of events, a finer division of the variable into

more categories would have resulted in imprecise effect estimates, leading to inconclusive findings.

We considered only previous divorces and not previous union dissolutions as SHARE collects only

information on the former.

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

630 https://www.demographic-research.org

Table 1: Sample characteristics. N = 53,311

Variable % or mean with SD in brackets

Silver splits 1.40%

Gender

Men 48.77%

Women 51.23%

Birth cohort

<1946 40.72%

1946–1955 36.42%

>1955 22.86%

Educational level

Primary 39.86%

Secondary 37.12%

Tertiary 23.02%

Employment status

Retired 45.26%

Working 36.84%

Other (e.g., unemployed, homemaker) 17.90%

Union duration (in years) 34.81 (12.87)

Has already divorced at least once 2.74%

Homeownership 82.92%

Financial stress

Can make ends meet easily 32.28%

Can make ends meet fairly easily 34.42%

Can make ends meet with some difficulty 24.99%

Can make ends meet with great difficulty 8.31%

Number of limitations with daily activities

No limitations 92.36%

At least one limitation 7.64%

Depression scale 2.23 (2.13)

Number of grandchildren

Childless 5.87%

Has children, no grandchildren 31.67%

Has one or two grandchildren 26.82%

Has three or more grandchildren 35.64%

Grandchildren and intensity of ties

Childless 5.87%

Has children, no grandchildren 31.67%

Has grandchildren, rare contact with children 8.58%

Has grandchildren, frequent contact with children 52.22%

Missing 1.66%

Wave of entrance

Wave 1 26.63%

Wave 2 10.27%

Wave 4 37.15%

Wave 5 21.21%

Wave 6 4.74%

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 631

3.3 Modelling

Using logistic regression, we modelled the probability of experiencing union dissolution,

taking into consideration demographic, socioeconomic, and health-related factors. We

replicated the same model twice in order to test both variables about grandchildren (i.e.,

the number of grandchildren and the intensity of family ties). We clustered the standard

errors at the country level to account for possible correlations in the error terms.

We computed the average marginal effects (AMEs) to facilitate substantive

interpretations. AMEs express the effect on P(Y = 1) as a categorical covariate changes

between categories or as a continuous covariate increases by 1 unit, averaged across the

values of the other covariates included in the model equations. In some instances, we also

present predicted probabilities with 95% confidence intervals for pairwise comparisons.

These intervals are centred on the predictions and have lengths equal to

2 × 1.39 × standard errors – which is necessary for achieving an average level of 5% for

Type I errors in pairwise comparisons of a group of means (Goldstein and Healy 1995).

After presenting some descriptive findings about country differences in late union

dissolution, we then scrutinise the relationship between children, grandchildren, and the

risk of a silver split before investigating the roles played by the other (potential)

correlates.

4. Results

4.1 Descriptive findings: Country differences in late union dissolution

Figure 1 displays the (adjusted) predicted probability of silver split by country. Denmark

had the highest (2.17%), followed by Sweden and Austria (2.13% and 1.98%,

respectively). Estonia, Spain, Belgium, and Switzerland also showed above-average

probabilities of late union dissolution (1.35%). Central-Eastern European countries (the

Czech Republic, Slovakia, and Hungary) and the other Southern European countries

(Italy and Portugal) had lower probabilities of union dissolution together with France,

Germany, and the Netherlands. We found Italy to have the lowest probability of union

dissolution after the age of 50 (0.59%). Generally speaking, late union dissolutions seem

to be a rare demographic phenomenon; however, the countries analysed displayed

interesting variability. Unfortunately, although we pooled six waves, the small number

of silver splits registered in each country prohibited any country-specific (nor country-

group-specific) analyses.

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

632 https://www.demographic-research.org

Figure 1: Adjusted predicted probabilities of union dissolution by country

Source: Authors’ elaboration on SHARE data, waves 1–7 (wave 3 excluded).

Notes: Silver split probabilities are adjusted by gender, birth cohort, education level, employment status, union duration, having children

and grandchildren, previous divorce experiences, homeownership, perceived financial stress, depression level, limitations in daily

activities, and entrance wave. Predicted probabilities refer to the population average.

4.2 The role of children and grandchildren for silver splits

Our primary focus is the exploration of the role played by grandchildren in shaping silver

split behaviours. Table 2 reports AMEs of the children and grandchildren variables from

logistic regression models of the probability of experiencing a late union dissolution (the

remaining variables are discussed in Table 3; see Section 4.3). Model 1 includes the

variable about the number of grandchildren, while Model 2 includes the variable

considering the intensity of family exchanges. Both models include identical control

variables. For Model 1, we found that individuals with children – even if without

grandchildren – have a lower risk of late union dissolution compared to childless

individuals (AME = –0.0038, p-value = 0.101). Moreover, having grandchildren seems

related to a further decrease in the probability of experiencing late union dissolution, with

people having one or two grandchildren and those having three or more being less likely

to experience a silver split by 0.65 and 0.79 percentage points (pp), respectively. When

2.17%

2.13%

1.98%

1.66%

1.64%

1.46%

1.38%

1.12%

1.04%

1.00%

0.85%

0.59%

1.35%

0.00%

0.50%

1.00%

1.50%

2.00%

2.50%

Predicted probability of union dissolution

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 633

setting individuals with children but no grandchildren as the reference category (not

shown in Table 2), we found that having one or two grandchildren decreases the risk of

a silver split by 0.27 pp (p-value = 0.121), and having three or more reduces the risk by

0.41 pp (p-value = 0.007). The results from Model 2 (i.e., considering the intensity of

family exchanges) suggest that the strength of the negative association between having

grandchildren and the risk of late union dissolution is stronger for individuals with

grandchildren and frequent contact with their children. We found that the respondents

with grandchildren have lower chances of experiencing a silver split by 0.57 pp, even if

they reported having only infrequent contact with their children, and by 0.77 pp if they

have regular contact with their children. This indicates that having grandchildren relates

negatively not only to late union dissolution but also the intensity of family exchanges.

Switching the reference category to people with children but no grandchildren, we found

that those who have grandchildren and are in infrequent contact with their children are

less likely to experience a silver split by 0.18 pp (p-value = 0.481), while those in frequent

contact with their children are less likely to experience a silver split by 0.37 pp (p-value

= 0.021).

Next, we explored if and how the relationship between having grandchildren and

silver splits changes across birth cohorts. We followed Brown and Wright (2019), who

suggest that cohorts tend to have different attitudes towards grey divorce. This begs the

question of whether grandchildren may (partly) explain such variation. We replicated the

analysis shown in the previous paragraph but added an interaction term between the birth

cohort and each of the two variables about grandchildren. The results concerning the

other explanatory variables remained virtually unchanged to those reported in Table 2.

Accordingly, we show only the results of the interactions. We calculated the predicted

probabilities of late union dissolution for different birth cohorts separately for each

category described by the two variables about grandchildren.

Panel A of Figure 2 shows the predicted probability with confidence intervals of late

union dissolution for childless individuals, those with children but no grandchildren,

those with one or two grandchildren, and those with three or more. Panel B shows the

predictions according to the variable about the intensity of family exchanges. First, both

figures clearly show that the probability of experiencing a silver split increases, on

average, among recent cohorts – especially among those born post-1955. Regarding

grandchildren (Panel A), the results suggest that the role of grandchildren in shaping

divorce behaviours has become increasingly important in younger generations. There is

a clear gradient among individuals born between 1946–1955 and those born after 1955,

suggesting that individuals with grandchildren are less prone to dissolve their unions after

the age of 50, especially if they have three or more grandchildren. For example, for those

born after 1955, the probability of late union dissolution for childless individuals was

approximately 2.6%; for individuals with children but no grandchildren it was 2.4%; and

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

634 https://www.demographic-research.org

it was roughly 1.8% for those with one or two grandchildren and 1.6% for three or more

grandchildren. Such a gradient is slightly less evident in the 1946–1955 cohort, even if

the difference in the probability of silver splits between childless individuals and those

with more than two grandchildren is relatively large. These differences virtually

disappear in the oldest cohort (i.e., those born before 1946). In the oldest cohort, the

(predicted) probability of silver splits is much smaller than for the other cohorts and

remains virtually unchanged among the respondents with children, with one or two

grandchildren, and with three or more grandchildren. Panel B shows that the intensity of

family exchanges is a particularly relevant dimension for individuals from the older

cohorts. Indeed, the risk of silver splits is higher among individuals with grandchildren

who report having rare contact with their children (0.008), as compared to both

individuals with children but no grandchildren (0.005) and those who have grandchildren

and frequent contact with their children (0.004). Among the two younger cohorts, our

findings confirmed that having grandchildren is linked to a decrease in the probability of

experiencing late union dissolution (although the estimates have a low statistical

precision).

Table 2: Logistic model for the probability of experiencing union dissolution

after the age of 50. AMEs for the variables about children and

grandchildren are reported. N = 53,311

AME

value

Model 1

number of grandchildren (ref. childless)

has children, no grandchildren

–

0.0038

0.101

has one or two grandchildren

–

0.0065

0.006

has three or more grandchildren

–

0.0079

0.000

Model 2

contact with grandchildren (ref. childless)

has children, no grandchildren

–

0.0038

0.100

has children and grandchildren but rare contact

–

0.0057

0.065

has children and grandchildren with frequent contact

–

0.0077

0.000

Source: Authors’ elaboration on SHARE data, waves 1–7 (wave 3 excluded).

Note: The models control for gender, birth cohort, educational level, employment status, union duration, previous divorce experience,

homeownership, perceived financial stress, limitations with daily activities, depression scale, wave of entrance, and country of

residence.

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 635

Figure 2: Adjusted predicted probability of union dissolution by number of

grandchildren and birth cohort. Confidence intervals are reported

(A)

(B)

Source: Authors’ elaboration on SHARE data, waves 1–7 (wave 3 excluded).

Notes: Predicted probabilities are adjusted by gender, education level, employment status, union duration, previous divorce

experiences, homeownership, perceived financial stress, depression level, limitations in daily activities, and entrance wave. Predicted

probabilities refer to the population average.

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

636 https://www.demographic-research.org

4.3 The other correlates of silver splits

Table 3 illustrates the relationship between the other variables and the risk of silver splits.

As expected, we found no difference by gender (AME = 0.0005, p-value = 0.505).

Individuals born between 1946–1955 and post-1955 (i.e., the baby boom cohorts) are

more likely to experience union dissolution than the oldest cohort (pre-1946). The model

highlighted no remarkable differences in the probability of silver splits according to

educational level (robust to different specifications of the education variable).

Regarding

employment status, retired individuals are more likely to experience a silver split than

those working or otherwise not retired. Our findings confirm that union duration is

negatively related to late union dissolution, meaning that the longer the marital duration,

the smaller the probability of experiencing a silver split. Previous divorce experiences

also play an important role in shaping silver split probability, with those who have already

divorced at least once being 9.95 pp more likely to experience a/another dissolution

compared to first-time divorcers. Homeownership is negatively related to silver split

probability, with a related AME of –0.0067 (p-value < 0.001). Regarding perceived

financial stress, we found that people who could make ends meet with difficulty or with

great difficulty had a higher probability of late union dissolution. Interestingly, regarding

health, we noted different results depending on the sphere of health considered. The

indicator for functional health revealed that a higher number of limitations in daily

activities related to a lower probability of late union dissolution (AME = –0.0020),

whereas we observed an opposing correlation for depression (AME = 0.0011). Finally,

those who entered the survey in the latest waves had a lower probability of late dissolution

(possibly due to their spending less time in observation).

We also tested education in the model as a binary variable with two different specifications: primary education

versus secondary or tertiary education, and primary or secondary education versus tertiary education.

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 637

Table 3: Logistic model for the probability of experiencing union dissolution

after the age of 50. AMEs are reported. N = 53,311

AME

value

gender (ref. male)

female

0.0005

0.505

birth

cohort (ref. before 1945)

1946

–

1955

0.0086

0.000

after

1955

0.0078

0.000

education (ref. primary)

secondary

–

0.0017

0.465

tertiary

0.0013

0.556

employment status (ref. retired)

working

–

0.0070

0.000

other (

e.g.,

unemployed, homemaker)

–

0.0093

0.000

union duration

–

0.0006

0.000

has already divorced at least once (ref. no)

yes

0.0995

0.000

homeowner

ship (ref. no)

yes

–

0.0067

0.000

making ends meet (ref. easily)

fairly easily

0.0024

0.214

with some difficulty

0.0041

0.017

with

great difficulty

0.0077

0.013

number of limitations with daily activities

–

0.0020

0.073

depression scale

0.0011

0.000

wave of entrance (ref. wave 1)

wave 2

0.0006

0.861

wave 4

–

0.0029

0.049

wave 5

–

0.0062

0.004

wave 6

–

0.0125

0.000

number of

grandchildren

YES

ountry fixed effects

YES

Source: Authors’ elaboration on SHARE data, waves 1–7 (wave 3 excluded).

4.4 Robustness checks and further analyses

Our findings are confirmed across various additional analyses and robustness checks.

Due to space constraints, the results are not shown here but are available upon request.

First, we added an interaction term between the number of children and the country of

residence in order to consider possible country-level differences but found no relevant

result (quite possibly due to reduced sample size). Moreover, to operationalise the ‘empty

nest’ concept, we also tested interaction terms between the number of grandchildren and

the fact that none, some, or all children had left the parental home. However, the cells

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

638 https://www.demographic-research.org

became too small and yielded inconclusive results. Regarding the role of children, we

conducted a specific analysis to check whether the number of children was also

significant and found that individuals with two or more children are less likely to

experience silver splits than those with only one child. Additionally, since we used

multiple imputation techniques to manage missing information, all models estimated on

the imputed dataset were replicated on the original dataset (i.e., without imputations).

While the estimates remained virtually unchanged, they clearly lost part of their statistical

precision. Finally, we replicated the analysis excluding individuals whose missing

information was recovered from previous waves. Again, the results remain virtually

unchanged but lost part of their statistical precision due to the reduced sample size.

5. Conclusions

Union dissolution in later life has become increasingly relevant as a social and

demographic phenomenon. Despite this growing importance, the correlates of silver

splits remain underexplored – especially in Europe. Using data from six waves of the

SHARE dataset, we explored the role of several factors as potential correlates of late

union dissolution, with a special emphasis on the role of children and grandchildren. We

studied union dissolution rather than divorce in the strict sense of the word to

acknowledge the increasing diversity of family life at older ages.

Our analysis suggests that having children and grandchildren is associated with a

lower probability of experiencing a silver split. Indeed, we found that late union

dissolution is less likely when individuals have children, let alone grandchildren. This

finding is unsurprising and aligns with prior research showing that having children is a

well-established factor that consolidates unions among the analysed birth cohorts (e.g.,

De Rose 1992; Hoem and Hoem 1992; Lyngstad and Jalovaara 2010; White 1990), and

that family-oriented individuals are both more likely to have children and less likely to

dissolve their union (Lesthaeghe and Moors 2000). Becker, Landes, and Michae (1977)

observe that children are a ‘marital-specific capital,’ thus representing a sign of family

harmony with positive implications for union stability. However, it remains to be seen

whether this association will be confirmed in future generations.

The effect of grandchildren in shaping silver splits is especially interesting. To

explore this effect, we employed two variables (the number of grandchildren and the

intensity of family exchanges) and explored their relationship with the risk of a silver

split. Our findings indicate that having grandchildren reduces this risk and that both their

number and the frequency of contact with children and grandchildren play a significant

role. Not only having grandchildren but also maintaining frequent contact with them

relates to a lower probability of late union dissolution. While it has been widely

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 639

established that having children is a strong inhibitor to divorce at younger ages (i.e., when

children are young), as children (and parents) age, this relationship may weaken. At later

ages, having grandchildren – in addition to that of children – and the intensity of family

exchanges influence the probability of union dissolution, possibly due to the positive

relationship between becoming grandparents and individual well-being (e.g., Arpino,

Bordone, and Balbo 2018). Interestingly, this correlation differs across birth cohorts. On

the one hand, younger cohorts have higher divorce rates than their older counterparts, and

childless individuals are always the most likely to experience late union dissolutions. On

the other hand, our findings suggest that, compared to individuals who only have

children, those who have also grandchildren have an even lower risk of silver splits –

especially among those born after 1946. This result may be explained by the fact that

more recent cohorts have young grandchildren whose grandparents are more involved in

childcare (which may be a form of positive engagement for the couple) compared to those

with older grandchildren who are less in need of care (Pasqualini, Di Gessa, and

Tomassini 2021). However, such an interpretation should be made cautiously since the

timing of fertility and childcare attitudes have changed over time and differ between

countries.

As an ancillary but important outcome of the study, we suggest that the European

correlates of silver splits are similar to those found previously in a North American

context. Regarding birth cohorts, we found that individuals born after 1946 were more

likely to experience late union dissolutions than those born previously. This aligns with

previous findings in the United States showing that the baby boom cohorts are the most

prone to divorce and sparked the silver split phenomenon as they aged (Brown and Lin

2012; Cohen 2019; Lin et al. 2018). We confirmed union duration to be negatively related

to silver splits, while previous divorce experiences proved to be important predictors of

silver splits at later ages. Retired individuals have a higher risk of experiencing union

dissolution after the age of 50 compared to those not retired. Regarding economic

conditions, our findings suggest a positive correlation between financial stress and silver

splits since individuals who struggle to make ends meet are more likely to dissolve their

unions after the age of 50, and homeowners are less likely to experience a silver split

compared to those who do not own a house. The latter finding suggests that a stable

housing situation may improve marital quality in later life, which accords with previous

research on the importance of housing for the aged (Angelini, Laferrère, and Weber 2013;

Vignoli, Tanturri, and Acciai 2016). Finally, our study supports the hypothesis of a

negative relationship between deteriorating mental health (i.e., depression) and late union

dissolution, thereby corroborating previous findings (Davies, Avison, and McAlpine

1997; Idstad et al. 2015; Kessler, Walters, and Forthofer 1998; Torvik et al. 2015).

Conversely, we found that poor physical health – measured through the number of

limitations to daily activities – was associated with a reduced likelihood of union

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

640 https://www.demographic-research.org

dissolution. Education level seemed irrelevant to silver splits. This is in line with existing

studies showing that educational level has only a limited effect on the probability of grey

divorce in the United States (Brown and Lin 2012; Lin et al. 2018).

It would be important to note several limitations to our study. First, the small number

of cases prohibited a country-specific analysis. This means that our findings reveal

average effects computed across several countries and thus obscure any potential

country-specific patterns. Furthermore, it is possible that we found no association

between some factors (e.g., education) and silver splits simply because opposing country-

specific effects averaged out. Another limitation relates to attrition. Different solutions

have been proposed on how to most effectively control for attrition, depending on the

mechanisms generating loss at follow-up (see e.g., Enders 2010; Little and Rubin 2002).

Although attrition effects are present in most panel surveys to various extents, their

consequences on model results are often disregarded in demographic research, which

may lead to a non-negligible bias (Alderman et al. 2001). Despite our efforts to include

a wide array of control variables in the models in order to mitigate attrition bias, such a

solution can reduce the consequences of attrition only to the extent that it depends on

observable characteristics. Nevertheless, it is worth noting that previous analyses have

found little evidence of selective attrition bias in SHARE (Bergmann et al. 2017; Kneip,

Malter, and Sand 2015). We also encountered certain data limitations. We were unable

to follow a couple approach because a substantial number of participants had (totally or

partially) nonresponding partners, who would have introduced a further selection in our

analyses. Besides, due to the reduced number of cases, we made no distinctions between

biological children and step-children, nor did we do so for biological grandchildren and

step-grandchildren. We found inconsistencies in marital and partner status across waves

(e.g., individuals who were married in one wave and reported being single in the

following one). Furthermore, approximately 5,000 individuals reported missing

information about their union duration, upon which we opted to impute these missing

values. Finally, we inferred family intensity through information concerning the

frequency of contact with children, which may possibly have introduced a bias.

Unfortunately, information about the frequency of contact with grandchildren was not

available, and data regarding whether the respondent cares for grandchildren were

accessible only if the latter was younger than 13 years of age.

Despite these limitations, this paper sheds some light on the correlates of a rare but

demographically and sociologically relevant phenomenon. Having children has been

widely shown to inhibit union dissolution – especially among older cohorts. Indeed,

parents may postpone their marital dissolution until after the ‘empty nest’ phase for the

sake of their children. However, grandchildren may serve to ‘refill the nest,’ thus

discouraging silver splits as grandparents assume new familial and social responsibilities.

Although predominantly exploratory, our study expands existing knowledge on the

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 641

factors related to silver splits in Europe and, we hope, will feed future research on the

topic (e.g., scrutinising the role of children and grandchildren specifically by country and

gender, or investigating potential differences between biological grandchildren and step-

grandchildren).

6. Acknowledgements

This paper uses data from SHARE Waves 1, 2, 4, 5, 6, and 7 (DOIs: 10.6103/

SHARE.w1.800, 10.6103/SHARE.w2.800, 10.6103/SHARE.w4.800, 10.6103/SHARE.

w5.800, 10.6103/SHARE.w6.800, 10.6103/SHARE.w7.800), see Börsch-Supan et al.

(2013) for methodological details. The SHARE data collection has been funded by the

European Commission, DG RTD through FP5 (QLK6-CT-2001-00360), FP6 (SHARE-

I3: RII-CT-2006-062193, COMPARE: CIT5-CT-2005-028857, SHARELIFE: CIT4-

CT-2006-028812), FP7 (SHARE-PREP: GA N°211909, SHARE-LEAP: GA N°227822,

SHARE M4: GA N°261982, DASISH: GA N°283646) and Horizon 2020 (SHARE-

DEV3: GA N°676536, SHARE-COHESION: GA N°870628, SERISS: GA N°654221,

SSHOC: GA N°823782, SHARE-COVID19: GA N°101015924) and by DG

Employment, Social Affairs & Inclusion through VS 2015/0195, VS 2016/0135, VS

2018/0285, VS 2019/0332, and VS 2020/0313. Additional funding from the German

Ministry of Education and Research, the Max Planck Society for the Advancement of

Science, the U.S. National Institute on Aging (U01_AG09740-13S2, P01_AG005842,

P01_AG08291, P30_AG12815, R21_AG025169, Y1-AG-4553-01, IAG_BSR06-11,

OGHA_04-064, HHSN271201300071C, RAG052527A) and from various national

funding sources is gratefully acknowledged (see www.share-project.org).

The authors acknowledge the financial support provided by the Italian Ministry of

University and Research under the 2017 MiUR-PRIN Grant Prot. N. 2017W5B55Y

(“The Great Demographic Recession,” PI: Daniele Vignoli). The authors would also like

to thank the project “Care, Retirement and Wellbeing of Older People Across Different

Welfare Regimes” (CREW), and the colleagues from the Unit of Population and Society

(UPS) of the University of Florence for their much-valued comments.

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

642 https://www.demographic-research.org

References

Aassve, A., Arpino, B., and Billari, F.C. (2013). Age norms on leaving home: Multilevel

evidence from the European Social Survey. Environment and Planning A 45(2):

383–401. doi:10.1068/a4563.

Alcser, K.H., Benson, G., Börsch-Supan, A., Brugiavini, A., Christelis, D., Croda, E.,

and Weerman, B. (2005). The survey of health, aging, and retirement in Europe –

Methodology. Mannheim Mannheim Research Institute for the Economics of

Aging.

Alderman, H., Behrman, J.R., Kohler, H.-P., Maluccio, J.A., and Watkins, S.C. (2001).

Attrition in longitudinal household survey data: Some tests for three developing-

country samples. Demographic Research 5(4): 79–124. doi:10.4054/DemRes.

2001.5.4.

Amato, P.R. (2010). Research on divorce: Continuing trends and new developments.

Journal of Marriage and the Family 72(3): 650–666. doi:10.1111/j.1741-

3737.2010.00723.x.

Angelini, V., Laferrère, A., and Weber, G. (2013). Home-ownership in Europe: How did

it happen? Advances in Life Course Research 18(1): 83–90. doi:10.1016/j.alcr.

2012.10.006.

Arpino, B., Bordone, V., and Balbo, N. (2018). Grandparenting, education and subjective

well-being of older Europeans. European Journal of Ageing 15(3): 251–263.

doi:10.1007/s10433-018-0467-2.

Bair, D. (2007). Calling it quits: Late life divorce and starting over. New York, NY:

Random House

Becker, G.S., Landes, E.M., and Michael, R.T. (1977). An economic analysis of marital

instability. Journal of Political Economy 85(6): 1141–1188. doi:10.1086/260631.

Berger, P. and Kellner, H. (1964). Marriage and the construction of reality: An exercise

in the microsociology of knowledge. Diogenes 12(46): 1–24. doi:10.1177/0392

19216401204601.

Bergmann, M., Kneip, T., De Luca, G., and Scherpenzeel, A. (2017). Survey participation

in the survey of health, ageing and retirement in Europe (SHARE), Wave 1–6.

Munich: Munich Center for the Economics of Aging.

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 643

Bernardi, L., Huinink, J., and Settersten Jr, R.A. (2019). The life course cube: A tool for

studying lives. Advances in Life Course Research 41: 100258. doi:10.1016/j.alcr.

2018.11.004.

Blossfeld, H.-P., de Rose, A., Hoem, J.M., and Rohwer, G. (1995). Education,

modernization, and the risk of marriage disruption in Sweden, West Germany, and

Italy. In: Oppenheim Mason, K. and Jensen, A.-M. (eds.). Gender and family

change in industrialized countries. Oxford: Clarendon Press: 200–222.

Börsch-Supan, A., Brandt, M., Hunkler, C., Kneip, T., Korbmacher, J., Malter, F.,

Schaan, B., Stuck, S., and Zuber, S. (2013). Data resource profile: The Survey of

Health, Ageing and Retirement in Europe (SHARE). International Journal of

Epidemiology 42(4): 992–1001. doi:10.1093/ije/dyt088.

Booth, A. and Johnson, D. R. (1994). Declining health and marital quality. Journal of

Marriage and the Family 56(1): 218–223. doi:10.2307/352716.

Brines, J. and Joyner, K. (1999). The ties that bind: Principles of cohesion in cohabitation

and marriage. American Sociological Review 64(3): 333–355. doi:10.2307/2657

490.

Brown, S.L. and 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. doi:10.1093/geronb/

gbs089.

Brown, S.L. and Wright, M.R. (2019). Divorce attitudes among older adults: Two

decades of change. Journal of Family Issues 40(8): 1018–1037. doi:10.1177/

0192513X19832936.

Brown, S.L., Lin, I.F., and Mellencamp, K.A. (2021). Does the transition to

grandparenthood deter gray divorce? A test of the braking hypothesis. Social

Forces 99(3): 1209–1232. doi:10.1093/sf/soaa030.

Bukodi, E. and Robert, P. (2003). Union disruption in Hungary. International Journal of

Sociology 33(1): 64–94. doi:10.1080/15579336.2003.11770264.

Canham, S.L., Mahmood, A., Stott, S., Sixsmith, J., and O’Rourke, N. (2014). ‘Til

divorce do us part: Marriage dissolution in later life. Journal of Divorce and

Remarriage 55(8): 591–612. doi:10.1080/10502556.2014.959097.

Chan, T.W. and Halpin, B. (2002). Union dissolution in the United Kingdom.

International Journal of Sociology 32(4): 76–93. doi:10.1080/15579336.2002.11

770260.

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

644 https://www.demographic-research.org

Charles, K.K. and Stephens, Jr, M. (2004). Job displacement, disability, and divorce.

Journal of Labor Economics 22(2): 489–522. doi:10.1086/381258.

Cohen, P.N. (2019). The coming divorce decline. Socius 5. doi:10.1177/237802311

9873497.

Crowley, J.E. (2019). Does everything fall apart? Life assessments following a gray

divorce. Journal of Family Issues 40(11): 1438–1461. doi:10.1177/0192513X1

9839735.

Cunningham-Burley, S. (1986). Becoming a grandparent. Ageing and Society 6(4): 453–

470. doi:10.1017/S0144686X00006267.

Daniel, K., Wolfe, C.D., Busch, M.A., and McKevitt, C. (2009). What are the social

consequences of stroke for working-aged adults? A systematic review. Stroke

40(6): e431–e440. doi:10.1161/STROKEAHA.108.534487.

Davies, L., Avison, W.R., and McAlpine, D.D. (1997). Significant life experiences and

depression among single and married mothers. Journal of Marriage and the

Family 59(2): 294–308. doi:10.2307/353471.

De Jong Gierveld, J., van Tilburg, T.G., and Dykstra, P.A. (2016). Loneliness and social

isolation. In: Vangelisti, A. and Perlman, D. (eds.). The Cambridge handbook of

personal relationships. Cambridge, MA: Cambridge University Press: 485–500.

doi:10.1017/CBO9780511606632.027.

De Rose, A. (1992). Socio-economic factors and family size as determinants of marital

dissolution in Italy. European Sociological Review 8(1): 71–91. doi:10.1093/

oxfordjournals.esr.a036623.

De Shane, M.R. and Brown-Wilson, K. (1982). Divorce in late life: A call for research.

Journal of Divorce 4(4): 81–91. doi:10.1300/J279v04n04_06.

Drew, L.M. and Silverstein, M. (2007). Grandparents’ psychological well-being after loss

of contact with their grandchildren. Journal of Family Psychology 21(3): 372.

doi:10.1037/0893-3200.21.3.372.

Drew, L.M. and Smith, P.K. (2002). Implications for grandparents when they lose contact

with their grandchildren: Divorce, family feud and geographical separation.

Journal of Mental Health and Aging 8: 95–120.

Elder Jr, G.H. (1994). Time, human agency, and social change: Perspectives on the life

course. Social Psychology Quarterly 57(1): 4–15. doi:10.2307/2786971.

Enders, C.K. (2010). Applied missing data analysis. Guilford Press.

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 645

Esterberg, K.G., Moen, P., and Dempster‐McCain, D. (1994). Transition to divorce: A

life‐course approach to women’s marital duration and dissolution. Sociological

Quarterly 35(2): 289–307. doi:10.1111/j.1533-8525.1994.tb00411.x.

Gaymu, J. (2003). The housing conditions of elderly people. Genus 59(1): 201–226.

Glenn, N.D. and Supancic, M. (1984) The social and demographic correlates of divorce

and separation in the United States: An update and reconsideration. Journal of

Marriage and Family 46(3): 563–575. doi:10.2307/352598.

Goldstein, H. and Healy, M.J. (1995). The graphical presentation of a collection of

means. Journal of the Royal Statistical Society: Series A (Statistics in Society)

158(1): 175–177. doi:10.2307/2983411.

Hank, K., Cavrini, G., Di Gessa, G., and Tomassini, C. (2018). What do we know about

grandparents? Insights from current quantitative data and identification of future

data needs. European Journal of Ageing 15(3): 225–235. doi:10.1007/s10433-

018-0468-1.

Hiedemann, B., Suhomlinova, O., and O’Rand, A.M. (1998). Economic independence,

economic status, and empty nest in midlife marital disruption. Journal of

Marriage and Family 60(1): 219–231. doi:10.2307/353453.

Hoem, B. and Hoem, J.M. (1992). Disruption of marital and non-marital unions in

Sweden. In: Trussell, J., Hankinson, R., and Tilton, J. (eds.). Demographic

applications of event history analysis. Oxford: Clarendon Press.

Hoem, J.M. (1997). Educational gradients in divorce risks in Sweden in recent decades.

Population Studies 51(1): 19–27. doi:10.1080/0032472031000149696.

Hoem, J.M. and Kreyenfeld, M. (2006). Anticipatory analysis and its alternatives in life-

course research. Part 1: The role of education in the study of first

childbearing. Demographic Research 15(16): 461–484. doi:10.4054/DemRes.

2006.15.16.

Idstad, M., Torvik, F.A., Borren, I., Rognmo, K., Røysamb, E., and Tambs, K. (2015).

Mental distress predicts divorce over 16 years: the HUNT study. BMC Public

Health 15(320). doi:10.1186/s12889-015-1662-0.

Jalovaara, M. (2001). Socio-economic status and divorce in first marriages in Finland

1991–93. Population Studies 55(2): 119–133. doi:10.1080/00324720127685.

Jalovaara, M. (2002). Socioeconomic differentials in divorce risk by duration of

marriage. Demographic Research 7(16): 537–564. doi:10.4054/DemRes.2002.

7.16.

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

646 https://www.demographic-research.org

Joung, I.M., Van De Mheen, H.D., Stronks, K., Van Poppel, F.W., and Mackenbach, J.P.

(1998). A longitudinal study of health selection in marital transitions. Social

Science and Medicine 46(3): 425–435. doi:10.1016/S0277-9536(97)00186-X.

Karraker, A. and Latham, K. (2015) In sickness and in health? Physical illness as a risk

factor for marital dissolution in later life. Journal of Health and Social Behavior

56(3): 420–435. doi:10.1177/0022146515596354.

Kennedy, S. and Ruggles, S. (2014). Breaking up is hard to count: The rise of divorce in

the United States, 1980–2010. Demography 51(2): 587–598. doi:10.1007/s13524-

013-0270-9.

Kessler, R.C., Walters, E.E., and Forthofer, M.S. (1998) The social consequences of

psychiatric disorders, III: Probability of marital stability. American Journal of

Psychiatry 155(8): 1092–1096. doi:10.1176/ajp.155.8.1092.

King, V. (2003), The legacy of a grandparent’s divorce: Consequences for ties between

grandparents and grandchildren. Journal of Marriage and Family 65(1): 170–183.

doi:10.1111/j.1741-3737.2003.00170.x.

Kneip, T., Malter, F., and Sand, G. (2015). Fieldwork monitoring and survey

participation in fifth wave of SHARE. SHARE wave 5: 102–159.

Kulu, H. and Boyle, P.J. (2010). Premarital cohabitation and divorce: Support for the

“Trial Marriage” Theory? Demographic Research 23(31): 879–904. doi:10.4054/

DemRes.2010.23.31.

Kulu, H. (2014). Marriage duration and divorce: The seven-year itch or a lifelong itch?

Demography 51(3): 881–893. doi:10.1007/s13524-013-0278-1.

Lee, K.J. and Carlin, J.B. (2010). Multiple imputation for missing data: Fully conditional

specification versus multivariate normal imputation. American Journal of

Epidemiology 171(5): 624–632. doi:10.1093/aje/kwp425.

Lesthaeghe, R. and Moors, G.B.D. (2000). Recent trends in fertility and household

formation in the industrialised world. Review of Population and Social Policy 9:

121–170.

Lin, I.F., Brown, S., Wright, M., and Hammersmith, A. (2018) Antecedents of gray

divorce: A life course perspective. The Journals of Gerontology Series B:

Psychological Sciences and Social Sciences 73(6): 1022–1031.

Little, R.J. and Rubin, D.B. (2002). Bayes and multiple imputation. In: Statistical

analysis with missing data. Hoboken: John Wiley and Sons, Inc.: 200–220.

doi:10.1002/9781119013563.ch10.

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 647

Lyngstad, T.H. and Jalovaara, M. (2010) A review of antecendents of union dissolution

Demographic Research 23(10): 257–292. doi:10.4054/DemRes.2010.23.10.

Martin, S.P. (2006). Trends in marital dissolution by women’s education in the United

States. Demographic Research 15(20): 537–560. doi:10.4054/DemRes.2006.

15.20.

Mastekaasa, A. (1994). Psychological well-being and marital dissolution – Selection

effects. Journal of Family Issues 15(2): 208–228. doi:10.1177/0192513X940

15002004.

Matysiak, A., Styrc, M., and Vignoli, D. (2014). The educational gradient in marital

disruption: A meta-analysis of European research findings. Population Studies

68(2): 197–215. doi:10.1080/00324728.2013.856459.

Office for National Statistics (2017). Marriage and divorce on the rise at 65 and over.

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriag

es/marriagecohabitationandcivilpartnerships/articles/marriageanddivorceontheris

eat65andover/2017-07-18.

Ono, H. (1999). Historical time and US marital dissolution. Social Forces 77(3): 969–

999. doi:10.2307/3005968.

Pasqualini, M., Di Gessa, G., and Tomassini, C. (2021). A change is (not) gonna come:

a 20-year overview of Italian grandparent–grandchild exchanges. Genus 77(1): 1–

25. doi:10.1186/s41118-021-00142-6.

Pezzin, L.E. and Schone, B.S. (1999). Parental marital disruption and intergenerational

transfers: an analysis of lone elderly parents and their children. Demography

36(3): 287–297. doi:10.2307/2648053.

Poortman, A.R. and Kalmijn, M. (2002). Women’s labour market position and divorce

in the Netherlands: Evaluating economic interpretations of the work effect.

European Journal of Population/Revue europeenne de demographie 18(2): 175–

202. doi:10.1023/A:1015520411449.

Prince, M.J., Beekman, A.T., Deeg, D.J., Fuhrer, R., Kivela, S.L., Lawlor, B.A., and

Reischies, F. (1999). Depression symptoms in late life assessed using the EURO–

D scale. The British Journal of Psychiatry 174(4): 339–345. doi:10.1192/bjp.

174.4.339.

Raley, R.K. and Sweeney, M.M. (2020). Divorce, repartnering, and stepfamilies: A

decade in review. Journal of Marriage and Family 82(1): 81–99. doi:10.1111/

jomf.12651.

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

648 https://www.demographic-research.org

Rotz, D. (2016). Why have divorce rates fallen? The role of women’s age at marriage.

Journal of Human Resources 51(4): 961–1002. doi:10.3368/jhr.51.4.0214-6224R.

Rubin, D.B. (1987). Multiple imputation for nonresponse in surveys. New York: Wiley.

doi:10.1002/9780470316696.

Solaz, A. (2021). More frequent separation and repartnering among people aged 50 and

over. Population Societies (2): 1–4.

Svarer, M. and Verner, M. (2006). Do children stabilize Danish marriages? Journal of

Population Economics 21(2): 395–417. doi:10.1007/s00148-006-0084-9.

Tomassini, C., Kalogirou, S., Grundy, E., Fokkema, T., Martikainen, P., Van Groenou,

M.B., and Karisto, A. (2004). Contacts between elderly parents and their children

in four European countries: current patterns and future prospects. European

Journal of Ageing 1(1): 54–63. doi:10.1007/s10433-004-0003-4.

Torvik, F.A., Gustavson, K., Røysamb, E., and Tambs, K. (2015) Health, health

behaviors, and health dissimilarities predict divorce: results from the HUNT

study. BMC Psychology 3(13). doi:10.1186/s40359-015-0072-5.

Tosi, M. and Van den Broek, T. (2020). Gray divorce and mental health in the United

Kingdom. Social Science and Medicine 256: 113030. doi:10.1016/j.socscimed.

2020.113030.

Uhlenberg, P., Cooney, T., and Boyd, R. (1990). Divorce for women after midlife.

Journal of Gerontology: Social Sciences 45(1): S3–11. doi:10.1093/geronj/45.

1.S3.

Uhlenberg, P. and Hammill, B.G. (1998). Frequency of grandparent contact with

grandchild sets: Six factors that make a difference. The Gerontologist 38(3): 276–

285. doi:10.1093/geront/38.3.276.

Vignoli, D. and Ferro, I. (2009). Rising marital disruption in Italy and its correlates.

Demographic Research 20(4): 11–36. doi:10.4054/DemRes.2009.20.4.

Vignoli, D., Matysiak, A., Styrc, M., and Tocchioni, V. (2018). The positive impact of

women’s employment on divorce: Context, selection, or anticipation?

Demographic Research 38(37): 1059–1110. doi:10.4054/DemRes.2018.38.37.

Vignoli, D., Tanturri, M.L., and Acciai, F. (2016). Home bitter home? Gender, living

arrangements, and the exclusion from homeownership among older Europeans.

Genus 72(1): 9. doi:10.1186/s41118-016-0014-y.

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 649

Weingarten, H.R. (1988). Late life divorce and the life review. Journal of Gerontological

Social Work 12(3–4): 83–97. doi:10.1300/J083v12n03_06.

White, L.K. (1990). Determinants of divorce: A review of research in the eighties.

Journal of Marriage and the Family 52(4): 904–912. doi:10.2307/353309.

Wilson, S.E. and Waddoups, S.L. (2002). Good marriages gone bad: Health mismatches

as a cause of later-life marital dissolution. Population Research and Policy Review

21(6): 505–533. doi:10.1023/A:1022990517611.

Wu, Z. and Penning, M.J. (1997). Marital instability after midlife. Journal of Family

Issues 18(5): 459–478. doi:10.1177/019251397018005001.

Wu, Z. and Penning, M.J. (2018). Marital and cohabiting union dissolution in middle and

later life. Research on Aging 40(4): 340–364. doi:10.1177/0164027517698024.

Yorgason, J.B., Booth, A., and Johnson, D. (2008). Health, disability, and marital quality:

Is the association different for younger versus older cohorts? Research on Aging

30(6): 623–648. doi:10.1177/0164027508322570.

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

650 https://www.demographic-research.org

Appendix

Sample attrition

We investigated the factors associated with the probability of not entering our analytical

sample (i.e., participating in the survey only once). Table A-1 shows the results of a

logistic regression for the probability of leaving the survey after only one interview. We

reported four-digit AMEs in order to avoid rough approximations due to the small

magnitude of coefficients. Respondents who died after their first participation were

excluded from the model (n = 1,802, or approximately 10% of those who left). The model

revealed no relevant gender difference, while people born after 1955 were more likely to

abandon the survey after one wave (it should be noted that we did not consider leaving

the survey due to death as attrition). The probability of attrition was 2.67% lower among

people with tertiary education. Regarding employment status, retired individuals were

the least likely to leave the survey, while the employed and otherwise unretired had a

higher attrition risk. Having children was negatively related to attrition: Individuals with

one child and those with two or more were 2% and 6.6% less likely to leave the sample,

respectively. Having already had previous divorce experiences did not relevantly affect

the probability of attrition. Homeownership was related to a lower probability of attrition,

and those who reported finding it easy to make ends meet were the least likely to leave

due to attrition. These last two findings suggest that individuals with higher

socioeconomic status may be less inclined to terminate their participation. Finally,

limitations in daily activities played no prominent role, while depression was associated

with lower attrition probability (AME = –0.003). All the factors considered in Table 1

were included in the models on the probability of union dissolution due to their

correlation with attrition probabilities.

Demographic Research: Volume 46, Article 21

https://www.demographic-research.org 651

Table A-1: Logistic model for the probability of leaving the survey after one

wave. AMEs are reported. N = 72,032

AME

value

gender (ref. male)

female

–

0.0022

0.631

birth cohort (ref. before 1945)

1946

–

1955

–

0.0071

0.507

after 1955

0.0274

0.098

education (ref. primary)

secondary

–

0.0108

0.202

ertiary

–

0.0267

0.000

employment status (ref. retired)

working

0.0221

0.012

other (

e.g.,

unemployed, homemaker)

0.0141

0.059

number of children (ref. childless)

one child

–

0.0201

0.062

two or more children

–

0.0659

0.000

has already divorced at least once (ref. no)

yes

–

0.0054

0.538

homeowner

ship (ref. no)

yes

–

0.0294

0.000

making ends meet (ref. easily)

fairly easily

0.0139

0.050

with some difficulty

0.0135

0.222

with great difficulty

0.0001

0.934

number of limitations with daily activities

0.0042

0.210

depression scale

–

0.0030

0.043

ountry fixed effects

YES

ave

YES

Source: Authors’ elaboration on SHARE data, waves 1–7 (wave 3 excluded).

Alderotti, Tomassini & Vignoli: ‘Silver splits’ in Europe: The role of grandchildren and other correlates

652 https://www.demographic-research.org

Figure A-1: Flow chart of the sample selection process

72,032 eligible observations

(i.e., they were (1) married or in a registered cohabitation or (2) in an

informal relationship involving cohabitation at the time of the interview)

18,721 observations lost at follow up (i.e.,

they were interviewed only once)

Analytical sample:

53,311 observations interviewed at least twice

Out of the 53,311

observations:

17,392 left the survey

before the last wave

(potential attrition)

3,448 left because of

death

135 left after

experiencing union

13,809 left because of

attrition

1,831 out of 13,809: information

recovered through the partner

135 left after experiencing

union dissolution