Keywords

Close kin, particularly parents and adult children , have, barring any dramatic estrangement, a distinctive role in individuals’ support networks by virtue of their lifelong connections and the extent to which people rely on them through time. According to a recent Pew Research Center report, 60% of Americans provided practical support, such as help with errands and housework, to their aging parents in the previous year (Pew 2015). Over half reported giving in-kind assistance and almost two-thirds financial assistance to their adult children (see also Robinson and Schoeni 2010; Schoeni and Ross 2005). Time-diary studies suggest that parents and their adult children interact even more frequently and that most of them engage in mundane yet meaningful social exchanges on a weekly, and often daily, basis (Fingerman et al. 2016). Relationships between parents and adult children are assumed to be “special” and different from other types of relations due to the high level of emotional involvement and strong feelings of commitment they entail (Finch and Mason 1993; Silverstein et al. 2006; Wellman and Wortley 1990). Yet variation exists in the functions and quality of close kinship ties. In recent decades, major demographic, economic, and normative changes have affected the availability of immediate kin and altered the dynamics of intergenerational relationships (see reviews in Bengtson 2001; Johnson 2000; Swartz 2009).

In the present study we use data from the first wave of the UCNets project, a longitudinal study of personal networks , life events , and health in the greater San Francisco Bay Area, to learn in an inductive way about the role that close kin play in people’s personal networks. Specifically, we ask: (1) Who has parents or adult children available and accessible to help them? (2) Given that such immediate family are available, who reports an active connection to parents or adult children? (3) For people who have an active connection to parents or adult children, what role do these kin play in their network? And (4) to what extent is their connection related to other characteristics of their relationships? Recent sociological studies of familial exchanges have typically examined relations with kin in general without making a distinction between immediate and more distant kin. By contrast, gerontological research has mainly focused on the parent-adult child dyad and treated it somewhat in isolation from its broader social context. Our analytic approach is different. We use an egocentric network methodology, which collects data on to whom individuals are connected and the characteristics of those connected people; it does not prompt respondents to specifically consider their ties to kin. Hence, this study provides an unusually rich exploration of the role played by close kin, not presuming their importance but instead locating them within people’s larger spheres of activity and personal networks.

Among the findings discussed below is evidence of the interdependence of generations and, in particular, of the interdependence both upward and downward by the “sandwich” generation; the pervasiveness of gender differences in how much and what kinds of support parents and children provide one another; differences suggestive of a gendered division of labor in generational relationships; the continuing importance of geographical proximity for many aspects of filial ties; and paradoxical class differences in ties to parents. Overall, our findings indicate that social involvement with close kin is high and that kin play an important role in support systems.

Background

Much variability exists in the extent to which people are involved with and rely on kin for support. We know, for example, that getting married often has the paradoxical consequence of creating new, formal kin ties with in-laws but also reducing interaction with other types of kin, including parents, and the amount of support received from them; widowhood and divorce often have the reverse effect (Gerstel and Sarkisian 2006; Sarkisian and Gerstel 2008; Guiaux et al. 2007; Kalmijn 2012; Morgan and March 1992). Another important determinant of involvement with kin is gender. Numerous studies show that for both cultural and structural reasons women have more frequent contact with family members and are more likely to engage in social exchanges with them than do men (e.g., Hogan et al. 1993; Fischer 1982; Roschelle 1997; Sarkisian and Gerstel 2004; Wellman and Wortley 1990). Specifically, adult daughters more often provide support to aging parents than do adult sons (Lawton et al. 1994; Rossi and Rossi 1990; Silverstein et al. 1995; Silverstein et al. 2006). Differences in involvement with kin by race and ethnicity have also been widely documented. Research portrays a rather complex picture with results varying by gender, generation, and the type of support examined. By and large they suggest that whites more frequently engage in the exchange of emotional and financial support with kin whereas blacks and Latinos are more likely to live with or in close proximity to kin. These differences, however, are mostly explained by the lower socioeconomic status of blacks and Latinos, which affects both their level of need and the amount of resources available for exchange (Hogan et al. 1993; Lee and Aytac 1998; Sarkisian and Gerstel 2004; Sarkisian et al. 2007).

We also know that education and income matter for social involvement with kin. More educated people tend to live farther from kin and are, at least proportionately, less involved with kin than are less educated people (Chan and Ermisch 2015; Compton and Pollak 2009; Fischer 1982; Kalmijn 2006). Yet, the higher-educated also tend to have kin networks with the greatest upward reach in class standing (Goldstein and Warren 2000), providing at least a latent source of greater social support . Although reliance on kin constitutes an important coping strategy in low-income people’s struggle to make ends meet (e.g., Domínguez and Watkins 2003; Edin and Lein 1997; Nelson 2005), research suggests that overall they receive lower levels of support, particularly financial support, than people with higher incomes (Hogan et al. 1993; Roschelle 1997; Goldstein and Warren 2000). The greater ability of wealthier parents to help their young adult children in the transition into adulthood has important implications for the reproduction of class, as well as racial, inequalities (see review in Swartz 2009).

Despite the abundant literature on kin relations, there is still much we do not know about the contexts, modalities, and shapers of kin support. One such shaper of people’s involvement with immediate family is simply their availability. Obviously, an older person with a single child who lives a thousand miles away is in a different situation than one with three children who live nearby. Important demographic shifts that have occurred in the last three to four decades have had a profound impact on the number and types of kin available for intergenerational relations. Most notably, declining mortality and fertility rates have led to longer years of shared lives between generations but with fewer adult children available to provide support to aging parents in times of need. At the same time, smaller families also means that parents have more resources to share with each of their adult children (Bengtson 2001). The higher prevalence of divorce and single-parenthood has further contributed to the diversification of family forms and intergenerational relations (see also Johnson 2000).

Another important source of variation is residential mobility. Movers are likely to move away from kin in response to better educational and occupational opportunities (Pugh 2015; Rosenfeld and Kim 2005). And while new communication technologies have made it easier for family members to stay in touch regardless of their geographical location, some forms of support, such as taking care of a sick relative or providing help with childcare, are facilitated by physical proximity. The relatively smaller role of kin among the better-educated that earlier research has found may “simply” reflect their higher tendency to move long distances (Fischer 2002), as well as their lower birth rates.

Yet availability and accessibility are not the whole story. People’s lifestyles vary from less to more kin-centered. Some spend considerable time with and rely extensively on kin; others choose to rely less on relatives and more on non-kin (Agneessens et al. 2006; Giannella and Fischer 2016) and those lifestyle variations may in turn reflect differences by socioeconomic status, generation, and culture. For example, Hansen (2005) shows in her in-depth study of the networks of care of children that families who adhered to the ideology of self-sufficiency and independence of the nuclear family, typically families of middle-class background, tended to restrict their involvement with the extended family and relied on them less often for help with childcare than other families. The quality of kin relationships may be especially important in this context. Some scholars have suggested that in contemporary society, relationships with close kin have become less motivated by felt obligation and, similarly to other ties, more motivated by felt closeness and affection (Beck and Beck-Gernsheim 1995; Finch and Mason 1993; Van Gaalen and Dykstra 2006). In support of this view, research has shown that people feel more motivated to help kin with whom they get along and to whom they feel affection and love (Fingerman et al. 2016; Silverstein et al. 2006).

Overall, the extensive research on involvement with kin suggests that family networks are flexible and dynamic. The major goal of this study is to examine the scope and correlates of the availability and accessibility of parents and adult children and, given that these ties exist, to gain insights into the role they play in people’s larger networks. To address these issues we draw on the first wave of the UCNets project, described below.

Data and Measures

The UCNets project has collected extensive descriptions of personal networks from two samples of respondents in the greater San Francisco Bay Area: 690 respondents aged 50 to 70 and 495 respondents aged 21 to 30 completed wave 1 of the panel survey.

Sampling

We drew samples from six San Francisco Bay Area counties, using address-based methods, sending solicitation letters to households randomly selected from 30 randomly-selected census tracts. The letters invited any member of the household who was either 21 to 30 years old or 50 to 70 years old to join the panel study, a commitment entailing being interviewed three times over about a 4-year period. The study focused on these two specific age groups to maximize the number of key transitions and life events respondents would likely experience between waves of the survey. The letter offered escalating payments for each interview in order to entice staying on the panel. It directed would-be respondents to call in or to use a web site to register. The screening procedure randomly assigned qualifying respondents to either a face-to-face interview (75% of cases) or a web survey (25%). The in-person and online instruments were substantively identical; later reports will address mode effects. This outreach procedure sufficed for the 50–70 year-olds; to reach enough 21–30 year-olds, however, we had to resort to extra means, as described below.

The overall yield from the letters was low, we estimate at about 10%, which would be expected given (a) the narrow age criteria for qualifying; (b) the multi-year commitment; and (c) generally declining survey response rates (National Research Council 2013). Young respondents were especially difficult to recruit. We therefore supplemented the initial, address-drawn sample of 162 young adults with an additional 36 recruited through previous respondents and an additional 297 recruited through social media (Facebook solicitation allows one to target a region—the Bay Area, here—and specific ages.)Footnote 1 Our field contractor, Nexant, collected the data from the middle to the end of 2015. In the end, 522 older respondents were interviewed face-to-face and 168 did the survey online; 141 young respondents were interviewed face-to-face and 354 did the survey online (which includes the referred and Facebook-recruited respondents). Our final sample, described in Table 6.1, skews toward women and the better-educated. It is, however, diverse in various demographic dimensions and we use those as controls in our models.

Table 6.1 Sample characteristics by age group: Percentages (n in parentheses)
Full size table

Measures

First, the UCNets survey instrument took a rough “census” on the existence and location of various types of kin. Most relevant to this paper, it asked respondents if their mothers and fathers were still alive and, if so, whether each of the parents lived within a one-hour drive.Footnote 2 It also asked respondents if they had any adult children or step-children who did not live with them, how many of each, and whether any of them lived within a one-hour drive.

Second, the UCNets instrument solicited the names of the people, or alters, to whom respondents were connected by asking several “name-eliciting’ questions. The protocol then applied several “name-interpreting” questions to the list of alters to obtain descriptions of the named individuals and of the ties they had with the respondents. We focus here on five types of name-eliciting connections. We asked respondents to name the people with whom they:

  1. 1.

    Socialized—the people with whom they usually got together and did social activities such as going out to restaurants, concerts, plays, clubs, sports, other events, or hanging out (up to 9 names);

  2. 2.

    Confided in or sought advice—two name-eliciting questions: those whom they confided in about relationships, important life experiences, and the like, and the people whose advice they sought or would seek to help make an important life decision, for example, about taking a job, family issues, or health problems (up to 6 names for each item here and for each of the remaining questions)Footnote 3;

  3. 3.

    Practical help – the people who had given the respondent practical help in the previous few months, such as moving furniture, doing repairs, picking up something at the store, looking after a child, and giving a ride;

  4. 4.

    Emergency help – The people whom the respondent would ask if she or he were seriously injured or sick and needed some help for a couple of weeks with things such as preparing meals and getting around; and

  5. 5.

    Provides support – the people whom the respondent helped out practically, or with advice, or in other kinds of ways at least occasionally. While topics 2–4 refer to the respondent as the actual or potential recipient of support, this one treats the respondent as a provider of support.

The instrument then asked respondents to specify how each person whom they named was related to them, choosing among a list of predetermined categories ( e.g., parent, child, sibling, neighbor, friend, coworker, and so forth). We identified mothers, fathers, adult daughters, and adult sons living outside the respondents’ households.Footnote 4 We were then able to calculate the percentage of kin elicited for each type of connection, such as the percentage of mothers (out of all mothers) who were named in the socializing question, or the percentage of sons (out of all sons) who provided practical help. Additionally, we created a global measure of inclusion in the network (referred to as “in network”), indicating whether the relative in question was named in at least one of the five name-eliciting questions.

Additionally, we used information about the characteristics of the relationship between the respondent ego and each named alter. These include emotional closeness , measured by a yes/no question asking whether the respondent feels close to the alter; and physical proximity, a yes/no question asking whether the alter lives within one hour drive from the respondent.Footnote 5

Finally, we examined a series of sociodemographic characteristics of the respondent. Gender is coded as a dummy (0 = “female”; 1 = “male”). Recall that UCNets gathered information among two age groups: 21–30 year-olds and 50–70 year-olds. We ran some of the analyses separately for each age group and included age as a control in others. In the latter case we made a distinction within the older cohort between respondents aged 50–59 and those aged 60–70. Married is coded as a dummy (0 = “no”; 1 = “yes”). In some of the analyses, for example when we estimated the factors associated with naming adult children in the network, we also controlled for the (natural log) number of adult daughters and sons the respondent had. Ethnicity/Race is measured with three dummies: Latino, Asian, black and other, with white used as the reference category. Education level is coded as a dummy indicating whether the respondent had a BA or higher degree. Family income refers to either the total household income for respondents who were married or living with a partner or other relative, or the individual income for respondents who lived alone (or with roommates). Income is measured with a dummy indicating whether the income (before taxes) was $75,000 or higher (0 = “no”; 1 = “yes”). We included two additional measures to capture the respondent’s residential history and potential migration: whether the respondent had been living in current town for two or fewer years, referred to as new resident in current town, and whether the respondent was born in California (0 = “no”; 1 = “yes”).

Table 6.1 shows the distribution of the sociodemographic variables by age group. Overall, both samples were predominantly female. Many more of the older sample, unsurprisingly, were married. The two age groups also appeared to significantly vary by ethnicity and race. About half of the young respondents were either Asian, Latino, black, or “other,” whereas the older ones were overwhelmingly white (more than 70%). Not surprisingly, older respondents had higher incomes. No meaningful difference was observed for level of education ; about 70% of respondents in both age groups had a BA or higher degree (about 30% of the young respondents were still engaged in schooling of some kind.) Sixty percent of the younger respondents had lived in their current town for two or fewer years, compared to only 6% among the older age group. The younger respondents, however, were more often born in California.

Findings

Descriptive Results: The Availability, Accessibility, and Mobilization of Close Kin

Table 6.2 presents a first, descriptive look at the availability, accessibility, and mobilization of kin . Because the raw numbers are subject to methods effects and because our samples are regionally and cohort specific, we should attend to internal comparisons rather than the absolute values. The results show differences by age group and the relative’s gender. Not surprisingly, almost all 21–30 year-olds reported parents who were alive, while most 50–70 year-olds had one or both parents deceased. Younger respondents were less likely to live near their parents, either mother or father, than older ones (about one fourth versus 40%). These results are somewhat unexpected. Considering their 30 or more years of opportunities for migration, older respondents might have drifted farther away from their parents. These results, however, may reflect the tendency of the young generation to move away from their family of origin in search of educational and occupational opportunities and that of the older generation to have their elderly parents live close to them so that they can more easily provide assistance to them. Given living mothers, the young were much more likely than the old to list their mothers in response to the name-eliciting questions (61% versus 36%). No such difference was found for reporting a father in the network (approximately 45%). We discuss in later analyses (see Table 6.5) the particular roles those parents played in each cohort’s lives.

Table 6.2 Kin availability: Percentage of respondents reporting parents or adult children who are alive, in close geographic proximity, and in the reported network, by age group
Full size table

Table 6.2 further indicates that among the 50–70 year-olds almost 40% had at least one adult daughter or adult son. Given a living child, it appears that sons were slightly more likely than daughters to live within one-hour drive from the parent, whereas daughters were slightly more likely than sons to be named in the network by the parent. Consistent with the latter trend respondents included on average slightly more adult daughters than sons in their network (means of 0.56 versus 0.50).

These numbers are consistent with the general findings in the literature that mothers play a more critical role in Americans’ lives than do fathers. They also suggest that daughters are only slightly more likely than sons to be part of the network but, as we will show in Table 6.5, the role that daughters assume in the network of their aging parents is substantially different from that of sons.

The Sociodemographic Correlates of the Availability, Accessibility, and Mobilization of Close Kin

In the next stage we examined the socio-demographic factors that are independently associated with the availability , accessibility , and mobilization of close kin . Tables 6.3 and 6.4 present the results of a series of logistic regression models accounting for the roles of parents and adult children, respectively. For each type of kin, we tested three models to predict the likelihood of (1) having the relative alive; (2) given that the relative is alive, that he or she lived within one hour drive; and (3) drawing on the name-eliciting data, that he or she appears in the respondent’s network in one or more of the five delineated roles. We present the effects as odds ratios. The results for parents, displayed in Table 6.3, show that having a mother or father alive was largely a matter of age , although there is a suggestion that high-income was also positively associated with the likelihood of having a living father. This result is not surprising considering the well-documented beneficial effect of socioeconomic status on health.

Table 6.3 Likelihood of kin availability: Odds ratios from logistic regression models predicting respondents’ reports on the availability, accessibility and network inclusion of parents
Full size table
Table 6.4 Likelihood of kin availability: Odds ratios from logistic regression models predicting respondents’ reports on the availability, accessibility, and network inclusion of adult children
Full size table

The next models examined the sociodemographic factors associated with living within one hour drive from the parent. Not much in the respondent’s background was independently associated with the chances that a parent lived nearby, with the exception of age , marital status, and state of origin. The models predicting accessibility show that, given that the parent was alive, by and large older respondents were more likely than younger ones to have a mother or father living within one hour drive from them. Married respondents were almost two times (OR = 1.81) more likely to have a father, but not a mother, who lived nearby than their non-married counterparts, a non-obvious finding. Whether the respondent was born in California appeared as an important determinant of having either parent within a one hour drive from them. Although we do not know about the full migration history of the respondents (e.g., they could have moved out of the state at some point before the survey and then come back to California), nor do we know about the migration history of the parent, this result suggests that respondents born in California and their parents remained rooted in their state of origin.

The models predicting inclusion in the network showed that older respondents were less likely than younger ones to name their parents in the network. Asians were less likely than whites to name living parents. Education level turned out to be an important determinant of network inclusion. Highly educated respondents (i.e., those with a BA or higher degree) were substantially more likely to name their mothers (OR = 2.15) and fathers (OR = 2.45) in the network than those with a lower level of education. Additionally, recently-arrived respondents more often named a parent in their networks. Finally, geographic proximity was another important factor; parents who lived nearby were likelier to appear as part of the respondent’s network than those who lived farther away. Later analyses (see Table 6.5) examine which specific kinds of interactions produced these associations.

Table 6.5 Types of connection to kin : Differences in mean percentage of respondents reporting a connection to kin , by gender of kin
Full size table

Table 6.4 repeated the analyses with the adult daughters and sons of the older cohort. Having an adult child living outside the household was largely a matter of demographics: Older and married respondents, as well as those born in California, were more likely to report having at least one adult child. Respondents with a BA or higher degree were less likely to have an adult child (significantly so in the case of daughters) than their less well educated counterparts.

Given that they had adult children and controlling for the number of adult daughters and sons, the oldest respondents (those aged 60–70) were more likely to report that at least one of their children, either a daughter or a son, vlived within one hour drive from them. These results most likely reflect the greater need for support of the oldest. Latinos were over four times (OR = 4.62) more likely to live near a son (but not a daughter) than whites. Respondents who were born in California were almost two times more likely to have at least one adult daughter (OR = 1.99) or son (OR = 1.74) living nearby compared to those born elsewhere.

As to respondents actually naming daughters or sons in their networks, conditional on having adult children and the number of adult children, Table 6.4 shows that 60–70 year-old parents were significantly more likely to mention at least one adult son in their network than were parents in their 50s. Latinos were substantially more likely to include at least one adult daughter in their network as compared to whites. Geographic proximity mattered a great deal. Respondents were significantly more likely to name an adult child in their network if at least one of their children lived nearby.Footnote 6 The independent association of having a daughter nearby (OR = 7.19) was especially pronounced and substantially higher than that of having a son nearby (OR = 3.61). Unlike previous research suggesting that relying on adult children for support varies substantially by the marital status of the parent (see reviews in Polenick et al. 2017; Swartz 2009), Table 6.4 shows no independent association. It is possible that the lack of effect resulted from our comparison of the married to the non-married, a category that included both divorced, never-married, and widowed respondents (unfortunately, due to the sample size we could not examine these groups separately).Footnote 7

The Role of Close Kin in the Network

We now turn to examine the question of what specific role parents and adult children reportedly played in respondents’ lives. For this analysis, we shifted from the respondent (i.e., ego) as the unit of analysis to the listed parents and children (i.e., alters) as the units of analysis: 424 mothers, 282 fathers, 251 daughters, and 213 sons. We focused on five kinds of connections to kin: socializing, confiding/being advised, receiving practical help, anticipating emergency help, and providing support to alter. The results displayed in Table 6.5 show systematic variations by type of connection, type of kin, the gender of the relative, and age group. For comparison purposes we also show the results for siblings (sisters and brothers), other relatives, and non-kin. These ties all play a distinct role in personal networks and their consideration is important in order to capture the broader picture of it, but this is beyond the scope of the present study. We plan to examine the meaning of different types of kin, as well as that of non-kin, for social involvement in future research. In this study we focus on intergenerational relations between parents and adult children.

Multiple comparisons are possible – by type of role, by type of relative, and by age group. We start by examining the role of parents in the networks of young respondents. Their answers suggest that they looked to their parents for emotional support and emergency help. More than two-thirds of the parents whom young respondents listed at all appeared in answer to the confide or advise questions (76% and 69% for mothers and fathers, respectively) and approximately half of them were mentioned as someone respondents would turn to in the case of an emergency. Approximately one-third of the parents were mentioned as providers of practical support. Relatively few parents of the 21–30 year-olds appeared as social companions or recipients of help. Also note that respondents were more likely to name their mothers than fathers as potential providers of emergency support (55% as compared to 45%) and more often indicated providing support to the mother than to the father (31% as compared to 17%).

Older respondents named their parents in distinctively other roles. They named their parents as confidants or advisers (although less so than younger respondents), but then most often as recipients of help and as social companions. Put simply, the young respondents appeared to rely on their parents in times for emotional and instrumental support, but were otherwise only modestly involved with them, while older respondents reported helping their parents and spending social time with them, but not much depending upon them. These results suggest that middle-aged parents provide a safety-net for their young adult children while elderly parents play a more complex role in the lives of their middle-aged children.

Table 6.5 further looks at the various roles played by adult children in the networks of the older cohort. Overall, involvement was quite high. Almost the same proportion of 50–70 year-old respondents’ grown children got help (about 60%) as did those respondents’ parents (about 54%).7 This finding provides a nice illustration of the crucial role of parents in late mid-life, “sandwiched” between two generations, in helping both the young and the elderly. It is noteworthy that respondents were as likely to help their adult daughters as sons in the network. At the same time, adult children apparently provided important support for their 50–70 year-old parents. More than 40% of both adult daughters and sons who were included in the network were named as emergency helpers and about half of them were named as social companions. Daughters, however, were more often named as confidants or advisors than were sons (46% versus 35%), while sons got called on for practical help more often than daughters (34% versus 24%), suggesting an emotional versus practical division of labor by gender of adult child.

A comparison to other kin and non-kin in the network further emphasized the distinct role played by close relatives as a major source of emotional and instrumental support in the lives of the respondents. By and large, the results of Table 6.5 show that non-kin mainly assumed the role of social companions in the network and were less often named as providing emotional support, especially among young respondents. By contrast, siblings, especially sisters, were often named as confidants or advisors and as providers of instrumental support. Their social involvement in the networks of the older respondents appeared to be as important as that of parents, but less so than that of adult children. Among the 21–30 year-olds, however, siblings assumed a less important role as potential helpers during emergencies than did parents, but they were more often mentioned as social companions.

Variation in the Role of Close Kin in the Network by Characteristics of the Relationship

Finally, in the last set of analyses (see Table 6.6), we tested the extent to which the roles assumed by parents and children in the network were associated with other characteristics of their relationships. What kinds of bonds to the kin went along with what kinds of support that they provided (or received)? Again using the named relatives as units of analysis, we focus on three major attributes of the relationship: degree of emotional closeness , geographic proximity, and frequency of contact. The results displayed in the first two columns of Table 6.6 show that parents, mothers and fathers, whom respondents reported as “close” tended to be the ones whom they said provided emotional support and emergency assistance. Respondents who felt close to their mothers were also more likely to socialize with her. No such effect was observed for fathers.

Table 6.6 Variations in connections to kin : Differences in mean percentage of respondents reporting specified connection to kin by relationship characteristics (closeness and geographic proximity)
Full size table

Similarly, adult children, both daughters and sons, were more often reported as “close” when they also played the role of confidant or advisors and also providers of practical support. By contrast, expecting emergency assistance from adult children was not related to how close the parent felt to them. This result seems to suggest that parents expect that they can turn to adult children for support when acute needs arise regardless of how close they feel, perhaps reflecting the norm of filial obligation. Interestingly, reporting that one provided help to mothers and daughters was likelier if the respondent also reported feeling closer to those mothers or daughters, but was not the case for either fathers or sons.

As expected, geographic proximity (examined in the next two columns of Table 6.6) was an important correlate of socializing with immediate kin. Mothers, fathers, daughters and sons were all more likely to be named as someone with whom the respondent socialized when they lived nearby. It makes sense to assume that geographic distance would also matter for the provision of mundane practical help, such as home repairs and childcare, but that it would be less of a factor in emergency situations. By and large, however, geographic proximity was an important correlate of both types of support. Respondents were more likely to name their mother, father, and adult daughter as the provider of both practical and emergency support when those relatives lived nearby rather than farther away. For sons, only emergency help was significantly related to geographic proximity. These findings may perhaps reflect the desire not to impose on close kin who lived far away. Geographic proximity also appeared to be an important factor when we examined the data from the opposite perspective, switching the unit of analysis from children to parents. Proximity did not matter for parents’ reporting that they provided support to their adult children, but it did matter for parents’ report of providing support to their own aging parents. Respondents were more likely to report helping their parents if they lived close to them than if they lived farther away.

Considering that recent technological developments have made it easier for people to communicate with members in their network regardless of their location, it is not surprising that physical distance was not associated with the likelihood of confiding or advising with adult children. This is consistent with previous research showing that emotionally supportive ties tended to be maintained over long geographic distances (Viry 2012). Interestingly, however, mothers and fathers were more likely to be named as confidents or advisors if they lived farther away than if they lived nearby. A possible explanation for this seemingly odd finding has to do with a selection effect by which parents get included in the network at all. In this case, it may result from having many respondents naming distant parents only in this role, as the high percentages reported in Table 6.6 (76% and 74% for mother and fathers, respectively) and the results in Table 6.5 seem to imply (cf. Fischer 1977, pp. 172–77).

Conclusion

Using the UCNets data, we have been able to place the dynamics of parent-child relations within the larger context of people’s support and exchange networks. Overall, our findings highlight the high level of connections between young adult children and their aging parents and the important place each has in the others’ systems of support. Yet we found substantial variation in the availability, accessibility, and mobilization of close kin.

We found that the likelihood of having adult children, as well as the likelihood of socially engaging with them, increased with age . By contrast, the likelihood of having parents and of naming them in the network declined with age. This trend may be due to the greater needs of older as compared to younger parents, which may be attended at least partially by their adult children, but it could also reflect the rise in intergenerational stake that both parents and children experience as they grow older (Giarrusso et al. 1995). Older parents reported spending time and engaging in social activities with their adult children. They also named their adult children as confidents or advisors and as an important source of support during emergency situations. These same parents, however, also reported providing much support to their adult children. In fact, and consistent with previous research showing that in most American families intergenerational support typically flows downstream from the parents’ to the children’s generation (Fingerman et al. 2011; Fingerman et al. 2013; Logan and Spitze 1996), it appeared that the 50–70 year-old parents in our sample played a greater role as providers of support to their young adult children than the reverse. About 60% of the parents indicated that they provided support to their adult children. This finding was echoed in the adult children’s report, with a large percentage mentioning their parents as providers of emotional and instrumental support.

Interestingly, and unlike previous research (e.g., Chan and Ermisch 2015; Lawton et al. 1994), we found that geographic proximity between generations increased with age . That is, elderly parents were more likely to live near their adult children than were middle-aged parents. This finding may have important implications for intergenerational relations because geographic proximity is likely to facilitate face-to-face interactions and the exchange of support (Grundy and Shelton 2001; Lawton et al. 1994; Ward et al. 2014). Indeed, we found that geographic proximity was a major determinant of the inclusion of close kin in the network. Mothers, fathers, adult daughters, and adult sons were all more likely to be named in the network if they lived within one-hour drive to the respondent than if they lived farther away. Our results showed that geographic proximity mattered much for the chances of socializing with close kin and for receiving support from them in both mundane and emergency situations. Altogether, they suggest that, even in the internet age and with the widespread availability of digital communication technologies, geographic proximity still matters (see review in Mok et al. 2010). By facilitating shared experiences, geographic proximity may contribute to reinforcing intergenerational bonds, which in turn may encourage children’s provision of support to their aging parents (Silverstein et al. 2006; Ward et al. 2014). Geographic proximity, however, did not matter for the receipt of emotional support. This finding is consistent with previous research showing the limited effect of geographic dispersion on the emotional supportiveness of personal contacts (Viry 2012).

Level of education was another important determinant of parents’ inclusion in the network. However, unlike previous research (Kalmijn 2006; Lawton et al. 1994; Greenwell and Bengtson 1997; Grundy and Shelton 2001), we found a positive, not negative, association between education and social engagement with aging parents. That is, highly educated respondents were more likely to include their mother and father in the network than those with a lower level of education. This finding resonates with studies showing more frequent social exchanges among families of higher as compared to lower socioeconomic status (Hogan et al. 1993; Roschelle 1997), and may perhaps suggest a lower level of family conflict among the better educated.

Furthermore, we did not find a significant association between level of education and the likelihood of living close to the parent’s home. Previous studies have indicated that the highly educated face greater labor market opportunities and are thus more inclined to migrate from their place of origin in search of jobs that would fit their education level and skills (Kalmijn 2006). Our findings, by contrast, seem to support the view that in contemporary society it has become more difficult for most people, especially in the young generation and regardless of educational status, to accumulate financial resources, develop careers, and obtain jobs that would provide stability and grant control over one’s geographic location (Greenwell and Bengtson 1997; Pugh 2015).

Although we did not find any differences by the respondent’s gender in the availability, accessibility, and general inclusion in the network of parents, we did find differences by the gender of the named relatives. Mothers played a very important and specific role in their adult children’s lives; they were substantially more likely to be named as a source of emergency support for their children than were fathers. With respect to adult children, consistent with the abundant literature on gender and involvement with kin (Rossi and Rossi 1990; Silverstein et al. 1995; Silverstein et al. 2006), we found that overall daughters were slightly more likely than sons to be included in their parents’ network. The gender gap was relatively small in size and therefore should not be overstated. More interesting, however, were the results suggesting a gendered division of labor in the caretaking of aging parents; we found that daughters were more often named as providers of emotional support than sons whereas sons, more so than daughters, tended to be called upon for practical help. Nevertheless, our results did not reveal a difference in the likelihood of mobilizing daughters versus sons during emergencies, which suggests that both daughters and sons may be motivated by a strong sense of filial obligation to provide support to aging parents when acute needs arise. Our finding that the parent’s mentioning of an adult child, regardless of the child’s gender, as someone to rely on in the case of an emergency was not related to how emotionally close the parent felt to that child further supports this possibility.

The quality of the relationship between parents and adult children, measured in this study with emotional closeness , was another important factor that helped explain variation in the role played by close kin in the network and the kin’s gender (see also Wellman and Wortley 1990). Overall, we found that respondents tended to mention their immediate kin as confidants or advisors when they felt emotionally close to them. Emotional closeness was also related to the perception that the parent, either mother or father, would provide support in the case of an emergency. Yet our findings further revealed that respondents were substantially more likely to report that they provided support to their mothers and adult daughters if they felt emotionally close to them, but no such effect was observed for fathers and sons. Previous research has suggested that the motivation to help kin differs by gender. For example, Silverstein et al. (1995) found that while daughters tended to provide support to aging parents out of affection for them, sons mainly helped their parents out of a sense of obligation and therefore the amount of assistance they provided to them was not related to the quality of their relationship. Following this rationale, our finding may reflect the different ways by which men and women frame their involvement and willingness to socially engage with close kin.

Finally, this study allowed for an examination of familial relationships across multiple generations and it underscores the particular position of middle-aged parents as a “sandwich” generation. This concept has been originally applied when referring to parents in their 40s and 50s who simultaneously care for dependent children and frail elderly parents. Scholars, however, have noted that in light of recent demographic changes a more common situation is that of parents in late mid-life (those 50–70 year-olds whom we examined in this study) who have both at least one surviving parent and one adult child who is still economically dependent on them (Grundy and Henretta 2006). These parents’ “sandwich” experience is the result of both the increase in longevity of the older generation and the longer time it takes today for the younger generation to transition into adulthood and reach independence (Fingerman et al. 2011; Fuerstenberg 2010; Swartz 2009). Our findings showed that parents in late mid-life were highly and simultaneously involved in the provision of support to both their adult children and elderly parents. This finding has important implications for the well-being of the middle generation, as well as for members of the generations above and below them. In future research we plan to examine how relationships and network dynamics across multiple generations change over time in response to the occurrence of various life-events and transitions , which are likely to affect both the level of need and amount of resources at the disposal of different family members.

To conclude, the findings presented here clearly suggest that even though families have undergone significant changes over the last few decades, changes that have been extensively discussed in the sociological literature, social involvement with close kin is high. Aging parents and their adult children socially engage with each other in a variety of ways and their relationships constitute an important source of emotional and instrumental support in both routine and emergency situations.