Abstract
Previous research has suggested that unmarried persons may be disadvantaged in personal networks and social support. However, little is known about whether the quantity and quality of social relationships differ by marital status among older Americans. Using data from the 2006 and 2008 psychosocial questionnaires of the Health and Retirement Study, we examined the quantity and frequency of contact of three types of social relationships (i.e., friends, children, and family members) as well as the quality of these ties across six marital status groups (i.e., first married, remarried, cohabiting, divorced/separated, widowed, and never married). Our analytic sample included 13,087 respondents aged 51 and above. Multiple linear regression was used for the analysis. Our results show significant differences in social relationships by marital status. In terms of relationships with friends, compared to first-married persons, all the unmarried groups (except the cohabitors) had more frequent contact with their friends and reported greater support as well as greater strain from their friends, controlling for demographic covariates. Remarried persons and cohabitors were largely similar to their first-married counterparts except that both had fewer friends they felt close to, and the cohabitors also reported greater strain with their friends. In terms of relationships with children, all the unmarried groups except for widowed persons had fewer children they felt close to. All the unmarried groups were also disadvantaged in contact frequency and perceived social support from their children. Remarried persons were significantly disadvantaged compared to first-married persons in contact frequency and relationship quality with their children. These associations were largely robust when health and socioeconomic conditions were controlled for. In terms of relationships with family members, there were only a few significant differences across marital status groups. Our findings show the crucial role of marital status in shaping social relationships in later life.
Keywords
Profound changes in American families have occurred over the past six decades: Divorce rates increased sharply starting in the 1960s and then plateaued in recent years (Amato, 2010; Carr & Utz, 2020). However, among those aged 50 and over, divorce rates doubled between 1990 and 2000 (Brown & Lin, 2012). Remarriage has declined but cohabitation has increased for all age groups. As a result, a significant and increasing proportion of Americans are entering late adulthood never married, divorced/separated, or cohabiting (Zhang, 2006). Previous research has suggested that unmarried older adults may be disadvantaged in personal networks and social support due to the lack of a spouse (Ermer, 2021; Wrzus et al., 2013); however, others argue that older adults tend to adapt and rebalance their networks following life transitions such as marital loss (Cornwell et al., 2021). Singlehood may promote social connections with friends, kin, and neighbors to compensate for the lack of marital support (Gerstel & Sarkisian, 2006; Sarkisian & Gerstel, 2016). Despite the importance of this debate, there are surprisingly few empirical studies of the quantity and quality of older adults’ social relationships differentiated by marital status using nationally representative data in the United States. Furthermore, we know little about how cohabitation and remarriage may affect multiple dimensions of social relationships in later life.
This paper extends the previous literature in three ways. First, our study utilizes a national sample to examine the quantity and quality of three major types of social relationships (i.e., friends, children, and family members) in late adulthood across six marital status groups (i.e., first married, remarried, cohabiting, divorced/separated, widowed, and never married) in the U.S. Second, we examine whether the link between different dimensions of social relationships and marital status can be explained by socioeconomic resources and health. Third, we assess whether the link between social relationships and marital status varies by gender.
Social Relationships and Marital Status: Married Versus Unmarried
Social relationships have multiple dimensions, including the structure and quality of social ties in one’s social network (Song et al., 2021). The structural dimension refers to the size and composition of social relationships, such as the number of friends/kin or the frequency of contact; the qualitative dimension encompasses subjective appraisal of social relationships, including perceived support or strain (Carr & Moorman, 2011; Hsieh & Liu, 2021). Two competing hypotheses have dominated the debate on the link between social relationships and marital status. The “marriage as an integrative institution” perspective suggests that marriage can foster and enhance multiple social relationships by providing resources and opportunities for various social interactions. For example, marriage may increase the size of one’s social networks through access to the spouse’s social ties including kin and friends (Fischer et al., 1989; Sarkisian & Gerstel, 2016). Marital dissolution such as divorce, on the other hand, may disrupt shared social relationships as well as neighborhood ties when divorcees change residence. New ties may not be easily formed given the challenges of dealing with the aftermath of a divorce, which may result in a smaller social network (Leopold, 2018). Most divorced men and women experience losses in their networks after divorce (Terhell et al., 2004; Wrzus et al., 2013). Furthermore, in terms of the quality of social relationships, married older adults reported higher satisfaction with their social networks than divorced, widowed, and never-married persons (McLaughlin et al., 2010). As for relationships with adult children, parental separation and divorce are associated with less contact and lower parent-child relationship quality, especially for divorced fathers. On the other hand, widowhood does not seem to affect parent-child contact overall (Shapiro, 2003; Ward et al., 2009, 2014). Moreover, widowhood may trigger an increase in support and contact from children, friends, and other relations shortly before or after the death of the spouse (Guiaux et al., 2007; Ha, 2008) as people rally to support the bereaved individual in a crisis. The uptick in support may subside as time passes and the widowed person adjusts to singlehood (Ha, 2008). Adaption and rebalance of social networks following life transitions keep social connectedness relatively stable in the long run (Cornwell et al., 2021; Guiaux et al., 2007). Theoretically speaking, we tend to regard marital status as a stable state that “shapes our daily and long-term opportunities, constraints and demands” (Umberson et al., 2009, p. 328), but it is important to recognize marital transitions as crisis periods that may bring forth sudden changes in social relations (Ha, 2008).
By contrast, the “marriage as a greedy institution” perspective argues that marriage may compete with and even undermine couples’ ties to their extended families, including parents and adult siblings, as well as their friends, because the demands of a modern marriage can leave little time for married people to invest in other social ties (Sarkisian & Gerstel, 2012). As Gerstel and Sarkisian (2006) put it, marriage can “demand a kind of intense emotional involvement that by itself detracts from collective life” (p. 19). A recent study found that married American adults are less likely to socialize with and provide help to their friends, parents, siblings, and neighbors than their single counterparts (Sarkisian & Gerstel, 2016). This finding was echoed by more recent work using data from the National Social Life, Health, and Aging Project (NSHAP), which found that married persons felt less close to their social network than widowed and divorced persons (Ermer, 2021). Furthermore, there are differences in social relationships among unmarried people. For instance, widowed persons felt closer to their social network than their divorced counterparts (Ermer, 2021).
Recently, scholars have also argued that some of the differences in social relationships by marital status may be partially explained by the unequal socioeconomic resources of married and unmarried persons. Lower socioeconomic status (SES) is a significant risk factor for divorce (Amato, 2010) and widowhood (Sullivan & Fenelon, 2014). Moreover, these marital transitions often lead to worsening financial security (Leopold, 2018; Lin et al., 2017). As a result, married persons usually have higher education and income than their unmarried counterparts (Sarkisian & Gerstel, 2016; Waite & Gallagher, 2000), both of which are often associated with larger network size and more ties to nonkin (McPherson et al., 2006; Moore, 1990). Although higher SES is associated with lower fertility (Dribe & Scalone, 2020), and thus potentially fewer children to rely on in old age, the empirical evidence is mixed about how SES is linked with relationship quality with children and other kin (Sarkisian & Gerstel, 2016). Given that married people are often healthier than unmarried people (Zhang et al., 2016), and better health is associated with increased social connectedness over time (Schwartz & Litwin, 2019), poorer health may be an additional barrier for unmarried persons to have more frequent contact with their friends and kin.
Remarriage, Cohabitation, and Social Relationships
Much less is known about how remarriage and cohabitation may influence older adults’ social relationships (as an exception, see Kalmijn & Broese van Groenou, 2005). Remarriage and the formation of stepfamilies are common in contemporary U.S. society. Based on estimates from the Health and Retirement Study (HRS), 40% of middle-aged and older couples with children are in stepfamilies (Lin et al., 2018). Because remarriage has often been described as an incomplete institution due to the lack of norms governing various social interactions (e.g., parent-stepchild relationships) (Cherlin, 1978), its influences on social relationships may differ depending on the relationship examined (e.g., friends, biological children, stepchildren, or kin). On the one hand, compared to divorced or widowed individuals, remarried individuals not only can gain a source of support from their new partner (Bildtgård & Öberg, 2017), but also may benefit from additional kin support through the new union, including emotional, financial, and emergency support from non-coresident kin (Curran et al., 2003; Raley & Sweeney, 2020). However, step-relationships vary greatly in quality and are generally not as close or supportive as biological relationships (Kalmijn et al., 2019; Patterson et al., 2022; Wiemers et al., 2019). Ties between step-grandparents and step-grandchildren are weaker than those between biological grandparents and grandchildren, and full siblings remain closer than do half-siblings (Raley & Sweeney, 2020; Thomson, 2017).
Most studies on remarriage and social relationships focus primarily on step-kinship, particularly relationships between stepparents and stepchildren. Few have examined other types of relationships such as friendship. As an exception, Lin (2016) used NSHAP data to show that remarried older adults have a higher proportion of friends in their networks compared to their first-married counterparts. This finding suggests that friendship likely takes a more important role in the social life of remarried people.
Cohabitation has also become popular among older adults due to multiple demographic and social changes in the past few decades. According to the latest estimate, the proportion of unmarried adults who were cohabiting doubled from 7% to 14% between 2000 and 2016. About 90% of cohabitors were either divorced/separated or widowed; cohabitation is similar to remarriage in that it tends to be a higher-order union (Brown et al., 2006). Also like remarriage, cohabitation is considered an incomplete institution (Brown & Wright, 2017). In terms of social relationships, older cohabitors are more like divorced and the remarried than the first-married persons, partly because among older cohabitors aged 51 and older, about 71% were divorced/separated (Brown et al., 2006). For example, older cohabitors were less likely to have friends who lived nearby than first-married and remarried persons (Brown et al., 2006). They also had lower contact frequency and less positive relationships with their adult children than first-married persons. Older cohabitors did not differ from divorced and remarried persons in terms of contact frequency and relationship quality with adult children (Wright, 2017). However, little is known about cohabitors’ relationship quality with friends and family members.
The Role of Gender
There is a significant gender difference in social relationships due to biological, social, and psychological factors (Harling et al., 2020). Women are often socialized to invest in social relationships and act as kin keepers of the family. Empirical research has found that women tend to have larger and more intimate social networks (Ermer, 2021; Tifferet, 2020). Thus, marital status and gender may interact to influence the structure and quality of social relationships. For example, divorce negatively affects men’s social networks and social support more than women’s (Dykstra & Fokkema, 2007; Lin et al., 2022). This is partly because, before divorce, wives usually maintain a family’s various ties to friends and kin while husbands largely rely on their spouses for social support. Therefore, divorce tends to sever more ties for men than for women. Moreover, when children are involved, mothers often keep custody and remain close to them over the life course. Many studies have shown that divorced men tend to have smaller social networks and lower levels of contact with their children than divorced women (Kalmijn, 2007; Shapiro, 2003).
The literature on how widowhood may interact with gender to influence social relationships is mixed, depending on the dimension of social ties examined. One study has found no gender differences in talking frequency with friends and neighborhood ties among widowed persons (Ermer, 2021). However, other studies have found that widowers tend to have less contact with their children and perceive less emotional support from kin than widows (Curran et al., 2003; Kalmijn, 2007). Research on remarriage and social relationships has largely focused on intergenerational ties too: Remarriage tends to negatively affect contact and relationship quality with adult children, and the effect is stronger for men (Kalmijn et al., 2019; Lin et al., 2022). It is not clear how gender interacts with cohabitation to influence social relationships after midlife. One paper on intergenerational relationships among cohabitors found that cohabiting mothers were similar to first-married, remarried, divorced, and widowed mothers in terms of positive relationship quality with their children. However, cohabiting mothers also reported higher levels of negative relationship quality with their children than first-married and widowed mothers. In comparison, cohabiting fathers reported lower levels of positive relationship quality with their children than first-married and widowed fathers but did not differ from fathers with other marital statuses (Wright, 2017).
In sum, previous research on social relationships and marital status is limited. Most studies tend to focus on structural aspects of social relationships, such as parent-child contact. Few have distinguished remarried persons from their first-married counterparts or included cohabitors as a distinct marital status group. Even fewer have examined the relationship quality of a range of social ties. In order to fully understand the relationship between marital status and social relationships in late adulthood, it is important to investigate a variety of social ties as well as the quality of these ties, because marriage may promote some intimate ties (e.g., with children) while undermining others (e.g., with friends and relatives).
Research Questions and Hypotheses
Building on previous work, we have developed competing hypotheses for the link between marital status and friendship and kin networks: (1a) First-married and remarried persons have more friends and relatives they feel close to and more frequent contact with them compared to cohabiting, divorced/separated, widowed, and never-married persons, according to the “marriage as an integrative institution” perspective; (1b) First-married and remarried persons have fewer friends and relatives and less frequent contact with them compared to cohabiting, divorced/separated, widowed, and never-married persons, according to the “marriage as a greedy institution” perspective. Due to the scarcity of prior research on the quality of relationships with friends and relatives across marital status groups, we will not test a specific hypothesis on this dimension of social relations and will just explore the associations.
Next, based on the literature on intergenerational relationships across marital groups, we hypothesize that (1) first-married persons have more children whom they feel close to compared to all other marital status groups. We also expect that (2) first-married persons have more positive and less negative relationships with their children and relatives compared to all other marital status groups. Finally, we hypothesize that (3) remarriage, cohabitation, divorce, widowhood, and singlehood have stronger effects on social relationships for men than for women.
Data and Methods
Data
We used data from the 2006 and 2008 waves of the HRS. The HRS surveys a nationally representative sample of noninstitutionalized adults aged 51 and over in the U.S. every 2 years either by telephone or in person. It oversamples Blacks and Hispanics and collects detailed information on cognitive, physical, economic, work, and family conditions as well as health behaviors. The original HRS cohort consists of respondents aged 51–61 in 1992. The panel has been refreshed every 6 years with a new sample of persons aged 51–56. Since 2006, the HRS has collected psychosocial and lifestyle data biennially in the form of self-administered questionnaires that are left with respondents upon the completion of an enhanced in-person interview. Respondents are asked to return the Leave-Behind Questionnaire (LBQ) by mail. A random half of the HRS sample was selected to participate in 2006, and the other half was selected to participate in 2008. Response rates for the LBQ were high: 87.7% in 2006 and 83.7% in 2008 (Smith et al., 2017). Some of our variables (e.g., household income) come from the RAND HRS data.
In this study, we combined the 2006 and 2008 LBQ samples (N = 13,826). Our analytic sample was restricted to community-dwelling respondents aged 51 and older who had no missing values of covariates such as age, gender, education, and marital status. Based on these criteria, our final analytic sample was 13,087 respondents. To maximize statistical power, we allowed our sample sizes to vary depending on the number of valid responses for each dependent variable. So, our final sample sizes ranged from 11,636 to 12,087.
Measures
Dependent Variables
We measured the quantity and quality of three types of social relationships: friends, children, and family members. Three indicators of network size measured the number of close relationships with friends, children, and family members (excluding spouse), respectively. Each respondent was asked how many of their friends, children, and immediate family members (e.g., parents, siblings, or grandchildren) they felt close to. Contact frequency was measured as the mean of three items that asked how often the respondents met up, spoke on the phone, or wrote/emailed with their friends, children, and family members they did not live with. The three items were scored on a scale of 1 (three or more times a week) to 6 (less than once a year or never). We reverse-coded the variables so that higher scores mean higher contact frequency. The sample for contact frequency was limited to those who reported having at least one friend, child, or family member.
The relationship quality of the three types of relationships was measured by both positive (social support) and negative aspects (social strain). Positive relationship quality with friends, children, and family members was measured as the mean of three items that asked about perceived social support: “How much do they really understand the way you feel about things?” “How much can you rely on them if you have a serious problem?” and “How much can you open up to them if you need to talk about your worries?” Items were scored on a scale of 1 (a lot) to 4 (not at all). We reverse-coded the variables, so higher scores mean greater social support. Negative relationship quality of each type of social relation was measured separately as the mean of three items that asked about perceived social strain: “How much do they criticize you?” “How much do they let you down when you are counting on them?” and “How much do they get on your nerves?” Items were scored on a scale of 1 (a lot) to 4 (not at all). We reverse-coded the variables, so higher scores mean greater social strain. The Cronbach’s alpha coefficients of the positive and negative relationship variables of the three types of social relationships range from 0.75 to 0.86.
Independent Variables
Our key independent variable is marital status. It was categorized into six groups: first married (reference), remarried, cohabiting, divorced/separated (divorced thereafter), widowed, and never married. Variables measuring SES included education (in years) and household income. Household income was assessed as the total earnings from all sources, including investment returns, pensions, annuities, and welfare payments in the year prior to the interview. We used the RAND version of household income data, in which missing values were imputed (RAND HRS, 2016). To adjust for the skewness of the income distribution and differences in household sizes, we divided the original household income by the square root of household size, added one to eliminate zero incomes, and applied the logarithm (Glymour et al., 2008). Health conditions included self-reported health and limitations in activities of daily living (ADL). Self-reported health was the participant’s rating of their current health status on a 5-point scale, in which higher values represented better health. Activities of daily living limitations indicated the number of routine activities (e.g., eating, bathing, and mobility) with which respondents had difficulty or needed help (range: 0–5).
Covariates included age (in years), gender (1 = female), race/ethnicity (non-Hispanic Whites, Non-Hispanic Blacks, Hispanics, and others), and an indicator for survey year (1 = 2006; 0 = 2008).
Analytic Strategy
We first examined the three types of social relationships across the six marital status groups. Next, we used multiple linear regression models to investigate the associations between the three sets of social relationships and marital status. We estimated three models for each outcome variable. Specifically, Model 1 tested the association between each type of social relationship and marital status, controlling for demographic covariates and survey year. To test whether associations found in Model 1 can be partially explained by SES and health, we added education, household income, self-reported health, and ADL limitations in Model 2. In Model 3, interaction terms between marital status and gender were added to Model 2 to examine whether the associations between marital status and each type of social relationship vary by gender.
All regression analyses used LBQ weights to adjust for selection probabilities and non-response. We conducted analyses using Stata 16.
Results
Descriptive Statistics
Weighted descriptive statistics for three types of social relationships across marital status (N = 13,087).
Note. Differences by marital status were examined using Pearson’s chi-square test for categorical variables, and t-test for continuous variables. SD = standard deviation.
aStatistically different from remarried adults at p < .05.
bStatistically different from cohabitors at p < .05.
cStatistically different from divorced adults at p < .05.
dStatistically different from first-married adults at p < .05.
eStatistically different from widowed adults at p < .05.
fStatistically different from never-married adults at p < .05.
In terms of relationships with friends, remarried, cohabiting, and divorced persons had significantly fewer close friends than their first-married counterparts. However, all the unpartnered groups had more frequent contact with their friends and reported more support from friends compared to first-married and remarried persons. First-married persons also had significantly less strain in relationships with friends than all the other marital status groups except widowed persons. Remarried persons were largely similar to cohabitors except that they reported less strain with friends than cohabitors. There were only a few significant differences among the unmarried groups. For example, divorced persons had fewer close friends but more strain with their friends than widowed persons.
In terms of relationships with children, first-married persons had more children they felt close to, had more frequent contact with their children, and reported more support and less strain from their children than all the other groups except for widowed persons. Remarried persons had more frequent contact with their children than cohabiting and never-married persons but less frequent contact than divorced and widowed persons. They also reported more support from their children than cohabitors.
As for relations with family members, widowed older adults stood out as the most advantaged group. For example, they reported having more close family members, more frequent contact, more support, and less strain than their first-married counterparts. Compared to first-married persons, remarried persons had fewer close family members, less frequent contact, less support, and more strain; cohabitors, divorced and never-married persons all had more strain with their family members, although never-married persons reported greater family support.
Weighted descriptive statistics of independent variables by marital status.
Note. Differences by marital status were examined using Pearson’s chi-square test for categorical variables, and t-test for continuous variables. SD = standard deviation.
*p < 0.05; **p < 0.01; ***p < 0.001 (reference group = first married).
Regression Results
Relationships with Friends
Linear regressions on relationship with friends.
Notes. Model 1 controlled for age, gender, race, and survey year; Model 2 controlled for age, gender, race, survey year, log household income, education, self-reported health, and ADL limitation scores; Model 3 added interaction terms between marital status and gender to Model 2 (only significant interaction terms were reported).
*p < 0.05; **p < 0.01; ***p < 0.001.
Linear regressions on relationship with children.
Notes. Model 1 controlled for age, gender, race, and survey year; Model 2 controlled for age, gender, race, survey year, log household income, education, self-reported health, and ADL limitation scores; Model 3 added interaction terms between marital status and gender to Model 2 (only significant interaction terms were reported).
*p < 0.05; **p < 0.01; ***p < 0.001.
Linear regressions on relationship with family members.
Notes. Model 1 controlled for age, gender, race, and survey year; Model 2 controlled for age, gender, race, survey year, log household income, education, self-reported health, and ADL limitation scores; Model 3 added interaction terms between marital status and gender to Model 2 (only significant interaction terms were reported).
*p < 0.05; **p < 0.01; ***p < 0.001.
Relationships with Children
In terms of relationships with children, the results in Table 4 show that in general, all five groups are disadvantaged compared to first-married persons. As shown in Model 1, with demographic controls, compared to first-married persons, all the unmarried groups except for widowed persons had fewer children they felt close to. Moreover, all the unmarried groups were disadvantaged in contact frequency and perceived social support from their children (ps < .001). In addition, divorced persons also reported greater social strain from their children (p < .01). While remarried people reported a similar number of children they felt close to as their first-married counterparts, they had less frequent contact, less support, and greater strain from their children (ps <.01). When we added SES and health variables into Model 2, again all the coefficients changed very little. The only exception is that in terms of social strain, the coefficient of divorced adults was reduced from 0.07 to 0.05 and was no longer statistically significant. Additional analysis (results not shown) showed that although remarried persons were disadvantaged compared to first-married persons in relationships with children, they were more advantaged in a few aspects compared to some of the unmarried groups. For example, they had significantly more children they felt close to than cohabiting, divorced, and never married persons (ps < .05), net of all the controls. They also had more frequent contact with their children than cohabiting and never-married adults (ps < .05). However, they reported less support from children than divorced and widowed adults, all else equal (ps < .01).
In Model 3 of Table 4, we tested whether the association between marital status and relationships with children varied by gender. We found several significant gender interactions. First, there was a significant gender interaction with widowhood on the number of close children (b = 0.42, p < .05). This suggests that the link between widowhood and the number of close children was stronger among men than among women (i.e., for men, bwidowed = −0.50, p <. 05; for women bwidowed = −0.08, p > .1). In terms of gender and marital status interactions on contact frequency, our hypothesis was largely supported. For example, although remarried, cohabiting, divorced, and never married adults had less frequent contact with their children, this association was significantly stronger for men than for women (ps < .05). Specifically, the coefficients for remarried X female, cohabiting X female, divorced X female, and never married X female were 0.31, 0.53, 0.26, and 0.78, respectively. Similarly, both remarried and cohabiting adults perceived less support from their children, and again this association was stronger for men (i.e., coefficients for remarried X female and cohabiting X female were 0.25 and 0.37, respectively; ps < .01). In addition, gender also interacted with cohabitation and divorce on social strains (ps < .05). In fact, cohabiting men were not significantly different from married men in perceived social strain from their children. Additional analysis (i.e., switching the reference category of gender) showed that cohabiting women reported significantly greater social strain with their children (bcohabiting = 0.20, p < .05) than first-married women. Divorced men (bdivorced = −0.10, p < .05) perceived less social strain with their children, while divorced women (bdivored = 0.14, p < .001) reported greater social strain compared to their first-married counterparts. Comparing the results across the different dimensions of relationships with children, it seems that while divorced men had fewer close children, less contact, and less social support from their children, these relationships also had reduced opportunities for social strains in later life.
Relationships with Family Members
As shown in Table 5, there are more similarities than differences in relationships with family members across marital status groups. Nonetheless, there are still a few significant differences. The most notable differences are between remarried and first-married persons. Model 1 shows that remarried persons reported fewer close family members (p < .01) and less contact with and support from their family members (p < .001). Cohabitors also reported fewer close family members than first-married persons (p < .05). On the other hand, divorced adults had less frequent contact (p < .05) but greater strains with their family members (p < .01). The never-married adults had both greater support (p <. 01) and strain (p <. 05) from their family members.
In Model 2, these differences found in Model 1 were largely robust after we controlled for SES and health indicators, except that divorced adults were no longer significantly different from first-married adults in all dimensions of social relationships with their family members. In Model 3, we tested whether marital status differences in family relationships varied by gender and found very few significant gender interactions. The only significant gender interactions were observed in perceived social support for divorced and never-married adults and frequency of contact for remarried adults. Interestingly, we found that although divorce and singlehood were associated with more support from family members, the links were stronger for men than for women (p < .05). Additional analysis showed that the positive association between divorce/singlehood and perceived support was only statistically significant for men, not for women (results not shown). The negative association between remarriage and frequency of contact with family members was also stronger for men than for women (p < .01).
Sensitivity Analysis
We also examined potential differences in social relationships between more recently widowed adults and longer-widowed adults, defining these groups as having been widowed for less or more than 2 years, respectively. In the full model (results not shown), we found a few significant differences between the two groups in relationships with children and family members. The more recently widowed had more frequent contact (p < .01) and more support (p < .05) from their children; they also had more frequent contact with their family members (p <. 001). We did the same analyses comparing more recently divorced (i.e., less than 2 years) and longer-divorced adults and did not find any significant differences between the two groups, except that the more recently divorced adults reported more support from their friends (p < .05). However, we have a small sample of more recently divorced adults, so the results should be interpreted with caution.
Discussion
Based on a national sample of older adults in the U.S., our study shows that marital status matters for multiple types of later-life social relationships. Our findings provide novel evidence on the variation in friendship by marital status in later life, while confirming several patterns shown in prior research. First, although unmarried persons (i.e., divorced, widowed, and never married persons) are disadvantaged in some dimensions of social relationships, especially in terms of relations with their children (supporting the “marriage as an integrative institution” perspective), they also show some advantages in their friendship networks. Notably, despite having a similar number of close friends as first-married adults, unmarried adults contact their friends more frequently and report both greater support and greater strain from friends, net of covariates. This finding supports the “marriage as a greedy institution” perspective that marriage tends to reduce engagement with other relationships, such as friends, and limits commitment to collective social life (Sarkisian & Gerstel, 2012, 2016). It also indicates that friendship may partially compensate for not having a spouse/partner (Guiaux et al., 2007; Ha, 2008) and that more interaction with friends may involve ambivalent feelings—senses of both support and strain (Hsieh & Liu, 2021).
Second, our results show that in addition to current marital status, marital history matters. There are significant differences between first-married adults and remarried adults in several dimensions of social ties. As an “incomplete institution,” remarriage had disadvantages in almost all dimensions of social relationships examined. Consistent with findings from prior research that step-relationships may not be as close as biological relationships (Kalmijn et al., 2019; Raley & Sweeney, 2020), our study shows that remarried adults have less contact with both their children and family members and perceive these relationships as less supportive and more stressful than their first-married counterparts. Moreover, as one of the few studies examining friendship in late adulthood, our research finds that remarried individuals also have fewer close friends than first-married individuals. This may suggest that the loss of friends from a previous marriage may not be fully compensated for by new friends gained from remarriage. Similarly, cohabitors also have fewer close friends, children, and family members and report poorer relationship quality in some of these social ties compared to first-married adults. While we hypothesized that SES and health conditions may partially explain the relationship differences across marital status groups (Schwartz & Litwin, 2019; Waite & Gallagher, 2000; Zhang et al., 2016), our findings suggest otherwise: socioeconomic and health factors explain little of the differences described above.
Lastly, we find the most salient gender difference in the relationship with children. While remarriage, cohabitation, and divorce are associated with less contact with children, these associations are stronger for men than for women. We also found that while divorced men had less social strain with their children, divorced women had greater social strain with their children compared to their first-married counterparts. Previous research has shown that women are more likely to assume the role of the kin keeper and caregiver in their family, taking on tasks such as facilitating communications among children and their spouses, and thereby bear more negative consequences of performing emotion and care work (Thomeer et al., 2015; Wong & Hsieh, 2019). This finding of greater strain with children among divorced women but not divorced men implies that while divorced women may be more connected with their children than men, as prior research has also shown, these relationships may also have costs and take a toll on their health and well-being. It is therefore critical for research to examine both positive and negative qualities of social relationships (Song et al., 2021).
There are a few limitations in our study that warrant further investigation. We provide a comprehensive examination of the association between marital status and three major relationship types in late adulthood based on a cross-sectional study design. Therefore, our findings do not indicate causal directions and pathways. Prior studies suggested mixed results on the contribution of social relationships to marital transitions such as divorce (Buyukkececi & Leopold, 2021; McDermott et al., 2013). We tested the possibility of this reverse causation—whether and how relationships with friends, children, and other family members in the baseline predicted the risk of divorce in the 2-year follow-up—and found little evidence for any linkage (results not shown). Taking advantage of the longitudinal feature of the HRS, future research should continue to explore how marital status may predict change in social relationships and vice versa over the life course. Moreover, due to the small sample size for some of the marital status groups in our sample (e.g., cohabiting and never married persons), we do not have enough statistical power to investigate whether the associations between social relationships and marital status vary by characteristics beyond gender, such as race/ethnicity, educational attainment, the length of singlehood, and sexual orientation. However, we did find that the length of widowhood mattered: The more recently widowed (<2 years) had more support from children and more frequent contact with their children and family members than those who were widowed for more than 2 years. Future studies will benefit from exploring heterogeneity across social groups if data allow. Finally, in our study social support was measured as perceived support rather than received support (e.g., in-kind and money transfers). We focused on perceived support because prior research has noted that it is more strongly and positively linked to health and well-being than received support (Haber et al., 2007; Reinhardt et al., 2006; Wethington & Kessler, 1986), and thus a comparison of perceived support across marital status groups may offer more implications on the health and well-being of older adults. However, future research should consider examining how received support varies by marital status to fully gauge variations in the actual amounts of social resources by marital status when there is a need.
Despite these limitations, our study advances the understanding of social life in late adulthood by assessing multiple domains of relationships (friends, children, and family members) in relation to marital status, examining both the quantity and quality (positive and negative traits) of those relationships, and testing differences between older men and women. Our findings can inform future aging research and elder service programs on the interconnection between marital status and social vulnerability for older adults.
Footnotes
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported in part by the National Institute on Aging (Grants R03 AG062936 and R01 AG061118); 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). An earlier version of the paper was presented at the 2022 Annual Meeting of the Population Association of America, Atlanta, Georgia.
Open research statement
As part of IARR’s encouragement of open research practices, the author(s) have provided the following information: This research was not pre-registered. The data used in the research are available. The data can be obtained at:
. The materials used in the research are available. The materials can be obtained by emailing:
