Parent–Adult Child Relationships and Repartnering After Gray Divorce

Abstract

Objectives

The rise in gray divorce has catalyzed repartnering in later life. However, the antecedents of older adult repartnering remain poorly understood, particularly the potential role of adult children. A form of ambiguous loss, marital disruption often leads to family boundary ambiguity, thereby weakening family ties. We investigated whether co-residence and proximity of, frequent contact with, and downward transfers to adult children were negatively associated with repartnering after gray divorce.

Methods

Data from the 1998 to 2018 Health and Retirement Study were used to estimate discrete-time event history models using logistic regression to predict repartnering after gray divorce among parents of adult children (N = 1,359).

Results

As expected, adult child co-residence as well as having a child who lived nearby and had frequent contact diminished the likelihood of repartnering. Parents whose children either did not live nearby or were nearby but did not have frequent contact were more likely to repartner than those with children nearby and in frequent contact or co-residing, signaling that both proximity and frequency of contact are related to repartnering. Additionally, the provision of financial support to adult children was associated with a lower risk of repartnering.

Discussion

Relationships with adult children shape parental repartnering behaviors. Echoing the concept of boundary ambiguity, our findings indicate that stronger family ties, as signaled by child co-residence or proximity and frequent contact as well as economic provision, tend to deter repartnering. These findings also indicate that absent strong family ties, older adults are more likely to form a new union.

Repartnering during later life is on the rise, reflecting the ongoing gray divorce revolution (Brown, Lin, Hammersmith, & Wright, 2018). Since 1990, the divorce rate for U.S. adults aged 50 and older has more than doubled (Brown and Lin, 2022). In fact, more than one in three people getting divorced in the United States are experiencing gray divorces. Unlike widowhood, gray divorce results in two single adults eligible to repartner. Nearly 40% of men and about one-quarter of women repartner within a decade following gray divorce (Brown, Lin, Hammersmith, & Wright, 2019). Those who repartner are a bit more likely to form a cohabiting union than a remarriage, illustrating the growing popularity of cohabitation in later life.

Yet, our understanding of repartnering in later life remains limited with only a few empirical studies examining the linkages between various sociodemographic factors and the formation of a new union (Brown et al., 2019; de Jong Gierveld, 2004; Schimmele and Wu, 2016; Vespa, 2012). A glaring omission is a lack of attention to the role of adult offspring in older adults’ decisions to repartner, particularly in light of research showing that gray divorce and subsequent repartnering have significant ramifications for parent–adult child relationship dynamics (Buyukkececi and Leopold, 2024; Lin, Brown, & Mellencamp, 2022, 2024).

Family transitions, including parental marital dissolution and repartnering, challenge existing family boundaries, which can be stressful for offspring who want to maintain family unity (de Jong Gierveld and Merz, 2013). A qualitative study in Europe illustrated the signal role of offspring in shaping older adults’ decisions about repartnering. The presence of a co-resident child deterred union formation, as parents reported capitulating to offspring demands to preserve the existing family unit (de Jong Gierveld and Merz, 2013). Indeed, the ability of adult children to exert influence on their parents’ decisions to repartner is likely a function of the relationship ties between adult children and parents. Stronger family ties, as signaled by co-residence or residential proximity, frequent contact, and economic transfers, are indicative of robust familial bonds and boundaries, which in turn may discourage parental repartnering.

To investigate whether and how parent–adult child relationship ties are linked to parental repartnering after gray divorce, we use data from the 1998 to 2018 Health and Retirement Study (HRS). The HRS is ideally suited for our study because its large sample size yields a generous number of respondents who experience gray divorce and are eligible to repartner. It also taps into multiple dimensions of parent–adult child relationships, including co-residence and proximity, frequency of contact, and economic transfers, allowing us to empirically test novel hypotheses about the importance of family ties for parental repartnering in later life.

Parent–child relationships and gray divorce

Gray divorce is a stressful life event that often disrupts parent–child relationships. From a family systems approach (Cox and Paley, 1997), the consequences of divorce are felt by all members of the family unit. This family-wide impact of parental divorce echoes a key tenet of the life course perspective, namely, the notion of linked lives (Elder, 1994). The lives of parents and their children are intertwined, with the events and experiences of parents shaping the trajectories of their children’s lives (Greenwood, 2012). A burgeoning literature shows that parental divorce has enduring consequences for children’s development through adulthood (Amato, 2010; Chase-Lansdale, Cherlin, & Kiernan, 1995; Thomas and Högnäs, 2015). Emergent evidence suggests that experiencing parental divorce as an adult can be detrimental to individual well-being, too, although the research is scant (Furstenberg and Kiernan, 2001; Hughes and Fredenburg, 2020). Likewise, relatively few studies have examined the ramifications of gray divorce from the perspective of parents, and the results are mixed (Aquilino, 1994; Buyukkececi and Leopold, 2024; Cooney, 1994; Lin et al., 2022; Shapiro, 2003).

Parent–adult child relationship dynamics change significantly in response to gray divorce, often with distinct patterns for mothers versus fathers. Recent studies indicate that mothers’ contact with adult children tends to increase following gray divorce (Buyukkececi and Leopold, 2024; Lin et al., 2022; Shapiro, 2003), although more dated studies found either a decrease in contact (Aquilino, 1994) or no change (Cooney, 1994). Maternal financial transfers to children tend to remain stable after a gray divorce (Lin et al., 2022). Studies have shown consistently that fathers’ contact with adult children declines in the wake of gray divorce (Aquilino, 1994; Buyukkececi and Leopold, 2024; Cooney, 1994; Lin et al., 2022; Shapiro, 2003). However, fathers more often provide economic transfers to their children after a late-life divorce (Lin et al., 2022). Little is known about co-residence and proximity to adult children following gray divorce, although one recent study showed that men more often live alone or with a new partner whereas women disproportionately live with others, typically an adult child (Brown, Lin, & Mellencamp, 2023). The unique patterns characterizing the parent–adult child relationships of mothers versus fathers align with gendered patterns of investment and kin keeping over the life course (Kalmijn, 2007), with mothers typically spending more time with offspring than do fathers (Sayer et al., 2004). As kin keepers, mothers often are the conduit between fathers and children, a linkage that can be strained or broken following divorce (Buyukkececi and Leopold, 2024; Townsend, 2002).

Ambiguous Loss and Boundary Ambiguity After Gray Divorce

The severing of the marital relationship contributes to ambiguous loss, which in this context describes a subjective perception that is rooted in the experience of marital disruption (Afifi and Keith, 2004). Marital disruption can occur through divorce or widowhood. Ambiguous loss is closely tied to boundary ambiguity, reflecting the fuzziness surrounding who is in vs who is out of the family. Spousal bereavement contributes to boundary ambiguity although it tends to be short-term, attenuating as family members work through the grief process (Blackburn, Greenberg, & Boss, 1987). Additionally, unlike divorce, spousal loss is a “clear” transition which helps to alleviate some of the boundary ambiguity post-disruption (Carroll, Olson, & Buckmiller, 2007). In contrast, following divorce, familial relationships must be reconstructed with both parents, exacerbating boundary ambiguity. Typically, parents no longer live together or act as a unit in the same way, which is likely to change how mothers and fathers relate to their children (Buyukkececi and Leopold, 2024; Schenk and Dykstra, 2012). Divorce can lead to boundary ambiguity by disrupting one’s sense of family membership. Family ties often require recalibration after divorce, underscoring the ambiguity divorce imposes on family configurations (Pett, Lang, & Gander, 1992). The familial unit requires reconstitution, which can entail a reckoning of sorts to reestablish the boundaries of family membership (Buyukkececi and Leopold, 2024; de Jong Gierveld and Merz, 2013; Schenk and Dykstra, 2012).

This boundary work may manifest through parents’ decisions to delay or forgo repartnering after gray divorce (de Jong Gierveld and Merz, 2013). Boundary ambiguity can deter repartnering because a new family member would challenge already tenuous familial ties, especially between parents and their children, potentially straining parent–child relationships. In an effort to solidify their relationships with their adult children and project a semblance of stability, parents may privilege their filial relationships over potential coresidential romantic relationships to avoid exacerbating the family boundary ambiguity associated with gray divorce. Older adults with children are more likely than their childless counterparts to form a living apart together (LAT) relationship than a coresidential one in the Netherlands, signaling the deterrent effect of children on late-life repartnering (de Jong Gierveld and Merz, 2013). Drawing on in-depth interviews with older adults in LAT relationships, de Jong Gierveld and Merz (2013) concluded that adult children played an outsized role in parental romantic relationship trajectories. Children were a driving force in older adults’ decisions to form a LAT relationship rather than to cohabit or remarry.

The Current Study

The rapid acceleration in gray divorce in recent decades foretells a corresponding rise in repartnering among older adults (Brown et al., 2018). However, the role of adult children in parents’ decisions about whether to repartner remains unclear. This is a notable gap in the literature because gray divorce alters parent–adult child relationship dynamics (Buyukkececi and Leopold, 2024; Lin et al., 2022). Moreover, the ambiguous loss entailed by marital disruption contributes to boundary ambiguity that can strain or weaken familial ties (Schenk and Dykstra, 2012). We anticipate that in families which recently experienced gray divorce, the tenuousness of familial ties could be especially acute and thus parents and adult children alike may have reservations about parents forming new coresidential unions because they could introduce still more boundary ambiguity.

Our aim is to chart patterns of repartnering after gray divorce for mothers and fathers and then test whether three indicators of parent–child relationship ties (i.e., co-residence and proximity, frequent contact, and downward transfers) are associated with lower risks of parental repartnering, paying attention to possible variation by parental gender. Based on the limited prior literature showing that child co-residence deters the formation of a coresidential romantic union (de Jong Gierveld and Merz, 2013), we anticipate that parents who report stronger ties to their adult children may be reluctant to jeopardize those ties and thus the strength of these ties are negatively associated with repartnering. Specifically, child co-residence or proximity (measured in the HRS as living within 10 miles of at least one child) may deter repartnering among mothers and fathers alike. We also assess the associations between frequent contact with adult children, which we define as weekly contact or more, and downward financial transfers to adult children which together provide a more nuanced, multidimensional approach to conceptualizing parent-adult child relationships. Based on the gendered patterns of change in these latter two indicators of parent-adult child relationships following divorce with contact rising among mothers and transfers increasing among fathers (Lin et al., 2022), we anticipate that contact and transfers may be uniquely salient for the repartnering behaviors of mothers and fathers, respectively. We test whether all three indictors are more closely tied to repartnering for mothers than fathers (cf. Kalmijn, 2007).

Our study accounts for several sets of factors associated with repartnering after gray divorce. Economic resources, such as education, employment, homeownership, and wealth, tend to be positively related to repartnering after gray divorce (Brown et al., 2019) and later life repartnering more broadly (Vespa, 2012). Good health, as signaled by having fewer chronic conditions or difficulties with (instrumental) activities of daily living (ADL/IADL), is also linked to repartnering following gray divorce (Brown et al., 2018, 2019). Social ties to friends and family may deter repartnering after divorce (Brown et al., 2019; Vespa, 2013). Among older adults, the marital biography is consequential for repartnering. Repartnering is most likely to occur shortly after gray divorce and attenuates over time. Older adults who divorce from long-term, first marriages are less likely to repartner than their counterparts who end shorter-term remarriages (Brown et al., 2019). Additionally, we account for the characteristics of the offspring of older adults, such as the number of adult children, the presence of a minor child, the age of the oldest child, and the presence of any daughters. Finally, we include demographic characteristics of older adults, including racial-ethnic group, age, and gender as repartnering after gray divorce is more common among Whites, younger people, and men (Brown et al., 2019).

Method

Data used in the analysis came from the 1998 to 2018 Health and Retirement Study (HRS), a longitudinal study of a nationally representative sample of adults aged 50 and older. The HRS is sponsored by the National Institute on Aging (grant number NIA U01AG009740) and is conducted by the University of Michigan. The HRS began interviewing a nationally representative sample of 33,855 older adults in 1998, who are re-interviewed every other year. Recent cohorts of adults aged 51–56 were added in 2004, 2010, and 2016. The HRS is the only longitudinal and nationally representative study that includes a sizable number of older adults who experience a gray divorce. Moreover, the study includes a variety of information regarding marital divorce, subsequent repartnering, and parent–adult child relationships and allows us to examine the role of parent–adult child relationship characteristics in the risk of subsequent repartnering over time. Information on respondents’ children came from the RAND Family Data File 1992–2018 (V2).

Analytic Sample

To identify gray divorces, we created a marital history file that tracks the marriages formed or dissolved by HRS respondents. A total of 1,533 respondents experienced a gray divorce at age 50 or older from 1998 to 2018. We captured divorces reported at the time of the respondent’s first interview or after they entered the 1998 HRS to avoid left-censoring bias. For respondents who reported more than one divorce after age 50, we used information from their first gray divorce. We removed 10 respondent who were in a same-sex union and 19 respondents who had a sample weight equal to 0 or missing. Finally, we removed 145 respondents who did not report any children aged 18 or older. The final analytic sample consisted of 1,359 gray divorced individuals (671 mothers and 688 fathers), of whom 482 (190 mothers and 292 fathers) repartnered by 2018.

Measures

The dependent variable, repartnered, was a time-varying covariate that assessed whether the divorced respondent formed either a remarriage or cohabiting union (0 = no, 1 = yes). We grouped together new cohabitations (n = 235) and remarriages (n = 247) into an umbrella category of repartnering because cohabitation is as common as remarriage after gray divorce (Brown et al., 2019) and cohabitation appears to function as a long-term alternative to remarriage in later life (Wright, 2020).

In the HRS, respondents were asked at every interview how close each adult child lived to them. RAND collapsed this geographic data into a variable indicating whether each child resided with the respondent, and if not, whether the child lived within a 10-mile radius at a given wave. We classified proximity as follows: no child nearby (i.e., those with no co-resident adult children and no adult children living within 10 miles); child nearby (i.e., those with no co-resident adult children but at least one adult child living within 10 miles); and co-resident child (i.e., respondents with at least one co-resident adult child). Frequent contact was gauged at every interview to ascertain whether the respondent reported at least weekly contact with an adult child in the past 12 months (0 = no, 1 = yes). Contact could be in person or by phone, mail, or email. Respondents were not asked about their frequency of contact with resident adult children, so at least weekly contact was assumed. However, this meant that the measures of proximity and contact were collinear because respondents were uniformly assigned as having frequent contact with any child who was co-resident. Roughly one-quarter of the sample had a co-resident adult child and thus treating these two constructs as distinct would introduce bias in our model. Consequently, we combined these two measures into one time-varying variable, proximity and contact, to distinguish among the following groups: no child nearby and no frequent contact; no child nearby but frequent contact; child nearby but no frequent contact; child nearby and frequent contact; and co-resident child (and frequent contact by definition; reference category).

Downward financial transfer indicated whether the respondent gave an adult child financial help totaling $500 or more in the past two years (0 = no, 1 = yes).

Family characteristics were assessed using four time-varying covariates. Number of adult children gauged the respondent’s number of children aged 18 or older. Any minor child indicated whether the respondent reported any child younger than age 18 (0 = no, 1 = yes). Age of oldest child captured the age of the respondent’s oldest child. Any daughters indicated whether the respondent reported having a daughter (0 = no, 1 = yes).

Economic resources were captured across four covariates. Education was measured as years of schooling completed by the respondent’s baseline wave (i.e., when the respondent first reported their gray divorce). Employment was a time-varying covariate that measured whether the respondent engaged in full or part-time employment (0 = no, 1 = yes). Homeownership was time-varying and assessed whether the respondent owned a home (0 = no, 1 = yes). Wealth was a time-varying covariate that captured the logged value of the respondent’s non-housing household wealth. All wealth values were adjusted to real 1998 dollars. Those with wealth beyond the 99th percentile were top-coded to the 99th percentile. Respondents reporting negative wealth were assigned a value of $1 before taking the natural log.

The respondent’s health was assessed using two time-varying covariates. Number of chronic conditions was a time-varying measure that summed whether a doctor had diagnosed the respondent with a psychiatric disorder, heart disease, hypertension, cancer, lung disease, arthritis, diabetes, or a stroke. Values could range from 0 to 8. Number of ADL/IADL difficulties was a time-varying covariate that summed the number of reported difficulties with activities of daily living (ADLs) and instrumental activities of daily living (IADLs). ADLs gauged difficulty walking, dressing, getting in and out of bed, bathing, eating, and toileting. IADLs tapped into difficulty preparing hot meals, shopping for groceries, making phone calls, taking medications, managing money, and using a map. ADL/IADL difficulties could range from 0 to 12.

Social ties were gauged using two time-varying covariates. Friends nearby indicated whether the respondent reported friends living in close proximity (0 = no, 1 = yes). Relatives nearby indicated whether the respondent reported relatives living in close proximity (0 = no, 1 = yes).

The respondent’s marital biography was captured using three variables. Years since gray divorce was a time-varying covariate that accounted for the number of years since the respondent’s gray divorce occurred. Duration of prior marriage was a time-invariant covariate that measured the number of years spent in the prior marriage. Dissolved a remarriage was a time-invariant covariate indicating whether the respondent was in a higher-order marriage at the time of gray divorce (0 = no, 1 = yes).

Finally, two demographic characteristics were included. Age was time-varying. Race-ethnicity included Non-Hispanic White (reference category), Non-Hispanic Black, Hispanic, and Other Race as reported by the respondent.

Analytic Strategy

We first calculated survival probabilities by years since divorce to estimate the timing of repartnering following gray divorce. Respondents were censored after a 10-year period of being at risk of repartnering, as such partnerships tend to form shortly after divorce (Brown et al., 2019). Although we hypothesized that parent–child relationship ties operate uniquely for mothers and fathers, we did not uncover any significant gender differences in the fully adjusted models and thus we present findings using the pooled sample of mothers and fathers to maximize statistical power. We calculated baseline bivariate descriptive statistics separately according to repartnership status to compare our indicators of family ties (proximity and contact as well as downward transfers) and various sociodemographic characteristics between those who repartnered versus remained single.

Finally, we performed logistic regression analyses using discrete-time event history models to predict the role of family ties, including parent–adult child proximity and frequent contact as well as downward transfers, in the risk of subsequent repartnering after gray divorce. Because the dates of divorce and repartnering were measured at time intervals, discrete time-event modeling was better suited than continuous time modeling. Additionally, discrete-time event history models allowed us to account for the timing of transitioning from the divorced state to forming a new partnership, net of time-varying covariates, which were measured at each interview wave. Log-linear methods were used for model estimation:

$${\displaystyle \begin{array}{l}{\displaystyle P_{it}=1/[1+exp(-{\alpha }_t\beta x_{it})]}\end{array}}$$

P _it was the hazard rate, defined by P_it = Pr(T = t, |T ≥ t), where T was the discrete random variable given the uncensored time of event occurrence (Allison, 1982). In other words, P_it was the conditional probability that repartnering occurred to person i at time t, given that it had not already occurred. We examined how this hazard rate was a function of time (α_t) and a vector of explanatory variables (x_it, with its coefficient vector β). Respondents contributed a series of person-waves that spanned the period during which they were “at risk” of or eligible to repartner. The initial person-wave, which we term the baseline, occurred when the respondent first reported they had divorced. This could occur during any HRS wave spanning from 1998 to 2018. Respondents contributed waves of data to the analysis until they were censored, which occurred at the wave they reported that they had repartnered or, if they had not repartnered, then censoring occurred as soon as one of the following transpired: 10 years had elapsed since gray divorce, at the 2018 interview, or at attrition. This strategy mirrored that used in other research on repartnering in later life (Brown et al., 2019; Vespa, 2012).

The full model shows how the time-varying parent–adult child relationship indicators as well as both the time invariant and time-varying covariates were related to repartnering after gray divorce. Regardless of whether a variable is time invariant or time-varying, its interpretation is similar. The key distinction between the two types is simply that time-varying variables can differ in their values over time whereas time invariant variables remain constant. In our model, time-varying variables were measured contemporaneously, an approach consistent with prior research (Brown et al., 2019; Schimmele and Wu, 2016; Vespa, 2012). Supplemental analyses were conducted using lagged measures of the time-varying variables to enhance time-ordering (see Supplementary Table 1), although it compromised time-ordering for the roughly 14% of respondents who reported having divorced and then repartnered at the same interview wave (lagging their covariates meant they were obtained while the respondent was still married to the spouse from whom they gray divorced). The pattern of results echoed those obtained in the main analysis, underscoring the robustness of our approach.

Missing data were minimal. A multiple imputation procedure, multivariate imputation using chained equations (MICE), was preformed using Stata’s mi impute chained command to handle missing cases for our analytic sample, in which the single value for a missing variable was estimated as a function of the other covariates (Raghunathan, Lepkowski, van Hoewyk, & Solenberger, 2001; van Buuren, Boshuizen, & Knook, 1999). The results presented are based on ten random, multiple-imputed replicates. Additionally, both the bivariate and multivariate estimates were weighted using the baseline (i.e., the time at gray divorce) weights (Ofstedal, Weir, Chen, & Wagner, 2011).

Results

Figure 1 shows the cumulative probabilities of repartnering for mothers and fathers by years since gray divorce. Repartnering was less common among mothers than fathers. Nonetheless, the repartnering patterns of mothers and fathers followed comparable trajectories over time. Repartnering was most likely in the first four years after divorce and slowed thereafter.

Figure 1.

Cumulative probability of repartnering after gray divorce, by gender. Source: Health and Retirement Study, 1998–2018.

Table 1 reports the baseline (i.e., at the time of gray divorce) characteristics of parents according to whether they eventually repartnered or remained unpartnered after gray divorce. As expected, parents who remained unpartnered disproportionately had a co-resident child at 32.63% compared with just 19.61% among parents who repartnered. In contrast, lower proximity to and infrequent contact with adult children was more common among parents who repartnered than parents who remained unpartnered. For example, 33.59% of parents who repartnered had no child nearby but frequent contact versus only 24.41% among those who remained unpartnered. Contrary to our expectations, downward transfers did not differ significantly between parents who repartnered versus those who did not. Just over 40% had provided an economic transfer to an adult child.

Table 1.

Weighted Means (SD) or Percentages of Baseline Characteristics by Repartnering Status

Characteristics	Repartnered	Unpartnered	Differences by repartnering status
Family Ties
Adult children’s proximity and contact
Co-resident child	19.61	32.63	**
Child nearby and frequent contact	23.80	29.17
Child nearby but no frequent contact	6.67	3.00	*
No child nearby but frequent contact	33.59	24.41	*
No child nearby and no frequent contact	16.32	10.79
Downward financial transfer	40.26	42.99
Family characteristics
Number of adult children	2.65 (1.46)	2.59 (1.57)
Any minor child	13.10	15.45
Age of oldest child	33.76 (8.58)	34.79 (9.72)
Any daughters	74.61	73.81
Economic resources
Years of education	13.33 (2.90)	12.87 (3.08)
Employed	50.08	62.17	**
Homeowner	53.76	51.25
Median wealth (in 1998 dollars)	$23,410	$8,155	*
Health characteristics
Number of chronic conditions	1.58 (1.43)	1.85 (1.59)	*
Number of ADL/IADL difficulties	0.46 (1.27)	1.02 (2.10)	***
Social ties
Friends living nearby	48.60	52.38
Family living nearby	18.78	23.49
Marital biography
Years since gray divorce	1.08 (0.74)	0.99 (0.75)
Duration of prior marriage	17.76 (13.35)	20.29 (13.00)	**
Dissolved a remarriage	68.06	57.95	**
Demographic characteristics
Age	58.59 (6.37)	59.98 (8.01)	**
Race-ethnicity
White	73.21	74.87
Black	10.40	12.62
Hispanic	11.93	9.38
Other	4.46	3.14
Gender
Man	61.85	47.00	**
Woman	38.15	53.00	**
Weighted percentage	36.65	63.35
Unweighted number	482	877

Notes: ADL = activities of daily living; IADL = instrumental activities of daily living; SD = standard deviation. Baseline refers to the interview wave at which the respondent reported their gray divorce. Data source: Health and Retirement Study, 1998-2018.

*p <.05.

**p < .01.

***p < .001.

Family characteristics were comparable for parents regardless of whether they repartnered. Turning to economic resources, parents who repartnered (50.08% and $23,410) were less often employed and enjoyed higher median wealth, on average, than parents who did not form a new union (62.17% and $8,155). As expected, health was linked to repartnering. Parents who repartnered (1.58 and 0.46, respectively) reported fewer chronic conditions and ADL/IADL difficulties than parents who did not repartner (1.85 and 1.02, respectively). In contrast, social ties were similar for parents regardless of repartnership status. In terms of marital biography, parents who repartnered (17.76 years) had exited relatively shorter marriages than parents who did not repartner (20.29 years). They (68.06%) also more often ended a remarriage (versus a first marriage) than did parents who did not form a new union (57.95%). Parents who repartnered were more than one year younger, on average, than parents who did not repartner. The racial-ethnic composition of parents who repartnered was comparable to that of parents who remained unpartnered. However, repartnering was more common among men (61.85%) than women, as expected.

Table 2 reports odds ratios from the discrete-time logistic regression predicting repartnering. An odds ratio indicates the relative odds of repartnering for a group compared with their counterpart or the change in the odds of repartnering for every one-unit increase in a continuous predictor with values greater than 1.00 signaling higher odds whereas values less than 1.00 corresponding with lower odds, net of covariates (at p < .05). As expected, family ties were negatively related to repartnering. Specifically, lacking either frequent contact or a nearby child (or both) was positively associated with repartnering. Relative to having a co-resident child, having no child nearby (with [OR = 2.29, p < .001] or without [OR = 3.49, p < .001] frequent contact) or a nearby child without frequent contact (OR = 3.71, p < .001) were all related to higher odds of repartnering. Notably, only those with a child nearby and frequent contact (OR = 1.54, p > .05) did not differ from those with a co-resident child in terms of their likelihood of repartnering. Comparing the various groups, we found that repartnering was more likely for parents with nearby children when they lacked frequent contact than when they had frequent contact (superscript b). Similarly, for parents with no child nearby, those without frequent contact had a higher likelihood of repartnering than those with frequent contact (superscript d). However, repartnering did not appreciably differ between parents with nearby children but no frequent contact and parents who did not have children nearby but did have frequent contact (indicated by the absence of a c superscript for no child nearby and frequent contact). Downward transfers were negatively related to repartnering among parents, which also was in line with our expectations. Repartnering tended to be selective of more advantaged parents as those who were wealthier, had fewer ADL/IADLs, and were younger were more likely to repartner. Having friends nearby was linked to a reduced likelihood of repartnering. Hispanic parents exhibited a higher likelihood of repartnering than did their White counterparts. And fathers were more likely to repartner than mothers.

Table 2.

Odds Ratios from Discrete-Time Logistic Regressions Predicting Repartnering

Variable	OR
Family ties
Adult children’s proximity and contacta
Child nearby and frequent contact	1.54	c , e
Child nearby but no frequent contact	3.71	***,b
No child nearby but frequent contact	2.29	***e
No child nearby and no frequent contact	3.49	***,b,d
Downward financial transfer	0.55	***
Family characteristics
Number of adult children	1.11
Any minor child	0.74
Age of oldest child	1.01
Any daughters	0.98
Economic resources
Years of education	1.05
Employed	1.01
Homeowner	1.29
Wealth (logged)	1.05	*
Health characteristics
Number of chronic conditions	1.07
Number of ADL/IADL difficulties	0.84	*
Social ties
Friends living nearby	0.68	*
Family living nearby	0.82
Marital biography
Years since gray divorce	0.95
Duration of prior marriage	1.00
Dissolved a remarriage	1.49
Demographic characteristics
Age	0.95	**
Race-ethnicity (reference = White)
Black	0.97
Hispanic	1.67	*
Other	1.93
Gender (reference = Man)	1.72	**
Constant	0.33

Notes: ADL = activities of daily living; IADL = instrumental activities of daily living; OR = odds ratio. Data source: Health and Retirement Study, 1998–2018.

^aReference category is coresident child. Odds ratios significantly differ at p < .05;

^bchild nearby and frequent contact;

^cchild nearby and no frequent contact;

^dno child nearby and frequent contact; and

^eno child nearby and no frequent contact.

*p <.05. **p < .01. ***p < .001.

Discussion

The rapid growth in gray divorce in recent decades is spurring family diversification in later life (Carr and Utz, 2020). In particular, it is contributing to rising levels of repartnering among older adults, who are much more likely than their widowed counterparts to form a new union (Brown et al., 2018). Despite these shifting family patterns, later life repartnering remains understudied (Raley and Sweeney, 2020; Sassler, 2010). Our work helps to fill this gap by investigating the role of filial ties for the repartnering behaviors of parents after gray divorce.

Parental marital disruption tends to worsen parent-child relationship dynamics, reflecting family boundary ambiguity as parents and children alike work to reestablish family membership and roles in the aftermath of divorce. Perhaps understanding that repartnering can introduce still more boundary ambiguity by challenging the composition of the familial unit yet again, many parents, particularly those with strong ties to their adult children, may be reluctant to repartner.

Building on a qualitative study conducted in The Netherlands that revealed the centrality of offspring to parental decision-making about repartnering (de Jong Gierveld and Merz, 2013), we assessed whether mothers and fathers who reported stronger family ties were less likely to repartner after gray divorce. A key contribution of our study was its multidimensional approach to conceptualizing family ties, which encompassed co-residence and proximity of, frequent contact with, and downward economic transfers to adult children. These dimensions (and other covariates) were gauged in a time-varying, prospective approach that allowed us to track how family ties following gray divorce were related to repartnering.

Consistent with our hypotheses, the results revealed that parents who reported stronger ties with their adult children were less likely to form a new union. In particular, co-residence with an adult child or having an adult child living nearby and having frequent contact with them were deterrents to repartnering. In contrast, repartnering was more likely among parents whose children either did not live nearby or if nearby they did not have frequent contact with them, signaling the importance of both proximity and contact. Additionally, the provision of downward transfers was negatively associated with repartnering. Notably, our findings persisted for mothers and fathers alike. Contrary to our expectations that these indicators of parent-adult child relationship ties operate uniquely by gender, we found no evidence of differential associations for mothers versus fathers, leading us to present a combined model for the entire sample.

More broadly, this study suggests that repartnering following gray divorce may serve a compensatory function, providing an alternative way to “do family” when one’s family ties are more fragile. The parent-adult child relationship may be frayed or even fractured, prompting parents to seek familial connection through co-resident partnerships. When family ties are strong, parents seem to be less inclined to form a new union, perhaps because they already have the support and affinity of their offspring. Alternatively, they may simply be hesitant to risk disrupting what could be precarious post-divorce relationships with their offspring. A shortcoming of the current study is that we were unable to explore whether parents pursued non-co-residential romantic relationships, including dating or living apart together (LAT) partnerships, which may be more palatable to adult children who are wary of their parent forming a new household and family with a co-resident partner (de Jong Gierveld and Merz, 2013). Unfortunately, the HRS does not collect information on non-co-residential romantic relationships, which constrains our view of relationship formation. Other work has shown that roughly half of unmarried partnerships among midlife adults are dating or LAT relationships, not cohabiting unions (Brown, Manning, & Wu, 2022).

A logical next step in this line of inquiry is to assess whether and how repartnering reshapes ties between parents and their children and in particular whether the form this takes depends on the contours of the parent-adult child relationship prior to repartnering. Prior work has shown that repartnering can spur conflict between parents and their children, diminishing family connectedness and intergenerational exchanges (Schenk and Dykstra, 2012), more so for fathers than mothers (Kalmijn, 2015). Based on our findings, we might expect that for parents who are distant from their children, repartnering may result in little appreciable change in the filial relationship. Alternatively, weak ties could devolve into estrangement following repartnering (Ganong and Coleman, 2017). A stressful family transition, repartnering has the potential to introduce new boundary ambiguity, straining existing (and likely already tenuous) ties between parents and their children (de Jong Gierveld and Merz, 2013; Ganong and Coleman, 2017). Repartnering also entails the formation of a stepfamily, introducing additional relationship challenges with the parent’s new partner and the partner’s children. Family boundaries are notoriously blurry in stepfamilies (Stewart, 2005) which can undermine parent–child relationships as they attempt to navigate the murkiness of family membership. Deciphering how repartnering can alter parent-child relationships is an important question for future research, particularly as we have shown that these ties play a pivotal role in decisions to repartner in the first place.

Even as our study fills a notable gap in the literature on later life repartnering by examining the role of parent–adult child relationship ties, it also has some limitations. First, we only have the perspective of the parent (cf. Cohen, Demeter, & Koren, 2024); how adult children appraise their relationship with their parent is not measurable in the HRS because offspring are not interviewed. In line with the concept of the intergenerational stake, parents may be more sanguine than offspring in their appraisals of family ties (Buyukkececi and Leopold, 2024). Second, although we were able to tap into parent–adult child relationships as a multidimensional construct, which is an improvement over much of the prior research on intergenerational family relationships that has been constrained to a single dimension (Schenk and Dykstra, 2012), nonetheless we would have liked to have additional indicators to capture closeness and affinity with adult children, for example. Such measures are only available every four years in the HRS psychosocial mail survey for the same half of respondents. A smaller sample with a longer time interval prevents us from including these measures in the analysis. Third, it is possible that some of our findings are due to in part to selection rather than causation. Despite our prospective approach that avoided left censoring bias, parental decisions about their geographic proximity to their children (or frequency of contact or transfers) may be endogenous to their preferences for repartnering. For example, a parent who is not interested in repartnering may be especially motivated to either co-reside with or live near an adult child. That is, parents who do not want to repartner may invest more in their filial relationships. To assess the robustness of our results, we estimated our multivariable model using lagged time-varying covariates. The lagged model yielded findings comparable to those shown in our main model, bolstering our confidence that our conclusions are not merely a function of endogeneity. Fourth, a related point concerning reverse causation is that, in anticipation of eventual co-residential repartnering, the parent–child relationship already could be shifting in response to the romantic relationship the parent has initiated (de Jong Gierveld and Merz, 2013). Our modeling strategy cannot account for this real possibility. Even lagged modeling does not fully address this issue (Leszczensky and Wolbring, 2022) because parental romantic relationship timing may not align with the survey intervals (for the HRS, lagging by one wave of data corresponds with roughly a two-year time period). Lin et al. (2022) showed that parent–child relationship dynamics change alongside both the transition to gray divorce and subsequent repartnering, although their approach relied on concurrent measures of union transitions and parent-child relationship features making causal order uncertain. Finally, as noted earlier, we were not able to examine transitions into non-co-residential romantic relationships which are becoming more common among older adults and offer more autonomy than either cohabitation or remarriage (Benson and Coleman, 2016).

Gray divorce is reshaping the aging experience for a growing number of older adults, many of whom go on to form new co-residential unions. Reflecting the rising family diversity in later life, repartnering is accelerating. Our study contributes to the nascent literature on repartnering after gray divorce by revealing the centrality of family ties in the aftermath of marital disruption. Strong parent-adult child relationships tend to discourage repartnering, which aligns with work showing that the filial tie increasingly assumes center stage as parents age (Carstensen, 1995; de Jong Gierveld and Merz, 2013). The effects of marital disruption reverberate within the familial unit, leading to uncertainty and boundary ambiguity about family ties and membership. Parents who can weather these challenges and forge strong bonds with their offspring are less likely to repartner than their counterparts for whom filial ties are tenuous or broken. Lacking strong family ties, many aging parents “do family” by forming a new co-residential union. An important next step in this line of inquiry is to assess how repartnering is related to health and well-being and whether these linkages depend on relationship ties with offspring.

Funding

This research was supported by a grant from the National Institute on Aging (R15AG047588). Additional support was provided by the Center for Family and Demographic Research, Bowling Green State University, which received core funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (P2CHD050959).

Conflict of Interest

None.

Data Availability

This paper uses public release data files from the Health and Retirement Study. For more information, please visit https://hrs.isr.umich.edu/about. This study was not preregistered.