This is an open access article distributed under the terms of the Creative Commons
Attribution License (Families are a source of social capital, as they involve connections among individuals who often provide instrumental, informational, and emotional support to each other. The importance of family relationships as social capital has therefore been stressed (Coleman, 1988; Furstenberg, 2005; Furstenberg & Hughes, 1995; Furstenberg & Kaplan, 2003), with attention being paid to various outcomes such as the degree of trust in personal networks (Ravanera & Rajulton, 2010), the likelihood of out-migration of families (Palloni, Massey, Ceballos, Espinosa, & Spittel, 2001), the social integration of families in communities and neighborhoods (Brisson & Usher, 2005; Freistadt & Strohschein, 2013; Hughes & Stone, 2006), marital satisfaction, divorce proneness of couples and their ability to invest in coparenting (Doan, Favez, & Widmer, 2013; Widmer, Kellerhals, & Levy, 2004), as well as school attainment of children (Hofferth, Boisjoly, & Duncan, 1998).
There have been doubts, however, about the ability of individuals in stepfamilies to benefit from social capital because of the ambivalences associated with various stepfamily ties, such as those between children and their non-residential parent or between children and their residential parent's partner (Hoffmann, 2002; McLanahan & Sandefur, 1994). However, a deficit of social capital may not concern large numbers of individuals in stepfamilies, as the presence of such a deficit may depend on the ways in which family boundaries are set up after divorce and remarriage (Cherlin & Furstenberg, 1994; Stewart, 2005). This contribution considers family boundaries set up by individuals in stepfamilies, in comparison with family boundaries set up by individuals in first-time families, and their consequences for the production and availability of social capital in families. It stresses that family boundaries constitute a decisive factor for social capital and suggests that bridging social capital may in many instances compensate for the weakness of bonding social capital in such family contexts.
Bonding and Bridging Social Capital
Social capital includes the relational resources stemming from the possession of a social network of mutual acquaintance or recognition (Bourdieu, 1986). Two kinds of social capital have been investigated in a variety of contexts beyond families (Patulny & Svendsen, 2007; Pettit & Collins, 2011; Szreter & Woolcock, 2004). Bonding social capital refers to networks with a high density of relationships between members, where most, if not all, individuals belonging to the network are interconnected because they know each other and interact frequently with each other. This situation enhances expectations, claims, obligations, and trust among individuals because of the increase of the collective nature of normative control and social support (Coleman, 1988). If any network member fails to conform to the others’ expectations, s/he is likely to have several other network members jointly react against the situation. In the family context, collective support of siblings to an elderly parent or intergenerational supervision of children refers to such a case. Bonding social capital is illustrated in Figure 1A which features a stepfamily configuration in which almost everyone exchanged emotional support with everyone.
Bridging social capital is an alternative to bonding social capital based on brokerage opportunities that some individuals develop in networks (Burt, 1995, 2002; Granovetter, 1973). The absence of some connections creates relational holes in the network that provide some persons, known as brokers, with opportunities to mediate the flow of information among members and, therefore, control and influence others. In Figure 1B picturing another stepfamily, the respondent and her daughter played a brokerage role to connect the current partner and a female friend to the rest of the network members. Such persons take advantage of being intermediaries between other individuals, otherwise not directly connected to each other, to develop more autonomy and agency in various contexts (Davidsson & Honig, 2003; Szreter & Woolcock, 2004). In the family context, children being able to draw separate pocket money or gifts from their divorced parents, as well as benefiting from a larger autonomy in their everyday life, or women mediating the contacts and the exchanges of emotional support between their partner and their own parents and children refer to such a case.
Figure 1
Exchanges of emotional support in two stepfamilies.
1A. (left): Nuclear configuration. 1B. (right): Post-divorce configuration.
The kinds of social capital, either bridging or bonding, may vary between first-time and stepfamilies. One important issue for family social capital relates to remarriage after divorce (Amato, 2010; Sweeney, 2010). Stepfamilies are often supposed to experience a lack of social capital, associated with a reduced investment in time and energy in children (Coleman, Ganong, & Fine, 2000). Non-residential parents as well as residential parents in a new partnership tend to reduce their involvement with their children because they devote their time and energy to new partners (Bray & Berger, 1993; Claessens, 2007; Hetherington & Stanley-Hagan, 2000). Stepparents also invest less social capital in stepchildren, as they may prioritize their couple and their own children from prior relationships (Bogenschneider, 1997). Correspondingly, children do not necessarily develop strong ties with the new partners of their parents (Bumpass, Sweet, & Castro Martin, 1990; Ganong & Coleman, 2004). Ties with former in-laws and blood relatives (Peletz, 1995) are also disrupted by divorce and it takes time to replace or reconstruct them (Kalmijn & Broese van Groenou, 2005; Terhell, Broese van Groenou, & van Tilburg, 2004). In many cases, family recomposition makes family ties more diffuse and more ambivalent in their lines of responsibility for providing support (Harknett & Knab, 2007; Shriner, Mullis, & Schlee, 2009; Wen, 2008). Therefore, some authors suggested using family structure as a proxy for social capital (Ravanera & Rajulton, 2010). This proxy should be critically examined, as it underestimates the specificities and strengths as well as the diversity of stepfamily contexts (Coleman et al., 2000). It also masks the fact that mixed results were found in regard to the predicted negative outcomes of stepfamilies (Coleman et al., 2000; Hetherington & Stanley-Hagan, 2000; Sweeney, 2010). In addition, individuals belonging to stepfamilies may learn advanced relationship and communication skills (Golish, 2003; Higginbotham, Skogrand, & Torres, 2009), actively and flexibly negotiate their role, and develop great autonomy. This series of empirical evidence suggests that individuals in stepfamilies are more prone to develop bridging social capital than bonding social capital, as the absence or weakness of some ties makes the likelihood of disconnection between some family members higher and, therefore, the central position of brokers more likely.
Family Boundaries
Individuals in both first-time and stepfamilies develop their own family boundaries, by including some alters and excluding others in their definition of their family contexts (Widmer, 2010). The definition of their family boundaries by individuals has to be taken into account when dealing with the production of social capital by families. In recent decades, family contexts have become more individualized and pluralized (Beck & Beck-Gernsheim, 1995; Giddens, 1992), making the boundaries of contemporary families more ambiguous (Allan, Crow, & Hawker, 2011; Carroll, Olson, & Buckmiller, 2007; Castrén, 2008; Cherlin, 2004; Cherlin & Furstenberg, 1994; Pill, 1990; Stewart, 2005; Visher, Visher, & Pasley, 2003). Family recomposition after divorce does not necessarily lead to the new extended family model where a large number of individuals with various statuses consider that they constitute a single family unit (Cherlin & Furstenberg, 1994).
Family boundaries more likely vary from case to case in stepfamilies. Indeed, after a divorce and a remarriage, some individuals maintain strong relationships with their previous partner, whereas others invest in their new partner and stepchildren. Still, other people focus on their own parents and children. Similarly, not all available relatives are necessarily involved in meaningful relationships and considered as significant family members (Cherlin & Furstenberg, 1994). Remarriage may create interdependencies and solidarity among a large number of individuals far beyond the household. Solidarity toward relatives is, however, not automatically achieved in stepfamilies (Allan et al., 2011). For instance, the likelihood that step-grandparents and step-grandchildren develop a positive emotional attachment increases when step-grandparents are younger than 70 and step-grandchildren younger than 10 when the relationship starts (Christensen & Smith, 2002). Non-residential fathers may or may not disengage from active parenting, thus being at risk of not being considered as significant family members by their children (Cherlin & Furstenberg, 1994; Hetherington & Stanley-Hagan, 2000). The status of stepparent is even more ambiguous and offers individuals considerable latitude when negotiating their roles (Cherlin, 2004; Graham, 2010; Schrodt, 2011). The definition of appropriate stepparent roles varies among individuals and ranges from substitute parents to friends or even to outsiders (Church, 1999; Ganong, Coleman, & Jamison, 2011; Mahoney, 2006; Marsiglio, 1992). Children in stepfamilies unequally acknowledge their stepfather as a family member (Cherlin & Furstenberg, 1994). Stepparents are more likely to include stepchildren in their personal networks suggesting that stepfamily boundaries have become more permeable over time (Suanet, van der Pas, & van Tilburg, 2013). Therefore, individuals in stepfamilies are expected to set up their family boundaries in a variety of ways.
The diversity of family boundaries is not unique to stepfamilies. It is also relevant to first-time families (Cherlin, 2009; Widmer, 2010). Scholars have stressed the uncertainty associated with the significance of many family members in adulthood, including siblings (White, 2001), aunts and uncles (Milardo, 2010), parents and grandparents (Mueller, Wilhelm, & Elder, 2002; Silverstein & Marenco, 2001) and even spouses (Cherlin, 2009). In some cases, pseudo-kinship ties, such as friends considered as family members, play a significant role (Allan, 2008). Overall, the diversity of significant family members and family boundaries also concerns individuals in first-time families.
Summary and Hypotheses
The diversity of family boundaries is likely to have consequences for social capital. In stepfamilies, some individuals are expected to develop configurations in which their family members are all interconnected, while others may develop configurations with some family members who do not support each other or interact with each other. Therefore, social capital may not only depend on being in a first-time family versus being in a stepfamily, but also on how individuals develop and define their family boundaries irrespective of family structures. As friends, step-relatives and ex-partners, in-laws and other relatives refer to distinct and often disconnected parts of the family, it is expected that their joint presence in a family configuration leads to bridging social capital, whereas a focus on the current partner, children and parents is expected to be associated with bonding social capital, as those individuals are more often interconnected. Therefore, depending on the ways in which family boundaries are developed by individuals in both first-time and stepfamily structures, we expect individuals to develop distinct kinds of social capital, either bonding or bridging.
Three hypotheses will be tested in this study. (1) While first-time families develop bonding social capital more often than stepfamilies, stepfamilies develop bridging social capital more often than first-time families. (2) Boundaries of family configurations vary within stepfamilies and within first-time families. In both stepfamilies and first-time families, we expect to find a diversity of family configurations unequally based on siblings, parents, in-laws, previous partners and friends (considered as family members). (3) This diversity of family configurations has consequences for the social capital made available to individuals. Family configurations primarily based on close blood relatives (parents and children) and conjugality (partners) are expected to produce bonding social capital, whereas family configurations primarily based on in-laws, step-relatives or friends are expected to produce bridging social capital. This is expected to be true both in stepfamilies and first-time families.
Methods and Measurements
Study Design
The respondents belong to a non-proportional representative random sample of 300 women drawn from the population of females living in the canton of Geneva (Switzerland), with at least one child of their own aged 5 to 13 years (referred to as the target child) and living with a partner (cohabiting or married). When respondents had more than one child, basic information on them was collected, but the main focus remained on this selected target child. In one half of the cases (150 women in stepfamilies), the target child was the child of the respondent, but not of her co-residential partner. The respondent or her partner might have had other children, either together or with another partner, living at home or elsewhere. In the other half of the cases (150 women in first-time families), the target child was the child of both the respondent and her co-residential partner. The respondent or her partner had no other children from previous relationships. The two halves of the sample were matched with regard to the age and the sex of the target child, and the educational level of the respondent, to allow for a comparison of stepfamilies and first-time families with similar characteristics. Respondents were located using a random procedure based on a list of all households comprising children in the canton of Geneva, a mostly urban area of Switzerland, and were interviewed between the spring of 2009 and the winter of 2010. They were first recruited through a survey institute and, if they agreed to an interview, they were then contacted by the research team. No incentive was paid, but the importance of doing academic research on stepfamilies was stressed. The response rate was 65%. Face-to-face interviews of about one hour and a half length were conducted by the research team. Interviews were done in different settings depending on the respondents' preferences (mostly in their home or at the university). The absence of other family members during the interview was required in order to ensure confidentiality.
Participants
The three hundred female respondents ranged in age from 29 to 55 with a median age of 40. Fourteen percent of the respondents had a foreign nationality (other than Swiss) and 32% were born outside Switzerland. Regarding the education level, 35% of the respondents had a university education, 30% a low vocational training, 20% a high vocational training, 12% an upper secondary education, and 3% a lower secondary education. Note that Swiss respondents with a university education were slightly overrepresented. The mean number of children was 2.24; 2.35 for women in first-time families and 2.14 for women in stepfamilies.
Independent Variable - Measure of Family Boundaries
A free-listing technique was used to delineate the family configuration of the respondents (Levin, 1993; Widmer, 2010). Respondents were asked to provide a list of all individuals whom they considered to be significant family members at the time of the interview. They were free to identify as many family members as they wanted. The term “family” was deliberately left undefined to approach their personal definitions of family. Respondents were instructed that the term “significant” referred to people who have played a role in their life, either positive or negative, during the past year. Based on this list of significant family members, a typology of nine family configurations was created (see the detailed procedure in Results).
Dependent Variables - Measures of Social Capital
Perceived emotional support has often been used as a measure of social capital (Lochner, Kawachi, & Kennedy, 1999). It refers to the ability to provide guidance and moral comfort. It was investigated with the following question whose validity and reliability were confirmed by previous testings (Widmer, 2010): “Who would give emotional support to X [i.e., each previously listed individual included in the respondent’s family configuration, considered one by one] during routine or minor troubles?” In accordance with previous research, respondents not only had to evaluate their own family relationships, but also the relationships among all family members. Following previous validation studies (Widmer, Aeby, & Sapin, 2013), social network measures were applied to investigate social capital in families (Carrington, Scott, & Wasserman, 2005; Hanneman & Riddle, 2005; Scott, 2000; Wasserman & Faust, 1994).
Three overlapping sets of family members were considered. The first set was defined as the respondent’s full family network because it included all individuals who were considered to be family members. The second set, technically known as the respondent’s in-neighborhood, included only family members to whom the respondent provided support. The third set was the respondent’s out-neighborhood, which included only family members from whom the respondent received support.
As an illustration, Figure 1 showed two family configurations stemming from a stepfamily structure. The arrows were the ties of emotional support exchanged among the configuration members. Graphically, a respondent's in-neighborhood was represented by arrows pointing toward the respondent, while, in a respondent’s out-neighborhood, arrows pointed away from the respondent toward the support providers.
Using validated indices capturing social capital in ego-network research (Carrington et al., 2005; Hanneman & Riddle, 2005; Scott, 2000; Wasserman & Faust, 1994), bonding social capital was measured by the density of emotional support. Density is the number of existing connections divided by the number of pairs of family members cited by the respondent. Since the configuration of Figure 1A had 18 connections out of the 30 potential ones, it had a density of 0.60. In contrast the configuration of Figure 1B had 20 connections out of 72 potential ones and, thus, a density of 0.28. High density indicates tight interconnections between the configuration members and, therefore, bonding social capital. Density was computed for the three sets of family members.
Bridging social capital was measured by the betweenness centralization of networks and the betweenness centrality of respondents. Betweenness centralization indicates whether a small number of individuals laid between all other members’ chains of connections. It was computed for the respondent’s full family network. The configuration of Figure 1A had a betweenness centralization of 0.18, whereas the configuration of Figure 1B had a betweenness centralization of 0.47. Closely related to the betweenness centralization, betweenness centrality captures the proportion of connections involving a specific individual, here the respondents. In the respondent’s in-neighborhood and out-neighborhood, the configuration of Figure 1A had respectively a betweenness centrality of 0.09 and 0. The configuration of Figure 1B had a betweenness centrality of 0.60 in the respondent’s in-neighborhood and a betweenness centrality of 0.67 in the respondent’s out-neighborhood. High centralization or centrality indicates a high probability of being an intermediary between other configuration members and, therefore, bridging social capital. All network analysis indices vary from 0 (low) to 1 (high), as they were standardized by the network size.
Control Variables
We included a series of variables known to relate with family relationships and social capital in order to control for potential confounding variables when testing the impact of family structures and family configurations on social capital. Two indicators related to the target child were used as control variables: sex and age. Research on stepfamilies has shown that girls have more problems adjusting to the entry of a stepparent than boys (Hetherington & Stanley-Hagan, 2000), suggesting that social capital may be lower in their case. Age has been found to be correlated with more adjustment problems, adolescence being a particularly difficult period (Hetherington & Stanley-Hagan, 2000). Indeed, in single parent families, some adolescents attain early autonomy and are used to have a role in family decision making, role which is being jeopardized by the stepparent (McLanahan & Sandefur, 1994). Therefore, we chose the age of 12 years old as a threshold, because it corresponds to the beginning of the transition to adolescence institutionally marked by the end of primary school and the entrance in lower secondary school. Three other indicators related to the respondents were selected as control variables: age, education level (dichotomized in either low or high), and participation in paid work. The age of the respondent is associated with demographic constraints (Puur, Sakkeus, Põldma, & Herm, 2011). Older respondents are less likely to have their parents still alive. Finally, the two last variables refer to cultural and economic resources creating inequalities among families (Coleman, 1988). Education is generally found to have either no association or a negative association with remarriage (Sweeney, 2010). Mothers' education was found to be negatively associated with benefits that non-residential fathers perceive in the parent role (Seltzer & Brandreth, 1994). Full-time employment is associated with higher quality of the martial relationship in stepfamilies, as it enhances equality between partners (Rogers, 1996).
Results
Family Structures
We computed the mean scores of all indices for each family structure, as well as the F-Test, Kruskal-Wallis test (χ2), and d of Cohen. Indicators of bonding social capital scored higher for first-time families than for stepfamilies. The density of emotional support received was significantly higher in first-time families (Mean first-time families = 0.74; Mean stepfamilies = 0.68; F(1, 297) = 7.15, p ≤ .01; χ2(1, N = 300) = 7.21, p ≤ .01; d = 0.31). In contrast, the indicators of bridging social capital scored higher for stepfamilies than for first-time families. Betweenness centralization of the full family network was significantly higher in stepfamilies than in first-time families (Mean first-time families = 0.21; Mean stepfamilies = 0.25; F(1, 298) = 6.18, p ≤ .01; χ2(1, N = 300) = 4.48, p ≤ .05; d = 0.29). Regarding emotional support given, the betweenness centrality of the respondents was also significantly higher in stepfamilies (Mean first-time families = 0.22; Mean stepfamilies = 0.29; F(1, 297) = 9.79, p ≤ .01; χ2(1, N = 300) = 7.69, p ≤ .01; d = 0.36). First-time families and stepfamilies did indeed provide distinct kinds of social capital.
Family Configurations
Regarding their significant family members, respondents mentioned a total of 154 different family terms. Almost all respondents (98%) included their children as significant family members (Table 1). Current partners were also prominent, as they were cited by 96% of respondents. Mothers of respondents were the second most frequently cited, as 76% referred to their mothers as significant family members. Fathers and siblings of respondents were also well represented. Blood ties were included first in the list of significant family members, right after partners. In-laws came after blood relatives. In 37% of cases, mothers-in-law were cited, while, in 22% of cases, fathers-in-law were cited. In addition to blood and in-law connections, female friends were commonly cited as family members (29%). Overall, females represented a larger share of family members than males did. Former partners (i.e., fathers of the target child) were mentioned by 21% of respondents. Given that only half of the sample could actually mention them, this equated to 42% for respondents in a stepfamily structure.
Table 1
Distribution of the 23 Most Commonly Cited Family Terms
| Terms | Number of respondents citing the term (n) | Percentage of the 300 respondents citing the term (%) | Percentage of the term cited (in reference to the total 2942) (%) | Cumulative percentage (%) |
|---|---|---|---|---|
| 1. Partner | 287 | 96 | 10 | 10 |
| 2. Mother | 229 | 76 | 8 | 18 |
| 3. Son | 225 | 75 | 12 | 29 |
| 4. Daughter | 211 | 70 | 11 | 40 |
| 5. Father | 168 | 56 | 6 | 46 |
| 6. Sister | 121 | 40 | 5 | 51 |
| 7. Brother | 115 | 38 | 5 | 56 |
| 8. Partner's mother | 110 | 37 | 4 | 59 |
| 9. Female friend | 86 | 29 | 6 | 65 |
| 10. Partner's father | 66 | 22 | 2 | 67 |
| 11. Former partner | 63 | 21 | 2 | 69 |
| 12. Partner's sister | 51 | 17 | 2 | 72 |
| 13. Partner's daughter | 37 | 12 | 2 | 73 |
| 14. Brother's partner | 35 | 12 | 1 | 74 |
| 15. Partner's son | 33 | 11 | 1 | 76 |
| 16. Sister's son | 31 | 10 | 2 | 77 |
| 17. Sister's partner | 29 | 10 | 1 | 78 |
| 18. Sister's daughter | 28 | 9 | 1 | 80 |
| 19. Partner's brother | 26 | 9 | 1 | 81 |
| 20. Male friend | 24 | 8 | 1 | 82 |
| 21. Former partner's mother | 21 | 7 | 1 | 83 |
| 22. Brother's daughter | 19 | 6 | 1 | 84 |
| 23. Brother's son | 18 | 6 | 1 | 84 |
Following standard factor and clustering procedures as applied to textual data (Tabachnick & Fidell, 1996), subsequent analyses focused only on the 23 terms cited by at least 5% of the respondents, the other terms being gathered into a residual category. These 23 terms included 84% of the 2942 citations made overall. In order to uncover the significant family configurations, factor analysis with a varimax rotation was run on the 23 terms and the residual category to extract the initial factors. Following standard practice in factor analysis (Tabachnick & Fidell, 1996), 10 factors with eigenvalues greater than 0.65 were retained which explained 53% of the variance.
Table 2
Cluster of Family Terms (Average Number of Citations for Each Terma, by Cluster)
| Friend | In-law | Brother | Sister | Kinship | Beanpole | Nuclear | Without partner | Post-divorce | Total | F(8, 291) | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Partner | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 0.00 | 1.00 | 0.96 | - |
| Mother | 0.72 | 0.84 | 0.87 | 0.71 | 0.88 | 0.88 | 0.26 | 0.85 | 0.84 | 0.76 | 8.82** |
| Children | 2.08 | 2.28 | 2.26 | 2.11 | 2.12 | 2.34 | 2.41 | 1.31 | 2.23 | 2.20 | 2.21* |
| Father | 0.39 | 0.56 | 0.61 | 0.54 | 0.72 | 0.75 | 0.09 | 0.77 | 0.61 | 0.56 | 7.38** |
| Sister | 0.58 | 0.28 | 0.22 | 1.75 | 0.44 | 0.52 | 0.00 | 0.62 | 0.43 | 0.52 | 19.94** |
| Brother | 0.36 | 0.47 | 1.13 | 0.57 | 0.52 | 0.42 | 0.00 | 0.46 | 0.43 | 0.45 | 6.87** |
| Mother's partner | 0.19 | 0.91 | 0.39 | 0.29 | 0.72 | 0.28 | 0.15 | 0.00 | 0.36 | 0.37 | 12.31** |
| Female friend | 2.78 | 0.25 | 0.83 | 0.14 | 0.40 | 0.20 | 0.00 | 0.31 | 0.23 | 0.56 | 42.89** |
| Father's partner | 0.03 | 1.00 | 0.26 | 0.21 | 0.44 | 0.02 | 0.03 | 0.00 | 0.18 | 0.22 | 39.19** |
| Former partner | 0.11 | 0.00 | 0.04 | 0.04 | 0.28 | 0.00 | 0.00 | 0.46 | 1.00 | 0.21 | 95.32** |
| Partner's sister | 0.17 | 0.47 | 0.22 | 0.32 | 0.48 | 0.11 | 0.03 | 0.00 | 0.20 | 0.21 | 3.16** |
| Partner's children | 0.06 | 0.28 | 0.13 | 0.43 | 0.16 | 0.28 | 0.15 | 0.00 | 0.66 | 0.27 | 2.79** |
| Brother's partner | 0.03 | 0.00 | 0.96 | 0.11 | 0.12 | 0.00 | 0.00 | 0.08 | 0.14 | 0.12 | 43.43** |
| Sister's son | 0.14 | 0.00 | 0.04 | 0.96 | 0.16 | 0.17 | 0.00 | 0.23 | 0.05 | 0.18 | 8.66** |
| Sister's partner | 0.00 | 0.03 | 0.13 | 0.50 | 0.16 | 0.06 | 0.00 | 0.15 | 0.07 | 0.10 | 7.81** |
| Sister's daughter | 0.06 | 0.03 | 0.04 | 0.68 | 0.32 | 0.11 | 0.00 | 0.23 | 0.05 | 0.14 | 6.70** |
| Partner's brother | 0.03 | 0.00 | 0.00 | 0.00 | 1.24 | 0.00 | 0.00 | 0.00 | 0.00 | 0.11 | 230.49** |
| Male friend | 0.64 | 0.03 | 0.17 | 0.00 | 0.00 | 0.08 | 0.00 | 0.08 | 0.00 | 0.11 | 9.09** |
| Mother's former partner | 0.03 | 0.00 | 0.00 | 0.04 | 0.16 | 0.00 | 0.03 | 0.00 | 0.32 | 0.07 | 8.55** |
| Brother's daughter | 0.00 | 0.00 | 0.61 | 0.00 | 0.20 | 0.02 | 0.00 | 0.08 | 0.07 | 0.08 | 10.97** |
| Brother's son | 0.00 | 0.00 | 0.52 | 0.11 | 0.16 | 0.02 | 0.00 | 0.08 | 0.02 | 0.07 | 9.19** |
| Others | 2.11 | 1.84 | 4.30 | 3.18 | 4.08 | 1.78 | 1.68 | 3.23 | 2.66 | 2.52 | 5.55** |
aThe number of citations can vary between 0 and 1 for terms such as partner, mother and father, and between 0 to 15 for other terms such as children, sisters, brothers, etc.
*p ≤ .05. **p ≤ .01.
Following exploratory multivariate statistical methods (Lebart, Morineau, & Piron, 2002), the 10 factors' scores were inputted in a hierarchical clustering analysis based on a measure of the Euclidean distance between individuals and on the Ward clustering algorithm. Solutions from 2 to 12 clusters were examined and a 9-cluster solution was chosen because of its balance between interpretability and statistical efficiency related with the sample size (Everitt, Landau, Leese, & Stahl, 2011). Table 2 presents the average number of citations for each term by cluster. These nine clusters were: Friend, In-law, Brother, Sister, Kinship, Beanpole, Nuclear, Without partner and Post-divorce family configurations.
Friend configurations focused on individuals who were considered to be family members while being related neither by blood nor by marriage or partnership. Respondents in this cluster included as many as 2.78 female friends and 0.64 male friends in their family configurations. Post-hoc comparisons using the Tukey HSD test indicated that the mean score of 2.78 for female friends in Friend configurations was significantly larger than in all other configurations. In-law configurations had a strong orientation toward the partner and the in-laws. Partners and the mothers of the partners were overrepresented, as well as other in-law relationships. Brother and Sister configurations included the respondent's siblings and their children and current partners. Post-hoc comparisons indicated that the mean score of 1.75 for sisters in Sister configurations and the mean score of 1.13 for brothers in Brother configurations were significantly larger than in all other family configurations. Kinship configurations included a variety of individuals related by blood and marriage, such as partners, parents, children, uncles, aunts, nieces, nephews, cousins, and grandparents. Beanpole configurations referred to families in which several generations co-exist, with only a few family members in each of them (Bengtson, Rosenthal, & Burton, 1996). They focused on blood relatives, with the inclusion of members of various generations, particularly grandparents from the mother’s and the father’s sides. They were vertically, rather than horizontally oriented, contrary to the Brother and Sister configurations. Nuclear configurations were almost exclusively centered on the partner and the children. They corresponded to a definition of the family as a co-residential unit. Without partner and Post-divorce configurations were only found in the stepfamily structure. Respondents in Without partner configurations did not include the present partner as a significant family member, although he lived, as in all other cases, within the same household as the respondent and her child. In contrast to Without partner configurations, Post-divorce configurations had two orientations: one toward the former partner and his relatives and the other toward the new partner and his relatives (including, when applicable, his children and, in some cases, his own ex-partner). Post-hoc comparisons indicated that the mean score of 1 for the former partner in Post-divorce configurations was significantly different from all other configurations.
Table 3 shows the distribution of family configurations across first-time families and stepfamilies. In first-time families, Beanpole configurations came first (25.3%), followed by Friend (15.3%), Nuclear (14.7%) and In-law (13.3%) configurations. In stepfamilies, Post-divorce configurations came first (29.3%), followed by Beanpole (18%), Sister (10%) and simultaneously Friend (8.7%) and Without partner (8.7%) configurations. Brother configurations (2.7%) were rare in contrast to Sister configurations (10%) in stepfamilies, whereas both types of configurations were well represented in first-time families. Note that overall, sixty-two percent of respondents in stepfamilies shared configurations similar to those found in first-time families.
Table 3
Distribution of Family Configurations According to Family Structures (%)
| First-time families
|
Stepfamilies
|
Total
|
|||
|---|---|---|---|---|---|
| Configurations | n | % | n | % | N |
| Friend | 23 | 15.3 | 13 | 8.7 | 36 |
| In-law | 20 | 13.3 | 12 | 8.0 | 32 |
| Brother | 19 | 12.7 | 4 | 2.7 | 23 |
| Sister | 13 | 8.7 | 15 | 10.0 | 28 |
| Kinship | 15 | 10.0 | 10 | 6.7 | 25 |
| Beanpole | 38 | 25.3 | 27 | 18.0 | 65 |
| Nuclear | 22 | 14.7 | 12 | 8.0 | 34 |
| Without partner | 0 | 0.0 | 13 | 8.7 | 13 |
| Post-divorce | 0 | 0.0 | 44 | 29.3 | 44 |
| Total | 150 | 100.0 | 150 | 100.0 | 300 |
Note. χ2(8, N = 300) = 77.51, p ≤ .001, Cramer's V = 0.51.
Social Capital in Family Configurations
Table 4 presents the means calculated for each family configuration, as well as the results of the F-Test and Kruskal-Wallis test (χ2), with their levels of significance and the proportion of variance explained (R2).
Table 4
Indicators of Bonding and Bridging Social Capital in Family Configurations - Mean by Family Configuration, F-test, Kruskal-Wallis (χ2), and R2 Tests
| Network for emotional support | Mean by family configuration
|
M | F | χ2 | R2 | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Friend | In-law | Brother | Sister | Kinship | Beanpole | Nuclear | Without partner | Post-divorce | |||||
| Emotional support given | |||||||||||||
| Density (bonding) | 0.40 | 0.49 | 0.46 | 0.42 | 0.41 | 0.47 | 0.59 | 0.49 | 0.44 | 0.46 | 4.11**a | 25.89** | .012 |
| Betweenness centrality (bridging) | 0.31 | 0.19 | 0.24 | 0.25 | 0.23 | 0.26 | 0.19 | 0.29 | 0.29 | 0.25 | 1.71†a | 17.84* | .000 |
| Emotional support received | |||||||||||||
| Density (bonding)b | 0.63 | 0.74 | 0.66 | 0.68 | 0.71 | 0.72 | 0.88 | 0.70 | 0.67 | 0.71 | 5.03**a | 34.57** | .012 |
| Betweenness centrality (bridging) | 0.59 | 0.50 | 0.45 | 0.54 | 0.44 | 0.58 | 0.38 | 0.39 | 0.59 | 0.52 | 1.49b | 11.59 | .000 |
| Full family network for emotional support | |||||||||||||
| Density (bonding) | 0.38 | 0.37 | 0.38 | 0.31 | 0.31 | 0.42 | 0.55 | 0.40 | 0.35 | 0.39 | 5.47**c | 32.21** | .008 |
| Betweenness centralization (bridging) | 0.28 | 0.20 | 0.23 | 0.21 | 0.21 | 0.25 | 0.19 | 0.21 | 0.25 | 0.23 | 1.39c | 13.53† | .001 |
Note. For χ2, df = 8, N = 300.
aF(8, 290). bF(8, 254). cF(8, 291).
†p ≤ .1. *p ≤ .05. **p ≤ .01.
Indices measuring social capital were significantly associated with family configurations. Nuclear configurations featured a high density of interconnections, a well as a low centrality of respondents and a low centralization of networks. Therefore, their social capital is oriented toward bonding. In-law and Without partner configurations also had high scores of density. To the opposite, Post-divorce and Friend configurations featured a low density of interconnections, and a high level of centrality and centralization. Therefore, their social capital is oriented toward bridging. Brother, Sister and Kinship configurations had average to low scores of either density or centrality and centralization, with no clear evidence about the dominance of either bonding or bridging social capital. Interestingly, Beanpole configurations featured a high centrality and centralization and an average to high density.
Table 5 includes a set of regressions which simultaneously estimated the effects of family structures (stepfamily or first-time family) and of the types of family configurations, while controlling for possible effects of age and sex of the children, as well as age, education and participation in paid work of respondents.
Table 5
Indicators of Bonding and Bridging Social Capital – Multivariate Regression Analysis
| Predictors | Emotional support given
|
Emotional support received
|
Full family network of emotional support
|
|||
|---|---|---|---|---|---|---|
| Density (bonding) | Centrality (bridging) | Density (bonding) | Centrality (bridging) | Density (bonding) | Centralization (bridging) | |
| Family Structures | ||||||
| Stepfamilies | -0.040* | 0.079*** | -0.062** | 0.088* | -0.042* | 0.064*** |
| Family Configurationsa | ||||||
| Friend | -0.067** | 0.073** | -0.088*** | 0.108* | -0.006 | 0.065*** |
| In-law | 0.014 | -0.049 | 0.015 | 0.018 | -0.030 | -0.015 |
| Brother | -0.014 | 0.013 | -0.068* | -0.026 | -0.014 | 0.022 |
| Sister | -0.034 | -0.015 | -0.025 | 0.029 | -0.060* | -0.028 |
| Kinship | -0.065** | -0.009 | 0.002 | -0.045 | -0.083** | -0.009 |
| Beanpole | 0.004 | 0.027 | -0.005 | 0.104** | 0.020 | 0.038** |
| Nuclear | 0.126*** | -0.042 | 0.156*** | -0.092 | 0.165*** | -0.021 |
| Without Partner | 0.049 | -0.014 | 0.032 | -0.176* | 0.034 | -0.053 |
| Post-divorce | -0.014 | 0.016 | -0.019 | 0.079 | -0.027 | 0.001 |
| Target child | ||||||
| ≥ 12 | 0.011 | -0.040 | -0.016 | -0.041 | 0.033 | -0.024 |
| Boy | -0.010 | 0.034 | -0.028 | 0.045 | 0.004 | 0.036** |
| Respondent | ||||||
| Age | -0.003 | 0.003 | -0.001 | 0.003 | -0.006** | 0.004** |
| High education | 0.004 | -0.008 | -0.014 | 0.030 | 0.004 | -0.013 |
| Paid job > 80% (reference) | ||||||
| Paid job 50-80% | -0.037 | 0.030 | -0.078*** | 0.135** | -0.051* | 0.023 |
| Paid job < 50% | -0.024 | 0.045 | -0.070** | 0.147** | -0.039 | 0.025 |
| Housewife or other | -0.078** | -0.041 | 0.007 | 0.061 | -0.110*** | -0.027 |
| R2 | .133 | .107 | .189 | .088 | .183 | .107 |
aWe used a deviation contrast method to deal with the various effects of the configurations. This made it possible to estimate the effect of each category of a covariate in comparison to its overall effect.
*p ≤ .05. **p ≤ .01.
Results show that first-time families triggered a high density of interconnections and a low centrality and centralization, representative of bonding social capital, whereas stepfamilies were associated with a low density of interconnections and a high centrality and centralization, which are representative of bridging social capital. Family configurations also had a significant effect, once family structures and control variables were taken into account. Overall, Nuclear configurations were significantly associated with high density. Without Partner configurations had low centrality for emotional support received. In contrast, Friend configurations featured low density and high centrality and centralization. Kinship, Brother, and Sister configurations were also associated with low density. Interestingly, Post-divorce configurations did not trigger centrality and centralization when the effect of family structures was statistically controlled for. Also, Beanpole configurations had an average density, but a high centrality of emotional support received.
Some of the control variables had a significant impact of their own. Considering the full family network, family configurations tended to be associated with bridging social when the child was a boy, and the age of the respondents was negatively correlated with the density of emotional support and positively associated with the centralization. Finally, respondents who did not have a paid job and respondents who were working less than 80% had configurations with a lower density of emotional support than those working more than 80%. In addition, respondents working less than 80% had a higher centrality of emotional support received than those working more than 80%.
Discussion
Whereas first-time families more often produce bonding social capital, stepfamilies more often develop bridging social capital, a resource helping individuals to develop autonomy and agency in a variety of social situations and settings (DiPrete, Gelman, McCormick, Teitler, & Zheng, 2011; Wellman & Hogan, 2006). Therefore, the hypothesis of a deficit of social capital in stepfamilies is rejected, as many individuals in stepfamilies replace bonding social capital with bridging social capital. Nevertheless, the availability of bridging social capital in stepfamilies may not mean the absence of adjustment problems; having a brokerage position costs energy and time and may prove stressful.
As predicted, in both first-time and stepfamilies, a variety of family configurations with distinct boundaries are present. Some individuals in stepfamilies maintain ties associated with their previous partnership within their family boundaries and add new ones, while others favor new family members and disrupt their older ties. Some individuals develop family boundaries which exclude their ex-partner and include their current partner and their cohabiting children. Some others focus on their children and exclude their current partner from their family configuration. Furthermore, numerous individuals have family configurations which focus on other significant family members such as grandparents, siblings, and friends. Overall, divorce and remarriage trigger a variety of configurations of significant family members. Some stem from the same stock of family members that is available to first-time families, whereas some others directly relate to the specific demographic features of stepfamilies. Consequently, only a minority of family recompositions after divorce lead to a configuration that corresponds to the new extended family model in which ex-partners, current partners, and children from a variety of partnerships are considered as significant family members and parts of one large family system. The development of family boundaries is likely to change from one person to the next depending on events and transitions having occurred in their life trajectories (Antonucci, Fiori, Birditt, & Jackey, 2010; Widmer, 2010).
The hypothesis of an effect of family boundaries on social capital is also accepted. Strategies concerning the development of family boundaries have implications for the social capital that is made available to individuals (Widmer, 2006). Bridging social capital is produced by family configurations that are primarily based on in-laws, step-relatives or friends, whereas bonding social capital is produced by family configurations that are primarily based on close blood relatives, conjugality and, by extension, the household. Indeed, individuals who include step-relatives or friends in their family configurations benefit from two separate family parts which they bridge while getting a larger structural autonomy in their family realm and an early exposure to weakly connected social contexts (Burt, 2002; Granovetter, 1973). The exception regarding the impact of blood ties was found in Beanpole configurations which produce bridging social capital, contrary to expectations regarding the impact of intergenerational relationships for bonding social capital (Coleman, 1988).
Finally, there are some limitations to this study that should be noted. First, the delineation of family boundaries was based on the interview of only one family member. It is likely that children in stepfamilies develop other boundaries than their mothers, as they may often consider their non-residential fathers as members of their family configurations even if their mothers no longer do. Fathers may also develop distinct family boundaries than their ex-partners. Therefore, a useful extension of this study would be to include several respondents reporting on the same family structure. Second, we collected cross-sectional data and, therefore, we could not study the development of family boundaries and social capital across time. Third, we used emotional support as the sole indicator of social capital. It is necessary to consider other dimensions of social capital in addition to emotional support, for instance exchanges of material resources. Finally, the distinction between bonding and bridging social capitals should be further investigated in relation to a wider array of family outcomes. Therefore, in order to overcome those shortcomings, future research should include multi-actor longitudinal designs encompassing various dimensions of social capital.
Despite these limitations, the findings of this study point out the importance of bridging social capital produced by stepfamilies and the prominence of collective dimensions of family contexts far beyond parent-child dyads. Such results may help practitioners to understand how individuals in stepfamilies set up family boundaries and produce social capital. This information may be crucial to detect potential personal problems of adults and children related with their family contexts. Indeed, divorce and remarriage require a complex reorganization of all family relationships. The focus of the family configuration may remain on the previous partnership, switch to the new one, or balance both. When new ties prevail over older ties in the family, children may have difficulties to maintain a sound relationship with their non-residential parent. Similarly, when the opposite happens, children may have difficulties to create a meaningful relationship with their residential stepparent. In addition, the potential key role of other family members, such as a grandparent or a friend considered as family, may be more easily detected and used as a mediator if needed. Furthermore, each way of setting up family boundaries has consequences for social capital. When family configurations produce bonding social capital, children are more closely watched out for, but they may have difficulties developing as autonomous individuals. In contrast, when family configurations produce bridging social capital, children may be allowed more space within family arrangements, but they may lack concerted guidance. More generally, paying attention to the collective dimensions of family-based social capital by using a social network approach of family relationships may bring a better understanding of coparenting, child development, and other key outcomes for stepfamilies and first-time families.