A Brief Report on Living Arrangements Following Gray Divorce

Abstract

Objectives

We offer new insights on how older adults in the United States navigate the aftermath of gray divorce (i.e., divorce that occurs among adults aged 50+) by describing their living arrangements upon divorce and tracking the stability of these configurations over time. Living arrangements are important to decipher because they are linked to health, well-being, and longevity.

Methods

Using data from the 1998–2014 Health and Retirement Study, we uncovered patterns of U.S. older adult living arrangements upon divorce (N = 1,057), distinguishing among those who lived alone, lived with others, and lived with a new partner. Multinomial logistic regression models were estimated to assess how individual characteristics (demographics, marital biography, economic resources, health, and social ties) were associated with these configurations. Cumulative survival probabilities gauged the relative stability of these 3 living arrangements.

Results

About half of U.S. adults lived alone upon gray divorce, another one-third lived with others, and the remaining 14% lived with a new partner. Adults living with a new partner tended to exhibit the most advantaged sociodemographic profiles, whereas those living solo or with others were largely comparable. More than 70% of adults experienced a subsequent living arrangement transition if they lived with others upon divorce, versus just 50% of those living alone and only 30% of those with a new partner.

Discussion

After divorce, older adults reside in a range of living arrangements, some of which are more stable than others. Future work should address whether and how these arrangements and their durability are related to postdivorce adjustment.

Recent decades have witnessed a rapid acceleration in gray divorce, which refers to divorce among adults aged 50 and older (Brown & Lin, 2022), but researchers are only beginning to decipher its consequences for individual health and well-being. Consistent with the notion that divorce is a stressful life event, emerging evidence indicates that postdivorce adjustment is more protracted among U.S. middle-aged and older adults compared with younger adults (Lin & Brown, 2020). Younger adults tend to recover from divorce within a year or two (Amato, 2010). In contrast, the adjustment process for those experiencing a gray divorce is distinct, marked by diminished economic (Lin & Brown, 2021; Sharma, 2015) and psychological (Lin et al., 2019; Tosi & van den Broek, 2020) well-being that typically persists for several years. Thus, the rise in gray divorce is contributing to growing vulnerability among the aging population.

The disruption that accompanies divorce often entails residential change (Speare & Goldscheider, 1987; Zilincikova & Schnor, 2021), although its contours remain opaque. Establishing the patterns and stability of older adult living arrangements after gray divorce is important because living arrangements are consequential for health and well-being (Dykstra, 2021; Greenfield & Russell, 2011; Herm et al., 2016) and thus play a role in the postdivorce adjustment process. Following gray divorce, we might expect that most adults live alone but presumably, some live with others, typically adult offspring. Coresidence can be an adaptive strategy to mitigate the economic strain and lack of social support that often accompanies divorce (Choi, 2003; Cooney, 1989; Moen & Wethington, 1992). Repartnering is another pathway that allows the recently divorced to pool resources and exchange support with another. After a gray divorce, roughly 40% of men and nearly 25% of women form new partnerships, whether through cohabitation or remarriage (Brown et al., 2019). Much of this repartnering occurs relatively soon after divorce although some occurs a decade or so later (Brown et al., 2019).

In this brief report, we describe the living arrangements of older adults upon gray divorce, distinguishing among those who live alone, live with others, or live with a new partner at the interview wave when they report their gray divorce. We also examine the associations between respondent characteristics linked to these three living arrangements. Finally, by tracking transitions out of these living arrangements, we address their relative stability. Nearly all prior work on older adult living arrangements is either cross-sectional, ignoring transitions from one arrangement to another, or spans only an abbreviated timeframe (Raymo et al., 2019; Richards et al., 1987). Our investigation of stability and change sheds new light on how adults navigate the aftermath of gray divorce.

Method

Data came from the 1998–2014 Health and Retirement Study (HRS), a nationally representative study of older adults in the United States over the age of 50 interviewed biennially with new cohorts added every 6 years to ensure the sample remains representative. The HRS is well-suited for our study because it provides a large sample of older adults who experience gray divorce and includes a detailed household roster for respondents at every wave. Note that any living arrangement that occurred only between waves was not captured by the HRS. The analytic sample included 1,057 adults who experienced a gray divorce while an HRS respondent and who reported on their living arrangements at the divorce wave and at least one subsequent wave.

Our analyses proceeded in three steps. First, we described living arrangements at divorce distinguishing among respondents who (1) lived alone, (2) lived with others, including relatives and/or nonrelatives but excluding new partners, or (3) lived with a new partner/spouse, at the interview they reported their divorce (which had occurred about 1 year prior to interview, on average). We compared these three groups across a range of individual characteristics at the bivariate level. Respondent characteristics included demographics (gender, age, non-White), marital biography (marital duration and having dissolved a first marriage vs remarriage), economic resources (education, wealth, income, employment, and homeownership), health (number of Activities of Daily Living (ADL) or Instrumental Activities of Daily Living (IADL) difficulties and number of chronic conditions), and social ties (friends living nearby, relatives living nearby, number of adult children, and any minor children). See Supplementary Table 1 for a detailed description of the measures. Second, we used multinomial logistic regression to assess how these characteristics related to the relative risks of being in each of these living arrangement configurations. Third, we performed life table analyses to estimate unadjusted cumulative survival probabilities for all respondents by years since their divorce wave to gauge the stability of these living arrangements over time. In Supplementary Table 2, we used logistic regression to estimate an event history model predicting the likelihood of experiencing a living arrangement transition out of the initial living arrangement reported at the divorce wave (1 = yes, 0 = no) net of respondent characteristics. All analyses were conducted with multiple imputation using chained equations (MICE) across 20 random, multiply-imputed data sets weighted using person-level weights measured in the respondent’s divorce wave to ensure a nationally representative sample (Ofstedal et al., 2011). The Supplement contains detailed information on the multiple imputation procedure (Section A) and the sampling weights used in these analyses (Section B).

Results

As shown in Table 1, just over half (54%) of individuals resided alone upon divorce. Living with others was the arrangement for nearly one-third (32%) of adults and the remaining 14% were living with a new partner by the time they reported their divorce. Most adults living alone (58%) or with a new partner (68%) were men, whereas those living with others were typically women (61% = 100%–39%). Adults living alone tended to be about 2 years older than those in the other two configurations. Roughly 30% of adults living with others identified as non-White compared with only 21% of those living alone or with a new partner. The average duration of the dissolved marriage was much shorter for those who had repartnered (5.6 years) than those in other configurations (about 21 years). Most (83%) people living with a new partner had dissolved a remarriage (as opposed to a first marriage) versus about two-thirds (64%) among those living alone and just over half (52%) for those living with others. Education, wealth, income, and employment levels were highest among those who had repartnered. Those living with others or alone were largely comparable on these dimensions. The three groups did not appreciably differ in levels of homeownership or health. Those coresiding with others reported more adult children (2.6) than did those living alone (2.3) or with a new partner (2.1). Relative to those living alone, adults living with a new partner less often had friends (41% vs 57%) or relatives (14% vs 23%) nearby.

Table 1.

Weighted Means (Standard Deviations) or Percentages of Individual Characteristics by Living Arrangements Upon Divorce

	Living alone	Living with others	Living with a new partner
Mana	57.95	39.20***	68.45*
Age	60.60 (7.79)	58.33 (7.00)***	57.49 (5.61)***
Non-Whiteb	20.85	30.17*	20.76
Marital duration	19.58 (12.83)	20.76 (12.61)	5.62 (8.71)***
Dissolved a remarriagec	64.22	52.36*	82.61**
Years of education	12.99 (3.29)	12.97 (2.94)	13.59 (2.48)
Median wealth	$13,306	$10,649	$31,466
Median income	$24,435	$18,956	$62,014***
Employedd	52.00	55.14	68.52***
Homeownershipd	54.22	49.55	58.50
ADL/IADL difficultiese	0.79 (1.82)	0.84 (1.78)	0.50 (1.34)
Chronic conditions	1.69 (1.42)	1.51 (1.47)	1.62 (1.39)
Friends living nearbyd	57.46	50.93	41.26*
Relatives living nearbyd	23.23	16.67	14.21*
Number of adult children	2.27 (1.68)	2.62 (1.62)**	2.08 (1.42)
Any minor childrend	17.26	13.83	19.82
Weighted percentage of persons	54.10	31.97	13.93
Unweighted number of persons	560	351	146

Notes: Asterisks denote significant differences between living alone and living with others or a new partner. Bolded coefficients indicate significant difference between living with others and living with a new partner at p < .05.

^aReference category is woman.

^bReference category is White.

^cReference category is dissolved a first marriage.

^dVariable is coded 1 = yes, 0 = no.

^eActivities of Daily Living or Instrumental Activities of Daily Living

*p < .05. **p < .01. ***p < .001.

Table 2 depicts the relative risk ratios from the multinomial logistic regressions predicting living arrangements upon divorce, net of individual characteristics. This multivariable model elucidates whether the bivariate patterns persist after accounting for other relevant characteristics. Relative to living alone, men were less likely to live with others but did not differ in their chances of living with a new partner. Age was negatively associated with both living with others and with a new partner. The racial differences obtained in the bivariate analyses were reduced to nonsignificance in the multivariable model (the inclusion of number of adult children was key). Living with a new partner was less common for those dissolving longer-term marriages, net of other factors. Dissolving a remarriage (vs first marriage) was negatively related to living with others or living with a new partner. Net of other factors, economic resources were largely unrelated to living arrangements upon divorce. The only exception was that income was positively associated with living with a new partner (vs living alone). Health also was not appreciably related to postdivorce living arrangements. Those with relatives living nearby were less likely to live with others versus alone. The number of adult children was positively related to living with others or a new partner relative to living independently. Compared with living with others, those living with a partner were especially likely to be men, had more often dissolved marriages that were relatively short-lived, and had higher levels of wealth and income.

Table 2.

Relative Risk Ratios From Multinomial Logistic Regressions Predicting Living Arrangements Upon Divorce

	With others vs alone	With a new partner vs alone	With a new partner vs with others
Demographic characteristics
Mana	0.48***	1.50	3.13***
Age	0.98*	0.95**	0.97
Non-Whiteb	1.31	0.87	0.66
Marital biography
Marital duration	0.99	0.83***	0.84***
Dissolved a remarriagec	0.63*	0.24**	0.38
Economic resources
Years of education	1.00	0.95	0.95
Wealth (scaled, logged)	0.98	1.05	1.08*
Income (scaled, logged)	0.96	4.61***	4.82***
Employedd	1.09	0.67	0.61
Homeownershipd	0.78	1.12	1.41
Health characteristics
ADL/IADL difficultiese	1.00	1.09	1.09
Chronic conditions	0.93	1.14	1.22
Social ties
Friends living nearbyd	0.92	0.65	0.71
Relatives living nearbyd	0.57**	0.95	1.68
Number of adult children	1.16***	1.18*	1.02
Any minor childrend	0.69	1.10	1.59
Constant	5.16*	19.24*	3.73

Notes: ^aReference category is woman.

^bReference category is White.

^cReference category is dissolved a first marriage.

^dVariable is coded 1 = yes, 0 = no.

^eActivities of Daily Living (ADL) or Instrumental Activities of Daily Living (IADL)

*p < .05. **p < .01. ***p < .001.

The three living arrangement configurations exhibited distinct levels of stability, as illustrated in Figure 1. Living with a new partner was the most stable arrangement, culminating in disruption for about 30% of those in this arrangement. Stability levels were moderate for those living alone upon divorce with just over half experiencing a subsequent living arrangement transition. By comparison, living with others postdivorce was a highly unstable arrangement that eventuated in a subsequent living arrangement transition for about 70% of those in this configuration. Perhaps living with others more often functions as temporary arrangement until one can transition to a more permanent situation. As shown in Model 1 of Supplementary Table 2, these differentials were robust to the inclusion of individual controls. Living arrangement instability also was more common among individuals identifying as non-White, those with higher incomes, and those lacking any friends nearby. As indicated by the nonsignificant interaction terms in Model 2, these differentials also were not contingent on time since divorce.

Figure 1.

Cumulative probability of transition out of living arrangement at divorce.

Conclusion

The graying of divorce raises new questions about postdivorce adjustment during the second half of life (Lin & Brown, 2020). Living arrangements have the potential to exacerbate or minimize negative outcomes following a stressful life event such as divorce (Cooney, 1989; Zilincikova & Schnor, 2021), underscoring the urgency of deciphering not only how older adults configure their living arrangements upon divorce but also the stability of these configurations.

We show that only about half of adults live alone upon divorce. Nearly one-third reside with others and about one in six have already formed a new coresidential romantic partnership. Adults living with a new partner tend to be the most advantaged, which is consonant with prior literature showing that repartnering in later life occurs disproportionately among those with more economic resources, better health, and stronger social ties (Brown et al., 2019; Vespa, 2012). Living alone is more common among men, whereas living with others is more popular among women yet the sociodemographic profiles of adults in these two arrangements are substantively similar, particularly in terms of economic resources and health. Although the greater proclivity of adults to live alone rather than with others following divorce is consistent with the widely held belief that older adults prefer solo living or intimacy at a distance to living with relatives or roommates, the largely comparable profiles of the two groups is striking.

In fact, these similarities alone do not translate into comparable residential stability. The risk of transitioning out of one’s postdivorce living arrangement is significantly higher for those living with others than alone, net of individual characteristics. This differential emerges soon after divorce with nearly half of those living with others experiencing a living arrangement transition within 2 years versus only about one-quarter for those living solo. Transitioning out of a living arrangement can be beneficial; for many, it could signal that they have the resources and wherewithal to achieve a more desired living configuration (e.g., from living with others to living alone). The advantages marking individuals who live with a new partner extend to their residential stability, which remains quite high in the years after divorce.

Our study not only illustrates the varied living arrangements of older adults upon divorce, but it also reveals that the durability of these arrangements differs across configurations. Despite these notable strengths, our study has a few limitations. The living with others group combined those living with adult children (70%) and those living with other relatives or nonrelatives (30%) because the latter group was too small to examine separately. For those living alone, some have adult children living nearby, whereas others do not, and this proximity may shape their postdivorce adjustment experience. Also, the HRS only captures living arrangements at each interview wave (i.e., every 2 years) and thus some respondents may have experienced an initial transition following marital dissolution and prior to their next interview. The rather sizeable levels of instability marking the living alone and living with other configurations are thus all the more concerning from the standpoint of how gray divorce affects health and well-being given that residential instability is often disruptive to social networks (Badawy et al., 2018; York Cornwell & Goldman, 2021). Adjustment to gray divorce tends to be prolonged, signaling it is a distinct experience for older versus younger adults (Lin & Brown, 2020). This portrait lays the groundwork for future research on whether and how living arrangements and their stability shape the postdivorce adjustment process for older adults.

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 has core funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (P2CHD050959).

Conflict of Interest

None declared.