Keywords

The life course paradigm (Elder 1985; Elder 1994; Elder 1995; Marshall and Mueller 2003) focuses attention on individual life trajectories composed of interrelated transitions into and out of states such as marriage or employment. It examines the manner in which lives unfold in connection with the intersecting social rhythms of multiple careers (in domains including, e.g., intimate relationships , childbearing and childrearing, and work), constraints and opportunities associated with institutional structures and historical events, and the parallel lives of other persons. The paradigm has a clear affinity with the study of social networks : life course transitions often imply the formation of new relationships or the dissolution of previously-existing ones. Indeed, the perspective’s foundational principle of “linked lives ” (Elder 1995: p. 112) emphasizes the interdependencies among the life histories of persons connected by ties of kinship (especially), friendship , and other bonds.

The literature on social networks likewise alludes to ideas involving the life course when developing accounts for variations between the social networks of individuals (e.g. Marsden 1987), and in particular when theorizing about network change (e.g. Wellman et al. 1997); the connection is only sometimes made explicit (e.g. Bidart and Lavenu 2005; Kalmijn 2003). The occurrence of a major life event such as marriage, parenthood, employment or retirement, however, is readily seen as an occasion that may alter opportunities to form or continue social ties, closer and more distal alike.

The premise of this volume is that making more direct connections between the bodies of research on the life course and on social networks will be mutually profitable. This chapter, rooted in social network studies, contributes to that enterprise by examining general-population social survey data on the life course positions and social networks of U.S. adults. Emphasizing a conceptualization of the life course as a set of transitions (Alwin 2012), it examines the manner in which life course events shape networks. After reviewing findings of some prior studies that link these phenomena, it proposes opportunity-based network theories (e.g. Blau 1977)—Feld’s (1981) focus theory in particular—as a basis for understanding how life course transitions may shape the size and composition of individual social networks . It then examines General Social Survey data on how network size , network composition, and network activity vary across life course states. In keeping with a prominent emphasis in life course theory (Moen 2001), it considers the prospect that particular transitions hold different consequences for men and women.

Background

A number of prior studies consider the interplay between the life course and network development. Many of these examine cross-sectional differences in network structure and composition across life course states or stages; less commonly, they use longitudinal designs that measure both life course change and network turnover. We review some of the theoretical arguments and findings of this research here: a recurrent theme is the idea that a life course event such as marriage, divorce , or retirement can put the continuation of existing relationships at risk, while at the same time providing opportunities to initiate new ones.

A classic work on social support networks (Kahn and Antonucci 1980) made a close and overt connection between the life course and social networks . Its central concept was the “convoy,” defined as “a structure within which social support is given and received”—that is, a set of family , friends, and others who provide social support to a focal individual (Kahn and Antonucci 1980: pp. 253, 267). A convoy is, in essence, a personal or egocentric network (Crossley et al. 2015) composed of those “alters” involved in supportive transactions with a focal actor or “ego”. Kahn and Antonucci invoked role theory to link the life course to change in convoys, arguing that life events involve the acquisition of new roles and/or the shedding of previously-held ones; the changes in expectations associated with roles imply changes in role-dependent affiliations with others. Their argument suggested that a convoy’s most central members—typically linked to the ego via strong, “multiplex” relationships comprised of multiple strands associated with different roles—remain relatively stable across transitions , and that network size tends to decline with the gradual exit from roles in later life.

Subsequent research finds that closer ties indeed are more durable (Degenne and Lebeaux 2005; Morgan et al. 1997; Wellman et al. 1997). While Antonucci and Akiyama (1987) found no differences in network size by age in their sample of adults over age 50, a later study of a similar sample (Cornwell et al. 2008) did find networks to be smaller among older people; Fischer (1982), Marsden (1987) and Kalmijn (2012) report inverse associations between network size and age for adult samples spanning a broader age range.

Longitudinal studies sometimes examine several transitions . In their panel study of young adults, Bidart and Lavenu (2005) found a steady dropoff in contacts with nonkin during the transition to adulthood; relatively large networks during the “teenage sociability” period usually became smaller as subjects left school, entered the labor market, and formed steady partnerships. Mollenhorst, Volker and Flap (2014) linked divorce and the death of a spouse or partner to the dissolution of previously-existing relationships; widowhood was also associated with forming new relationships in new settings, however.

Other studies of changes in social networks over the life course focus on particular changes in life course states. An early study of adult men (Stueve and Gerson 1977) found marriage and parenthood to be associated with a shift in the sources of “best friends”; friendships drawn from work, neighborhood , and association settings tended to supplant those rooted in school and childhood experience. Wellman et al. (1997) likewise reported that entry into marriage was linked to substantial turnover in networks over a decade.

Hurlbert and Acock (1990) examined differences in confiding networks by marital status, finding that the networks of both married and widowed persons tend to be composed of higher proportions of family members than are those of single, divorced, or separated persons. Kalmijn (2003) examined the “dyadic withdrawal” hypothesis holding that marriage and cohabitation are associated with a general decline in network size and a concomitant rise in the interpenetration of the networks of spouses/partners. He found that the number of friends reported fell as subjects entered into partnerships and raised children ; as well, the dissolution of a partnership via divorce was linked to a decline in friendships. Part of the latter may reflect the tendency toward increased jointness in spousal networks over the course of a marriage, which rendered friendships within them vulnerable to disruption in the aftermath of divorce. Sarkisian and Gerstel (2016) reported that rates of socializing and of support exchanges with parents and siblings alike were highest among never-married people and lowest for the currently married, while those of ever-married personsFootnote 1 lay in between.

An early study on the transition to parenthood by Hammer, Gutwirth and Phillips (1982) found that it reconfigured social networks in several ways: by increasing the emphasis on kin rather than nonkin relationships, and by reducing the frequency of contact with those in the network. It reported no general difference in network size between parents and non-parents, though, and called attention to “child-linked contacts” as by-products of day care or participation in other child-related activities such as parent-teacher organizations , scouting, or athletic activities (see also Chua et al. 2016; Small 2009). McCannell (1988) studied a convenience sample of women over time, from mid-pregnancy to one year post-partum. She observed a notable decline in overall network size , but no reduction in the number of persons providing several specific types of social support . Moore (1990) reported that confiding networks were slightly smaller among adults with children at home.

Transitions in the domain of work and employment too have been examined. As noted, Bidart and Lavenu (2005) reported that nonkin contacts fell upon labor market entry. Fischer (1982) found that employed persons had more contacts with nonkin, but Moore (1990) reported no difference in confiding network size between full-time or part-time workers and other adults—including both the unemployed and those outside the labor market. In the study of older U.S. adults by Cornwell et al. (2008), confiding network size was higher among retired persons than those who remained active in the labor market.

Moen (2001) wrote about the gendering of the life course , observing that men and women of a given age may be differently situated within the life course, and further that the consequences flowing from a given life course position may differ for women and men. Examining associations between life course and social network phenomena separately for women and men is therefore common. Many such studies find fewer gender -related interactions than they anticipate; those differences that do appear often reflect a disparate influence of parenthood on the networks of women and men. Fischer and Oliker (1983) found that during early parenthood, women have fewer relationships with nonkin than do men; after children have left the home, however, women maintain more such relationships. Moore (1990) reported that employment was associated with fewer kin ties among women, but not among men. Munch, McPherson and Smith-Lovin (1997) focused on parenthood, finding confiding networks to be smallest among women—but not men—when the youngest child in a household was aged 3 or 4. Chua et al. (2016) found no overall difference in network size or span between women and men, but attributed differences in the types of contacts maintained to varying life course experiences.

This selective and undoubtedly incomplete survey of related research nonetheless provides ample warrant for anticipating that the size and composition of social networks varies across life course states, and moreover for expecting certain changes in social networks to follow adult life course transitions . In the next section, we suggest one route through which this connection arguably operates: life course transitions entail entry into and exit from social foci or settings, which in turn expand or limit opportunities for network formation.

Supply-Side Theories of Social Structure

Among the foundational axioms of Blau’s (1977) theory of social structure is that “rates of social association depend on opportunities for social contact” (Blau and Schwartz 1984: p. 29). It calls attention to the potentially available associates found in any arena in which social ties might be formed; I have elsewhere (Marsden 1990) termed it a supply-side theory of social structure, contrasting it with demand-side theories that stress individual choices and preferences. The number and composition of the others found in a social setting constrains the scale and types of the networks and relationships that can form within it. Several other opportunity-related theories of social networks , including Fischer’s (1977) choice/constraint theory and Verbrugge’s (1977) distinction between “meeting and mating,” draw similar contrasts.

We place special emphasis here on an opportunity-related concept developed by Feld (1981), termed a “focus of activity.” Feld (p. 1016) defines a focus as “a social, psychological, legal or physical entity around which joint activities are organized,” observing that a variety of settings—including persons, places, and groups, among others—may serve as foci (p. 1018). The foci of principal interest here are groups such as families or workplaces. Feld argues that foci serve to channel and organize social relations: “[t]hey may actively bring people together or passively constrain them to interact” (p. 1018), so that two persons jointly situated within a focus are more apt to interact with one another than are two who have no arena in common.

Foci can enable—or even require—the development or elaboration of social relationships . Feld observes that foci vary from one another (1981: p. 1019) in at least two crucial respects. First, they differ in size; people involved in a large focus are apt to encounter more meeting opportunities than are those in a smaller one. McPherson and Smith-Lovin (1982), for example, note that while men and women—on average—belong to the same number of voluntary organizations , men tend to be involved in larger groups linked with economic institutions, while otherwise-comparable women tend to participate in smaller groups concerned with domestic and community affairs (see also Chua 2013). They reason that these differences imply that the potential contacts men are prone to encounter in associations are more varied and valuable than those that women tend to meet.

Second, while all foci promote associations among those affiliated with them, the extent to which they do so differs considerably. At one extreme are “total institutions” (Goffman 1961)—such as prisons or nursing homes —that organize most or all aspects of a person’s life. At the other pole lie more discretionary settings, such as voluntary associations, in which participation is both less obligatory and much more episodic. The more time and attention that a focus requires of participants, the greater the constraint it poses on their opportunities to form and sustain social ties.

The focus concept is key for linking the life course and social networks because major life course transitions usually involve entry into new foci of activity, withdrawal from old ones, or changes in levels of commitment to foci in which one is already embedded. The formation of a romantic partnership creates a new focus—often an intense and demanding one—and secondarily links someone to the family and friendship networks of the partner or spouse; divorce or separation have the reverse consequences. The presence of children intensifies family commitments—particularly during childrearing—ordinarily entailing affiliations and obligations that extend throughout one’s life. Entering the labor force implies affiliation with one or more workplaces, and can draw someone into related organizations such as professional associations or trade unions; these constraints likely bear more strongly on those who have committed larger segments of their lives to work, e.g. full-time rather than part-time workers, or those who take on management responsibilities (Jonczyk et al. 2016). Events such as unemployment or retirement often sever such connections.

By the logic of Feld’s theory, life course events that alter the configuration of foci in which someone’s activities are situated will tend to expand or limit the pools and types of potential associates to whom one is exposed, and thereby the relationships that actually form. Variations in the size of the foci within which people participate imply variations in network size , while differences in the types of foci in which one is engaged, and the intensity of engagement within them, may shape network composition.

Our emphasis on shifts in foci as sources of network change has some resonance with Kahn and Antonucci’s (1980) arguments about the acquisition and shedding of roles as a source of network change. Roles certainly hold implications for relationships, and life course transitions surely involve entry into and exit from roles, or expansion/contraction in the degree of commitment to roles. Because Feld’s focus concept highlights variations in the size and composition of the sets of potential alters with whom one can associate while in different life course states, it has particular appeal as a device for understanding how life course transitions may prompt network change.

As noted, much prior work (e.g. Moen 2001; Munch et al. 1997; Kalmijn 2003) offers grounds for anticipating that the consequences of life course transitions may not be the same for men and women, given that women often assume the role of “kin keeper” (Moore 1990). In particular, the presence of (young) children may channel mothers and fathers into very different social worlds, which in turn could imply different changes in networks. Most analyses undertaken in this chapter therefore consider the possibility of gender differences.

Before presenting those analyses, I enter one disclaimer. The perspective adopted here posits that life course phenomena are explanatory variables that lead to differences and changes in network phenomena. The life course and social networks , however, likely have a bidirectional relationship, such that features of social networks might also prompt life course change. Lois (2016), for example, suggests that people with “family-centered” rather than “family-remote” networks are more apt to become parents ; Balbo and Barban (2014) find that childbearing by friends is associated with a greater likelihood of entering parenthood . Bernardi and Klärner (2014) present an overview of such research, suggesting that four network mechanisms (learning, pressure, contagion and support) underlie such effects. It is certainly possible that individuals, at least to some extent, alter their networks in anticipation of upcoming life course changes. It seems unlikely that all such consequences will be foreseen, however, or that individuals will be aware in advance of all of the new potential associates they later encounter in connection with a life course transition .

Data and Measures

The analyses that follow draw on data assembled by the General Social Survey (GSS) project, a continuing survey of U.S. adults stressing over-time replication of social indicators. The GSS began in 1972 and has been conducted every year or two since then. It uses a repeated cross-sectional design: each round draws a new sample of Americans aged 18 and over, and measures numerous sociodemographic variables—including important life course states—together with many behaviors and attitudes. The study regularly includes topical modules on subjects of current social science interest, including social networks . See Marsden and Smith (2012) for an overview of the GSS’s basic study design and content, and http://www.gss.norc.org/ for many more details and access to GSS data.

Analyses of network composition presented below draw on egocentric network data obtained in topical modules administered in 1985 and 1987.Footnote 2 Cross-sectional analyses of network activity use measures of socializing that appeared in the 28 GSSs conducted between 1974 and 2014; data from those studies are combined here. The network activity analyses also examine GSS panel data assembled between 2006 and 2014. Respondents to the 2006, 2008, and 2010 GSSs were subsequently reinterviewed 2 and 4 years later, yielding three-wave panels covering 2006–2010, 2008–2012, and 2010–2014; we pool these to examine linkages between life course transitions and changes in socializing.

Network Measures

The 1985 GSS obtained the first egocentric network data representative of a national population. It focused on “core” or “confiding” networks (Marsden 1987) consisting of those other persons (or “alters”) with whom a GSS respondent had spoken about “important matters” during the recent past. Those deemed to be part of a respondent’s network were elicited using the following “name generator” (Burt 1984) question:

From time to time, most people discuss important matters with other people. Looking back over the last six months, who are the people with whom you discussed matters important to you? Just tell me their first names or initials.

Subjects who initially provided less than five names were probed once for additional names. The 1987 GSS used the identical name-generating question, but probed only when fewer than three names were given at first. In both years, the number of names given ranged from 0 to 6 or more; as shown in panel B of Table 5.1, the mean network size was about 3 in 1985 (Marsden 1987) and modestly smaller (2.5) in 1987—likely due to the lower threshold used for probing.

Table 5.1 Descriptive statistics for network indicators and explanatory measures
Full size table

After naming alters, respondents were asked to describe each of them in several ways. Network composition was assessed by way of a question about the role relations that connect the respondent to an alter. We consider whether the subject deemed the alter to be any one of five types of kin , a neighbor, or a coworker.Footnote 3 Relatives comprised more than half of those cited in both 1985 and 1987 (Table 5.1, panel B); about a tenth were labeled as neighbors and a sixth as coworkers.

The analyses of network activity rely on this measure of informal socializing :

  • Would you use this card and tell me which which answer comes closest to how often you do the following things?

    1. A.

      Spend a social evening with relatives

    2. B.

      Spend a social evening with friends who live outside the neighborhood

    3. C.

      Spend a social evening with someone who lives in your neighborhood

    4. D.

      Go to a bar or tavern

Responses ranged from “never” (scored 1 here) to “almost every day” (scored 7). Socializing with relatives is most common: on average it occurs several times a month (panel C, Table 5.1). A typical respondent spends an evening with friends about once a month, and one with neighbors less often. Nearly half of respondents “never” visit a bar or tavern; an average respondent does so once to several times per year.

Life Course Measures

We focus here principally on transitions involving the family and the labor force. Within the family, we distinguish currently, ever-, and never-married persons, and also examine differences in networks associated with the presence of children in different age brackets (0–5, 6–12, and 13–17) in the respondent’s household. For labor force involvement, we compare respondents who are employed full-time, employed part-time, in retirement, and in some other work status (e.g. unemployment, keeping house, education ). Additionally, we consider the ways in which mobility across residential settings may be linked to network composition, contrasting persons who live in cities, in suburbs or unincorporated areas surrounding cities, and in towns or smaller areas. Descriptive statistics for these predictors are shown in panel A of Table 5.1.

Controls

All regression analyses control for four additional sociodemographic background measures—age , education , race , and sex. The alter-specific analyses of network composition also take account of the ordinal position in which the respondent named an alter; because closer alters tend to be named earlier (Burt 1986), this distinguishes alters by tie strength to some extent. As such, it contrasts those who are more central in a respondent’s network—and hence apt to remain stable (Kahn and Antonucci 1980)—and those with whom someone has weaker, more role-dependent relationships that are apt to turn over more rapidly. To assess conjectures that life course differences in networks are gendered, we examined interactions involving sex and the life course measures discussed.

Network Size and Life Course States

One of Alwin’s (2012) life course concepts stresses age -graded regularities in social phenomena, so we begin with a brief examination of cross-sectional differences in the size of confiding networks by age and life course states, relying on the 1985 and 1987 egocentric data. Table 5.2 presents data on average network size by selected explanatory variables.

Table 5.2 Mean confiding network size, by life course states and sociodemographic background
Full size table

In accord with prior findings (e.g. Fischer 1982; Marsden 1987; Cornwell et al. 2008), Table 5.2 indicates that networks tend to be smaller among older persons. In the 1985 data, average network size falls steadily from a mean of 3.4 among those under age 30 to one of just over 2 among those aged 70 and above. Age differences are somewhat more modest in the 1987 data.Footnote 4 Formerly married people have somewhat smaller networks than do the currently or (especially) never-married; employed respondents tend to have slightly more confidants than do retired persons (particularly) or those in other labor force statuses. For both the 1985 and 1987 data, the largest confiding networks are found among those living in suburban or exurban settings and the smallest ones among rural residents, while those of urban dwellers are close to the average size. No notable association of network size with the pr esence of children in the household is evident.

Among the sociodemographic variables considered here, differences by education appear largest. In the 1985 data, those completing 11 or fewer years of education cite just over two confidants on average, while those who attended at least one year of college name nearly four. Black respondents cite notably fewer alters than do those who are either white or nonwhite/nonblack. Table 5.2 indicates that men and women tend to have networks of roughly the same size, on average.

These life course and sociodemographic indicators are correlated with one another, of course. In regression analyses (not displayed) that enter all of them as predictors, education differences emerge as the largest and most consistent, together with the difference between black and nonblack respondents. In the 1985 data, age differences in network size remain statistically significant after adjusting for other explanatory variables; they do not in the 1987 data, however.

Cross-Sectional Differences in Network Composition by Life Course States

To examine differences in network composition, we asked whether persons in different life course states are more or less apt to cite alters drawn from three foci: the family, the residential neighborhood , and the workplace. Table 5.3 presents estimates for logistic regression analyses in which the life course and control variables predict binary indicators of whether a respondent described a given alter as a relative, neighbor, or coworker; alters are nested within respondents.

Table 5.3 Network composition and life course states (logistic regression coefficients)
Full size table

Opportunity-based theories anticipate that those whose lives are more deeply embedded within families will rely more on relatives as confidants . The findings for marital status are in keeping with this logic. Citation of relatives is substantially more likely among currently-married people; the odds that a given alter is described as a relative are more than three times higher among married than never-married respondents, in both 1985 and 1987. Formerly married (separated, widowed, or divorced) people are somewhat more apt to cite relatives than are those who have never married.

Table 5.3 does not, however, suggest that living in households having many children enhances the likelihood of citing family members as confiding contacts, with the exception of one significant coefficient (for 1987) indicating that those with more pre-teenage children tend to name relatives. Nor are relatives cited more often by those who have larger numbers of siblings.

Employment, however, is inversely linked to naming family members. The odds that an alter is a family member are more than 25% lower for the full-time employed than for those outside the labor force. Part-time employment also is negatively associated with naming family members, significantly so in the 1987 confiding data. These findings could well reflect competition among foci, as work-related activities come to consume more of someone’s time and energy.

Turning to the control variables, we see that women consistently name family members as confidants more often than men do, in keeping with Moore’s (1990) prior findings based on the 1985 data. Respondents with more education are less likely to name family members, while white respondents appear more likely to do so than nonwhites. In both years, earlier-cited alters tend to be relatives, suggesting that respondents tend to have closer, more multiplex relationships with kin .

Few of the factors considered are predictive of whether an alter is a neighbor. The only consistent finding here is that full-time employed people are less likely to cite neighbors; the odds that a full-time worker does so are about half as large as those for someone not in the labor force. In the 1987 data, neighbors tend to be among the later-cited (and hence less close) alters.

Both full- and part-time workers are much more likely to cite coworkers than non-employed people are; the odds of doing so grow by factors of between 3 and 5. Work consumes more attention from employed persons, of course, and they also have ready access to coworkers. Controlling for employment status, women are less likely to cite coworkers than men are, and blacks appear somewhat less apt than whites to do so. Later-cited alters are modestly more likely to be coworkers, as is to be expected for these role-dependent relationships.

In both years, substantial intraclass correlations (rho) are present for all three types of alters. These indicate that—after adjusting for all predictors considered—respondents vary in the extent to which they are embedded in particular foci of activity. If one alter is (or is not) a relative, others also tend to be (or not to be). Such clustering appears especially pronounced for citation of neighbors.

Gender Differences

To examine possible gender differences in how life course states and confiding are linked, interactions of sex with marital status, the numbers of children in the household, employment status, and residence were estimated. Few systematic sex differences in associations of life course states with network composition were evident. The most consistent patterns involved differences in the citation of coworkers: employment—both full- and part-time—is more strongly linked to naming work colleagues among women than among men (see Moore 1990). This suggests, perhaps, that the states of membership and non-membership in the labor force are more sharply differentiated experiences for women than for men. Less pronounced was a finding that having more teenagers in the household is more negatively associated with citing coworkers among women, perhaps indicating a gender difference in how work-family conflicts are experienced; it is the only statistically significant interaction involving children. Beyond these, the estimates for 1985 (but not those for 1987) suggest that employment and non-rural residence may be negatively linked to citing relatives among women, but not men. Overall, however, inspection of these interactions leaves an impression of similarity rather than difference in the way that life course states predict network composition for men and women.

Network Activity and the Life Course: Cross-Sectional Differences

We next examine differences across life course states in the frequency of informal socializing . Socializing may involve less intense network contacts than does than confiding about important matters, likely a mixture of stronger and weaker relationships. We begin by examining cross-sectional associations using the pooled 1974–2014 GSSs. For analyses focused on time trends in socializing over most of this period, see Marsden and Srivastava (2012).

Opportunity-oriented theories of network formation would anticipate that events marking family formation , including marriage and the arrival of children , would intensify one’s involvement in family-related foci and hence tend to increase the frequency of socializing with relatives while decreasing non-familial social activity. Some results of regression analyses presented in Table 5.4 are in accord with this, most notably the finding that married and formerly-married people socialize with relatives more often. Also of interest here is that those with more pre-school age children in the household tend to see more of their relatives. The reverse holds, however, for those having more children over 5 years of age; this is associated with small decreases in the frequency with which relatives are seen, perhaps reflecting engagement in school- and community-based activities involving children of these ages.

Table 5.4 Socializing levels and life course states (regression coefficients), 1974–2014 GSSs
Full size table

The negative associations between marriage and the presence of children with the other forms of socializing measured in the GSS (spending social evenings with friends and neighbors, and visiting bars and taverns) also align with expectations based on focus theory. Differences between currently- and never-married people are most notable here, but the presence of children of any age is inversely, though weakly, associated with non-familial socializing.

Differences in socializing related to labor force status are comparatively modest. The full-time employed are less apt to spend evenings with neighbors and more likely to visit bars and taverns than are those outside the labor force; socializing with friends is most common among part-time employees and retired persons. These findings likely reflect the availability of time together with involvement in activities that are either competitive or synergistic with workplace demands. Employment status is unrelated to socializing with relatives.

Differences by residential setting mirror those reported by Marsden and Srivastava (2012): respondents living in more urbanized places tend to socialize more with friends outside their neighborhoods than do rural dwellers, and are also more apt to visit bars and taverns, while those living in towns or rural areas are more likely to socialize with relatives and neighbors. These differences can be attributed to variations in both the availability of the different types of associates and the varying ecologies of these residential settings: urban settings make a diversity of potential friends readily accessible, thereby facilitating “chosen” relationships (Fischer 1982), while rural ones facilitate contact with neighbors.

Figure 5.1 displays standardized sheaf (or multiple-partial) regression coefficients (Heise 1972; Whitt 1986) that summarize the relative magnitudes of the differences in socializing in Table 5.4 that are related to family-related factors (both marital status and children in the household), employment status, and residential setting. Overall, family status appears to have the most pervasive associations with socializing, particularly with friends and in bars/taverns. In this sense, family may be the most constraining focus of the three. Relatively speaking, employment status appears most consequential for neighboring, while residence is most important for socializing with relatives and neighbors.

Fig. 5.1
figure 1

Sheaf coefficients summarizing strength of associations of family -related, employment, and residential factors with socializing, based on Table 5.4

Full size image

We remark briefly on some findings for the control variables in Table 5.4. Women are significantly more likely to socialize with relatives than are men, and less apt to be engaged in the other types of informal contact, again consistent with Moore’s (1990) image of “kin-keeping.” Socializing of all types grows less frequent with age .Footnote 5 Those with more years of education are less likely to socialize with relatives, but more likely to see those outside the family. The 40-year span covered by the data saw slight rises in familial socializing, and slight declines in the non-familial forms.

Gender Differences

To assess the prospect that socializing varies with life course states in different ways for men and women, we estimated interactions of gender with marital status, the presence of children, employment status, and residence. The large sample available for the cross-sectional socializing analyses allows detection of more such differences than for confiding. Most gender differences found had to do with family-related life course states, particularly marital status. Table 5.5 presents all pairs of conditional regression coefficients that differ significantly between men and women.

Table 5.5 Gender differences in associations between socializing levels and life course states (regression coefficients conditional on gender), 1974–2014 GSSs
Full size table

The way in which socializing varies with marital status differs by gender. Being married is more strongly linked to frequent contact with relatives among men than among women, though both see their relatives more than the never-married do. Formerly-married women, however, remain more likely than their never-married counterparts to socialize with relatives, while formerly married men do not. Both of these findings suggest that women maintain a more enduring connection to family than do men.

Outside of the family, the negative association between being currently married and seeing friends is somewhat stronger among men than among women. Similarly, the tendency for both currently- and ever-married people to socialize less with neighbors is significantly larger among men. Together with the difference in socializing with relatives already mentioned, these results indicate that marriage realigns the social lives of men more than it does those of women. Currently married men and women alike are much less apt to visit bars and taverns than are the never married (Table 5.4); formerly-married women remain unlikely to do so, however, while formerly-married men do not differ from otherwise comparable never-married ones.

A few gender differences in links between children and socializing were found, though many others proved to be insignificant. Those that were identified are consistent with the idea that children may impact the social lives of women more strongly than those of men. Women with teenage children in their household are somewhat less likely to see their relatives frequently, while no such difference is evident for men. The presence of very young children has no association with neighboring among men, but women in households with children under 6 tend to socialize less often with neighbors.

For the most part, socializing does not vary with employment status in notably different ways for men and women. An exception that resonates somewhat with the above findings about confiding has to do with seeing neighbors: both full- and part-time employment is more strongly linked to less neighboring among women than among comparable men. The difference in socializing with friends between women who live in suburban or exurban places and those in rural ones is also larger than that found among men.

These findings lend some credence to assertions that life course phenomena hold different implications for the social networks of women and men. As such arguments anticipate, the most consistent findings revolve around family-linked events, especially marriage. It is worth noting that most gender differences found here are of modest magnitude, and hence can be detected only with abundant data, but those differences that were isolated are of considerable substantive interest.

Transitions and Changes in Network Activity

To further probe the links between the life course and socializing, we turn to longitudinal analyses of the GSS panel data. Some respondents changed life course states (e.g., became married or entered retirement) during the 2 years that elapsed between successive panel interviews, so we can examine the associations between such transitions and changes in socializing that were reported. Table 5.6 presents estimates for fixed-effect models that predict socializing from life course states; these adjust for all time-constant respondent characteristics, observed or unobserved. The sample size for these analyses is much smaller than that in Table 5.4, so much less power to detect significant associations is available. Across occasions of measurement, respondents display relatively strong proclivities toward particular forms of socializing, as indicated by the rho values in Table 5.6.

Table 5.6 Fixed effect estimates for socializing levels (regression coefficients), GSS panel data, 2006–2014
Full size table

In general, regression coefficients in Table 5.6 are similar in sign but smaller in magnitude than the corresponding estimates from the cross-sectional analyses (Table 5.4). Entry into marriage is associated with significant decreases in the frequency of socializing with friends and visiting bars. It is not, however, linked to increased time with relatives. While almost all estimated associations between rises in the numbers of children in a household and changes in non-familial socializing are negative, they are small and insignificant for the most part. There are indications that having very young children slightly reduces neighboring, however, and that visits to bars decline a bit among those with additional children of elementary school age .

Several other findings regarding employment and residence also align with those in the cross-sectional analyses. Full-time employment is linked with fewer visits with neighbors and with an increase in socializing at bars, but not with spending more time with friends. Additional socializing with friends does, however, accompany entry into part-time employment. Residential moves from rural places into urban or suburban ones entail a rise in socializing with friends and a decline in seeing neighbors; the estimated coefficients for seeing relatives are also negative, but not statistically significant.

Table 5.6 contains one significant finding that did not emerge in the cross-sectional analyses: entry into the labor force—on either a full- or part-time basis—is linked to spending fewer social evenings with relatives, while no such association is evident in Table 5.4. One might conjecture that this divergence in findings reflects the short-term (2-year) changes captured by the panel data, and that labor force participants might adapt and restore their familial contacts to pre-entry levels after longer durations of employment. Alternately, reduced contact with family upon employment could be a phenomenon specific to the recent period (2006–2014) covered by the panel data, rather than the four decades spanned by the cross-sectional data.

Inspection of gender-related interactions yields only a few suggestive findings bearing on the prospect that transitions affect men and women differentially. The three significant differences found, however, take a similar form, indicating that a transition reduces socializing among women while making no difference among men. Specifically, relocating from a town or rural area to a suburb is accompanied by less socializing with relatives among women, but not among men; additional pre-school children lower socializing with friends among women only; and adding teenage children is associated with fewer visits to bars for women, but not men. None of these interactions emerged in the cross-sectional analyses, however (see Table 5.5).

The longitudinal analyses of network activity in Table 5.6 lend support to the main premise on which this chapter rests, that life course transitions can prompt changes in social networks . Though the panel and cross-sectional analyses are not consistent in all respects, the fixed-effect estimates offer stronger evidence that changes involving marriage , children, employment, and residence alter the rhythms of informal social lives.

Summary and Conclusion

This chapter argues that life course events shape social networks by both creating and eliminating opportunities for contact with others. Entering different life course states makes new types of people accessible, while limiting contact with other types of potential associates. Transitions in the life course can prompt changes in the composition of social networks , as well as in the frequency of different forms of network activity.

The General Social Survey data on social networks and informal socializing activity presented here offer considerable support for the perspective set forth. Family-related life course states, particularly marital status, shape contacts with relatives: currently married persons tend to confide in kin and spend more time socializing with them, and are correspondingly less involved with non-familial associates. Entry into the paid workforce expands the number of readily available coworkers, making it more likely that confiding networks will include workplace colleagues. Some competition across foci of activity is evident, as employed people appear somewhat less apt to draw their confidants from the family or the residential neighborhood , and likewise have fewer social contacts with neighbors.

In line with Moen’s (2001) observations about the gendering of the life course, the chapter investigated the possibility that life course states and social networks covary in different ways for men and women. The findings obtained are far from conclusive, and many of the differences detected are only modest in magnitude. They do convey several hints, however, that marriage and the presence of children may indeed hold different implications for the social lives of wives/mothers and husbands/fathers.

Notwithstanding our emphasis on life course transitions as events that shape opportunities for contact and thereby serve to constrain network formation, individual preferences and the human agency stressed by Elder (1994) surely affect the social networks that actually take form. Certainly individuals are able to exercise discretion within a structure of opportunities. Indeed, some may opt to enter a context or focus of activity because of the prospects it offers for network building (Burt 1992). Settings also may be chosen for other reasons, though, and not all network consequences of contextual choices are anticipated ex ante (Small 2009). Selection into a context or focus, then, serves to narrow—sometimes dramatically—the range of alternatives within which individuals may exercise agency.

The lines of analysis pursued here could be productively extended in several ways. More extensive network data, covering forms of social contact other than confiding and socializing (e.g. instrumental and informational assistance) could be informative. Longer-term longitudinal studies could enable a better parsing-out of the degree to which networks reflect the availability of contact opportunities versus preferences for associates of particular types. More recent data on confiding too would be helpful, especially in light of ongoing changes in the positions of women and men within families and the workforce .

As well, many questions can be posed about whether particular combinations of life course states—e.g. of marital, parental, and employment statuses—give rise to unique social network configurations. One might also ask, in line with Elder’s (1985) emphasis on historical time as a context in which life courses unfold, whether particular states or transitions are more strongly linked to network phenomena during different periods. Finally, pursuing the directions opened by Kalmijn’s (2003) study of the shared social networks of spouses and partners could yield insights into social networks as a means via which life courses are linked, and how transitions in a subject’s life course may ramify into the networks of his/her alters. The idea that life course states are associated with structured opportunities for contact, however, is likely to be one element to be taken into consideration while pursuing any of these interesting directions beyond this chapter’s line of inquiry.