Introduction

The importance of family functioning for individual (e.g., Fosco et al. 2012), couple (e.g., Kerig 1995), and child well-being (e.g., Dickstein et al. 2009) is well established. Romantic attachment is one factor that is thought to be associated with family functioning particularly, family cohesion and adaptability (Mikulincer and Florian 1999). It has been theorized that if adult attachment difficulties affect the marital relationship there will be implications for family functioning more generally (Mikulincer et al. 2002). No studies have considered the mediating role of marital satisfaction in the link between romantic attachment and family cohesion, family adaptability, and triangulation. This is unfortunate because investigating the mechanisms that connect romantic attachment and family functioning may inform the development of effective interventions with distressed families. Also little understood is how romantic attachment relates to triangulation of the child, a family process with significant implications for children’s psychological well-being (see Kerig and Swanson 2009 for a review).

Hazan and Shaver (1987) proposed that Bowlby’s (1982) attachment theory could be extended to romantic relationships, presenting the concept of “romantic attachment”. Romantic attachment has been defined as comprising two independent dimensions: anxiety (an individual’s worries about being abandoned and rejected) and avoidance (discomfort with closeness and dependency, reflecting a person striving to maintain emotional distance from partner; e.g., Brennan et al. 1998). Fraley and Waller (1998) suggested that continuous dimensions of attachment are more likely to explain variation in adult attachment than the traditional categorical measures defining attachment styles.

Olson et al. (1982) identified two central dimensions in family functioning: cohesion and adaptability. Family cohesion concerns the emotional bond between family members, while family adaptability (or flexibility) pertains to the capacity of the family to change its leadership, rules and roles in response to contextual and developmental demands. A strong body of empirical work documents the impact of family cohesion and adaptability both on family health and the well-being of its members (e.g., Olson and Gorall 2003), highlighting the need to keep on exploring variables that may enhance these aspects of family functioning. Research shows, for example, that family adaptability predicts change in maternal depression and child behaviour (Baker et al. 2011), and family cohesion improves adult social and emotional health (e.g., Fosco et al. 2012), as well as children’s socioemotional adjustment (e.g., Leidy et al. 2010). Moreover, although many factors have been identified to influence these dimensions of family functioning (e.g., marital satisfaction; Lindahl and Malik 2011), less is known about the association between spouses’ romantic attachment and family cohesion and adaptability.

Previous research has focused on how the family of origin may predict romantic attachment in dating adults (e.g., Conger et al. 2000), or links between adult attachment with respect to family of origin and interaction in the nuclear family (Paley et al. 2005). Only a handful of studies have examined the effects of spouses romantic attachment on family interaction in general, (e.g., Dickstein et al. 2004; Dickstein et al. 2009; Sibley and Liu 2006), and among these the specific association between romantic attachment and family cohesion and adaptability has been neglected. Mikulincer and Florian (1999) found that securely attached spouses reported high family cohesion and adaptability while avoidant spouses reported low levels of these family dimensions. An anxious attachment style was associated with high family cohesion and low adaptability. To our knowledge, since Mikulincer and Florian’s work, only one study has tried to replicate these findings (with couples where the husband had problems with drug abuse; Finzi-Dottan et al. 2003), but never with married and cohabiting couples in a community sample. This research is promising but limited, and the extent to which romantic attachment may contribute to our knowledge of family dynamics remains largely unexplored.

Another central dimension of family functioning is triangulation of the child. Triangulation involves a distortion of parent–child boundaries whereby parents attempt to create a coalition with their child that excludes or undermines the other parent (Minuchin, 1974). There are several reasons to consider triangulation as an important indicator of family functioning. First, Bowen (1978; Kerr and Bowen 1988) postulates triangles (a three-person emotional configuration) as the basic building block of the family, and the smallest stable relationship unit. The process of triangulation is central to Bowen’s theory, given that a two-person system is basically unstable, and when tension increases it will involve a third person to reduce anxiety (i.e., the dyad will triangulate a third person). Therefore, some triangulation is normal and even healthy within family interactions, as the involvement of a third party can support a dyad in overcoming impasses and coping through stressful times, turning the family unit to a stable emotional equilibrium (Bowen 1978). However, although Bowen did not proposed that triangulation was necessarily dysfunctional, he argued that it could become problematic when a third person involvement prevented the dyad’s members from solving their relationship issues (Kerr and Bowen 1988). Second, there is empirical evidence that family cohesion and triangulation are related to distinct family outcomes, highlighting the need to investigate family cohesion and adaptability along with triangulation. For example, triangulated family relationships are associated with lower levels of marital adjustment and higher levels of marital conflict, than cohesive families (Kerig 1995). Finally, triangulation damages marital and family functioning as it prevents couples from resolving their conflicts, drawing the child into the parental subsystem and increasing tension in family and marital relationships (Grych et al. 2004; Peris et al. 2008). Triangulation creates tension between the child and the excluded parent (Fosco and Grych 2010), enmeshment with the allied parent (Margolin et al. 2001) and has been associated with psychological adjustment problems in children (e.g., anxiety and depression; Buehler and Welsh 2009; Wang and Crane 2001; aggression; Fosco and Grych 2008). However, despite the importance of triangulation for understanding family dynamics, little is known about how romantic attachment relates to triangulation of children. This is unfortunate, considering the empirical evidence previously mentioned highlighting the unique impact of triangulation on family relationships and mental health (Kerig and Swanson 2009).

Unfortunately, much less is known about the psychological processes that may connect the relationship between romantic attachment and family cohesion, adaptability and triangulation. To our knowledge, no published studies have examined mediating effects between these variables. This represents a relevant gap in the literature if we consider that understanding the mechanisms involved in the romantic attachment–family functioning link may be important for several reasons. Firstly, it could help clinicians to identify and support distressed families. In addition, although there is evidence that attachment patterns can change (see Davila 2003 for a review), the exact factors that may change a person’s attachment style and promote attachment security (and, consequently, family functioning) are still not clear (Fraley et al. 2011). The stability of romantic attachment over time remains a controversial topic in the literature (Fraley et al. 2011), and thus it could be helpful to identify more proximal predictors of family dynamics that may mediate the link between romantic attachment and family functioning. Finally, concepts of adult attachment are being integrated in family and couple therapy models (Johnson 2008), but most of the empirically supported couple and family treatments still do not incorporate attachment framework to treat relationship problems (Davila 2003; Johnson 2008), which may hamper the efficacy of intervention efforts.

One potential explanatory mechanism that may mediate the relationship between spouses’ romantic attachment and family functioning is marital satisfaction, “an individual’s subjective evaluation of and personal sentiments toward the marriage” (Thompson 1988, p. 95). The rationale to investigate marital satisfaction as a mediator in the romantic attachment—family functioning link is both theoretical and empirical. Assuming that family relationships are interdependent and reciprocal, the family systems perspective suggests that the marital relationship affects the functioning of family as a whole (Cox and Paley 1997). Accordingly, marital satisfaction may provide spouses with the emotional resources necessary to create a nurturing family environment (Feldman et al. 1990). Also, the substantial conceptual overlap between the constructs of attachment security and marital satisfaction (Mikulincer et al. 2002) underscores the need to disentangle the differential impact that both romantic attachment and marital satisfaction may have on family dynamics, by testing mediating effects. Thus it is logical to think that marital satisfaction may be an important mediator between romantic attachment and family functioning. Moreover, there is empirical support to investigate marital satisfaction as an important explanatory mechanism relating romantic attachment and family functioning. Several studies document the association between romantic attachment and marital satisfaction, as well as between marital satisfaction and dimensions of family functioning. Attachment theory suggests that marital satisfaction depends on the extent to which spouses meet their needs for proximity, a safe haven, and a secure base (Mikulincer and Shaver 2007). Thus, marital satisfaction can be expected to increase as spouses become available and reliable sources of closeness and intimacy, effective providers of support and security (safe havens), and secure bases from which they participate in autonomous growth-oriented activities (Mikulincer et al. 2002). In other words, attachment security may work as a psychological resource that fosters marital satisfaction (Mikulincer and Shaver 2013). Accordingly, a large body of research confirms the link between romantic attachment (whether measured in terms of styles and dimensions) and marital satisfaction (for a review, see Mikulincer and Shaver 2007).

Research has consistently shown that spouses with a secure attachment (low levels of anxiety and avoidance) report higher levels of marital satisfaction, while an insecure attachment (avoidant or anxious styles) is associated with relationship dissatisfaction (e.g., Dickstein et al. 2009; Feeney 2002, 2008; Shaver et al. 2005). Though the majority of research has been cross-sectional (e.g., Banse 2004; Meyers and Landsberger 2002; Moller et al. 2006), longitudinal studies have also replicated these findings (e.g., Cobb et al. 2001; Crowell et al. 2002; Dickstein et al. 2009; Feeney 1994). For example, in a study examining two overlapping longitudinal samples of US couples with children over a period of 15 years, Hirschberger et al. (2009) found that attachment security in the couple relationship was related to both own and partners’ marital satisfaction. Furthermore, research using interview (e.g., Alexandrov et al. 2005; Treboux et al. 2004) and diary methods (e.g., Diamond et al. 2008) to evaluate attachment orientations, has demonstrated similar patterns of results.

Others have focused on assessing attachment dimensions and their distinct link with marital satisfaction. The majority of research shows that both anxiety about abandonment and avoidance are related to low levels of marital satisfaction, (e.g., Davila et al. 1999; Marchand 2004). However, results are discrepant regarding the importance of different attachment dimensions for relationship satisfaction. It has been suggested that attachment anxiety may play a greater role in marital relationships for own and partner satisfaction (e.g., Birnbaum 2007; Feeney 1994, 2002), while attachment avoidance may be a better predictor of relationship satisfaction in dating couples (e.g., Shaver et al. 2005). Yet, in a systematic review, Mikulincer and Shaver (2007) concluded that anxiety and avoidance equally predicted females’ marital satisfaction, whereas avoidance was more consistently related to males’ marital satisfaction. Also, some studies reported no differences between the strength of avoidance and anxiety as predictors of relationship satisfaction (e.g., Marchand 2004).

There is some evidence that men’s relationship satisfaction is more strongly predicted by females’ anxiety about abandonment than by females’ avoidance, whereas female relationship satisfaction appears to be more influenced by males’ avoidance than by males’ anxiety (Collins and Read 1990; Simpson 1990). These findings have been explained in relation to sex differences in the need for closeness and autonomy, consistent with sex-role stereotypes (Collins and Read 1990; Mikulincer and Shaver 2007). Still, some discrepancies can be found in the literature. There is evidence that only females’ avoidance predicts males’ perception of relationship quality (Davila et al. 1999; Shaver et al. 2005) while other studies indicate that males’ anxiety is the most consistent predictor of female dissatisfaction (e.g., Davila et al. 1998). In their review, Mikulincer and Shaver (2007) concluded that despite the inconsistency in gender differences, partner’s attachment anxiety seems particularly likely to decrease the other partner’s marital satisfaction. Further research that examines links between attachment security and partner marital satisfaction is warranted.

The association between marital quality and family functioning is also well documented in the literature, with empirical evidence showing that marital satisfaction is related to positive family functioning outcomes. More precisely, family cohesion seems to be positively associated with prenatal marital satisfaction (Lindahl et al. 1997), and couples with happy marriages show greater investment, support and warmth during family interactions than unhappy couples (e.g., Cowan et al. 1994; McHale 1997; Lindahl and Malik 2011). More recently, Froyen et al. (2013), found that higher marital satisfaction was related to higher levels of positive and lower levels of negative family emotional expressiveness. Conversely, research has also shown that lower levels of warmth among family members are present in distressed marital relationships (McHale 1995), and that marital dissatisfaction is related to lower levels of family cohesion (Henderson et al. 2003). Although most of the studies that have investigated the association between marital satisfaction and family functioning have been cross-sectional, longitudinal research has also replicated these findings (e.g., Yates et al. 1995). For example, Shek (1999), (2001) showed that marital satisfaction was related to both parents and children’s perceptions of family functioning, over time.

It has also been proposed that females and males’ marital satisfaction may affect family functioning in distinct ways. With parents of preadolescent boys, Feldman et al. (1990) found that females’ marital satisfaction, but not males’ marital satisfaction, was related to good family functioning. Furthermore, research also shows that marital satisfaction is related to triangulation. Kerig (1995) found that parents in triangulated families (where the child was triangulated) reported less marital adjustment than cohesive couples. Likewise, Wang and Crane (2001) showed that low levels of MS were related to triangulation, and Lindahl et al. (1997) found that marital dissatisfaction before parenthood was associated with child’s triangulation.

Despite an array of empirical work documenting links between romantic attachment, marital satisfaction, and family functioning, no study has investigated the mediating role of marital satisfaction in the romantic attachment–family functioning link. Paley et al. (2005) showed that marital interaction (but not marital satisfaction specifically) and spouses’ early attachment to parents predicted family functioning, but they did not test mediation effects. Another study has found that security in marital attachment significantly related to family-unit functioning for men and women, and to marital satisfaction, for women but not for men (Dickstein et al. 2001). However, more recent work from Dickstein et al. (2004) showed that marital attachment security was not associated with marital satisfaction or family functioning (Dickstein et al. 2004). More importantly, pre-adolescence is considered a critical period in the family life cycle. Encompassing dramatic developmental changes, the onset of puberty in offspring is regarded as the lowest point in couple’s marital satisfaction and a period of high risk for divorce (Gottman and Levenson 2000). Because attachment dimensions are more obvious in stressful circumstances, the association between romantic attachment, marital satisfaction, and family functioning should be more clearly displayed in this period. Also, marital satisfaction may be particularly important during times of family challenge, such as early adolescence.

This study aims to extend the work of Mikulincer and Florian (1999) by investigating the mediating role of males and females’ marital satisfaction, in the associations between romantic attachment dimensions, and family cohesion, family adaptability and triangulation. Drawing on attachment (Bowlby 1982) and family systems theory (Minuchin 1974), as well as past work documenting links between spouses’ attachment, marital satisfaction and family dynamics, and the importance of marital satisfaction for family relationships, we examined whether marital satisfaction mediated the relationship between males and females attachment anxiety and avoidance, and family functioning (including family cohesion and adaptability, and triangulation of the child (Fig. 1) (H1). Specifically, we hypothesized that adult anxiety and avoidance would be negatively associated with marital satisfaction (H1a). Based on findings by Mikulincer and Shaver (2007), we expected that anxiety and avoidance would both be associated with females’ marital satisfaction, and avoidance would be more consistently related to males’ marital satisfaction. On the other hand, we expected that males and females’ marital satisfaction would be positively associated with family cohesion and adaptability, and negatively related to triangulation of the child (H1b). Therefore, another contribution of our study is to help to clarify the explanatory mechanisms in the romantic attachment–family functioning link by adopting structural equation modelling to simultaneously assess the relative role of different theorised mediating pathways. Based on the systemic theory that conceives family processes as interdependent (Minuchin 1974), our study also aims to advance previous research by evaluating crossover effects between partners, as well as the unique contributions of men and women for family functioning. As previously mentioned, the literature has been inconsistent regarding gender differences and spouses’ crossover in the association between romantic attachment and marital satisfaction. Therefore, although using the couple as the unit of analysis would provide the exploration of linkages from the perspective of the dyad, the present study considers the interdependence and differences between spouses. This kind of analysis may reveal valuable information concerning the implications of each partner’s romantic attachment and marital satisfaction for family dynamics. Specifically, we hypothesized cross-spouse associations to occur between attachment avoidance and anxiety and self and partner’s marital satisfaction (H2). In particular, some differences were expected between males and females regarding these associations. Following work by Collins and Read (1990), it was hypothesized that males’ marital satisfaction would be more strongly related to females’ anxiety about the relationship (H2a), whereas females’ marital satisfaction would be more associated with males’ attachment avoidance (H2b). We were also interested in analysing gender differences in the hypothesised model. Consistent with past research showing that women’s marital satisfaction (but not men’s marital satisfaction) was related to overall family functioning (Feldman et al. 1990), it was hypothesised that women’s romantic attachment dimensions and marital satisfaction would have a stronger association with family functioning compared to men’s romantic attachment and marital satisfaction (H3).

Fig. 1
figure 1

Conceptual model showing relationships between females’ and males’ romantic attachment and family functioning

Full size image

Method

Participants

The data for this study were collected from a larger investigation examining aspects of family life among a panel of participating families living in Portugal. The larger study explored how variables of marital relationship (e.g., marital conflict, coparenting) were related to parent–child interaction (e.g., parenting practices). The original panel of participants consisted of 615 females and males. Family types other than two-parent families (because of the primary focus on marital satisfaction), non-biological parents, and families in which only one of the parents filled out the questionnaire, were excluded from the present study. The final sample comprised 519 married or cohabitating couples with at least one child between 9 and 13 years old (Mean age = 10.97 years old; SD = .92). The sample combined married or cohabitating couples as both had been in their relationship a similar length of time. We performed independent t tests to investigate whether relationship length (mean = 16.47 years; SD = 4.51) was related to relationship status. No significant differences were found between married and cohabiting couples [t(517) = 1.62, p > .05]. For the sake of clarity, we refer to spousal marital satisfaction in descriptions of the study constructs. The majority of families were from Lisbon (59.2 %) and Portugal West Coast (38.3 %), with 2.5 % from other areas of Portugal. The average age of females and males was 40.53 years (SD = 5.12) and 43.12 (SD = 6.05), respectively. Approximately 33.7 % of females had less than secondary education, 26.3 % completed secondary education and 40 % completed university. Of the males, 29.1 % had less than secondary education, 32.7 % completed secondary education, 38.2 % completed university. Most of the participating parents were Caucasian (98 %) and 2 % were Black.

Procedure

Participants were recruited from 14 schools. Ethical approval and permission to contact parents was obtained from each school board before the study began. After receiving permission from schools to conduct the study, parents were contacted by giving the fifth (mean age = 10.23 years) and sixth graders (mean age = 11.73 years old) letters describing the study and inviting families to participate. Parents provided written consent to participate in a study about the association between marital and parent–child relationships. Parents’ questionnaires were sent home with the child, in a sealed envelope. Separate questionnaires with instructions were included for males and females. Each spouse was asked to complete the questionnaires independently. Questionnaires contained measures relating to family demographics, marital satisfaction, romantic attachment and family functioning. A contact number for concerns or queries was provided. Prepaid envelopes were provided to return the questionnaires, although participants could also return the envelopes to their children’s teacher, if they preferred. Participants’ anonymity was guaranteed, and they were assured that no information would be used for anything other than research purposes.

Measures

Romantic Attachment

The Portuguese version of Experiences in Close Relationships Scale (ECR; Brennan et al. 1998) was used to evaluate romantic attachment. ECR is a 36-item self-report questionnaire that assesses romantic attachment dimensions: anxiety over abandonment and proximity avoidance. Participants were asked to answer based on their experiences in their current relationship on a 7-point scale ranging from (1) Disagree strongly to (7) Agree strongly. Eighteen items comprise the anxiety scale (e.g., “I worry about being abandoned”) and eighteen items constitute the avoidance scale (e.g., “I try to avoid getting too close to my partner”). Two scores were computed for each participant by averaging the respective items. Higher scores indicate higher levels of anxiety and avoidance. Internal consistency estimates were good for males and females, for both scales, respectively: avoidance (α = .91, α = .92) and anxiety (α = .84, α = .85).

Marital Satisfaction

Marital satisfaction was assessed using the Marital Life Areas Satisfaction Evaluation Scale (EASAVIC; Narciso and Costa 1996), a self-report 44-item Portuguese instrument that assesses marital satisfaction in areas of marital life (e.g., emotional intimacy). Items include: “The way my spouse expresses the way he/she feels about me”. Responses can range from Not satisfied (1) to Completely satisfied (6). Higher scores reflected greater marital satisfaction. Internal consistency estimates were excellent for males (α = .98) and females (α = .96).

Family Functioning: Cohesion and Adaptability

Family cohesion and adaptability were assessed using the Portuguese versions of the FACES-II (Olson et al. 1982), a 30-item measure comprising two subscales: family cohesion and family adaptability. Items are rated on a 5-point Likert scale, ranging from Almost Never (1) to Almost Always (5). The cohesion scale comprises 16 items (e.g., “Family members are supportive of each other at difficult times”) and the adaptability scale includes 16 items (e.g., “Each family member has input in major family decisions”). A linear scoring and interpretation procedure were used to compute a total cohesion score and a total adaptability score (Olson 2000). Higher scores on the subscales indicate more balanced ways of family cohesion and adaptability, whereas lower scores indicate unbalanced ways of cohesion and adaptability. Internal consistency was acceptable for males and females, for the cohesion (α = .80, α = .82) and adaptability subscales (α = .71, α = .71).

Family Functioning: Triangulation

Triangulation was measured with the triangulation subscale of the Portuguese version of the Coparenting Questionnaire (Questionário da Coparentalidade–QC, Pedro and Ribeiro 2015; Margolin et al. 2001). Females’ scores evaluated their perceptions about males’ triangulation behaviour and males’ scores assessed their perceptions about females’ triangulation behaviour. The Triangulation subscale comprises four items related to parents attempt to establish a coalition with the child that undermines the other parent, assessed on a five-point scale from (1) Never to (5) Always (e.g., “My spouse uses this child to get back at me”). Higher scores captured greater triangulation. The internal consistency for males (α = .75) and females (α = .81) and was acceptable.

For parsimony, and given the number of items that compose romantic attachment and marital satisfaction measures, we used parcels of the respective items as indicators of the latent variables of anxiety, avoidance, and marital satisfaction, following recommendations by Little et al. (2002).

Statistical Analyses

Analyses aimed to explore the mediating role of males’ and females’ marital satisfaction in the relationship between romantic attachment and three dimensions of family functioning: cohesion, adaptability, and triangulation. Analyses proceeded in five steps. First, the pattern of associations between variables was analysed and preliminary analyses were conducted to assess the fit of the measurement model. Following the guidelines proposed by Kline (2005), the fit of the models was assessed with the Chi square statistic, the comparative fit index (CFI), the root mean square error of approximation (RMSEA) and the standardized root mean square residual (SRMR). According to Hu and Bentler (1999) a CFI cutoff of .95 or higher, RMSEA values below .06, and SRMR values below .08 indicate a good model fit. A CFI cutoff of .90 indicates an acceptable fit (Hu and Bentler 1999), and values for the RMSEA between .06 and .08 suggest an acceptable fit (Browne and Cudeck 1993). Second, we tested the conceptual model (Fig. 1), associating spouses’ romantic attachment and family functioning, via each partner’s perception of marital satisfaction. Third, alternative models were tested to exclude the possibility of alternative hypotheses that may account for the relationships between the variables. Fourth, SEM analyses were conducted to test the indirect effects of a trimmed model with only significant paths (Fig. 2) using Structural Equation Modelling with maximum likelihood estimation, performed using Amos 19 software (Arbuckle and Wothke 1999). The Expectation Maximization (EM) algorithm was used to treat missing data. The amount of missing data across the study measures ranged from 1.5 to 13.9 %, which can be regarded as small to moderate (Widaman 2006). Following MacKinnon et al. (2004) recommendations, we used a bootstrap approach (Shrout and Bolger 2002) to test our mediation hypotheses. This procedure tests the significance of the indirect effect. We performed a nonparametric resampling method (bias-corrected bootstrap; Preacher and Hayes 2004) with 5,000 resamples drawn with replacement from the original sample to derive the 95 % confidence interval for the indirect effect.

Fig. 2
figure 2

Standardized coefficients for mediation model. MS1–MS3 = parcels from the EASAVIC, An1–An3 = parcels from the anxiety scale of ECR, Av1–Av3 = parcels from the avoidance scale of ECR, Tri = Triangulation, Co = Cohesion, Ad = Adaptability, M = mother’s report, F = Father’s report. For simplicity, only significant paths are presented

Full size image

Results

Descriptive statistics and correlations are presented in Table 1. The correlations were generally consistent with the theorized pattern of relationships. For example, attachment anxiety and avoidance were negatively associated with marital satisfaction. Conversely, marital satisfaction was positively associated with family cohesion and adaptability, and negatively associated with triangulation. Previous research suggests that relationship status and romantic attachment may be related (e.g., Adamczyk and Pilarska 2012; Kirkpatrick and Davis 1994). Thus, relationship status was included in the matrix of correlations. Given the small to non-significant correlations between relationship status and key study variables (including marital satisfaction), analyses proceeded without considering relationship status as part of model estimation.

Table 1 Correlations among spouses’ romantic attachment, marital satisfaction and family functioning
Full size table

Separate confirmatory factor analyses were performed to test the measurement model. Nine conceptual latent variables were specified and allowed to correlate (females’ avoidance, females’ anxiety, males’ avoidance, males’ anxiety, females’ marital satisfaction, males’ marital satisfaction, family adaptability, family cohesion and triangulation). Results showed an acceptable fit to the data [χ2 (210, N = 519) = 898.15, p < .001, CFI = .92, RMSEA = .08, SRMR = .06]. All manifest indicators had significant loadings on their latent variables (.56 to .95). These results were compared to an alternative measurement model in which all items loaded on a general factor. This model showed a very poor fit to the data. Full results are available upon request.

Prior to testing the full theoretical model, the direct effects between romantic attachment and family functioning were examined. Results showed direct effects between females’ avoidance and adaptability (β = −.23, p < .05, respectively), and between males’ avoidance and triangulation (β = .34, p < .01). We then tested our proposed model (Fig. 1). Results showed an acceptable fit of the model to the data: χ2 (225, N = 519) = 876.09, p < .001, CFI = .93, RMSEA = .07, SRMR = .06.

The results partially support our hypothesis regarding the associations between spouses’ romantic attachment and marital satisfaction (H1a). Females’ avoidance and anxiety were directly associated with females’ marital satisfaction. Only males’ avoidance was directly associated with males’ marital satisfaction. As expected, marital satisfaction was related to the three dimensions of family functioning (H1b), albeit just for females’ marital satisfaction. Against predictions, males’ marital satisfaction was not associated with family cohesion and adaptability but it was associated with the triangulation of the child. Moreover, our predictions regarding cross-spouse associations (H2) were partially supported: one inter-spouse effect was found between females’ attachment avoidance and males’ marital satisfaction. However, contrary to our hypotheses (H2a and H2b), males’ perception of marital satisfaction was related to females’ avoidance but not to females’ anxiety, and males’ attachment avoidance was not associated with females’ perception of marital satisfaction.

We then compared the fit of the proposed model with the fit of three alternative models: a nested model with only direct paths between attachment avoidance and anxiety and family functioning dimensions (alternative model 1); a model testing the mediating role of romantic attachment in the link between each spouse’s marital satisfaction and family functioning (alternative model 2); and a model that was the “reverse” of our model, where family functioning was related to attachment avoidance and anxiety through each partner’s marital satisfaction (alternative model 3). Results showed that the model we proposed provided a better fit to the data than alternative model 1 [χ2 = 1,451.78, df = 222; ∆χ2 (14, N = 519) = 600.13, p < .001], alternative model 2 [χ2 = 919.48, df = 209; ∆χ2 (1, N = 519) = 67.83, p < .001], and alternative model 3 [χ2 = 984.20, df = 211; ∆χ2 (1, N = 519) = 132.55, p < .001].

Indirect Effects Between Romantic Attachment and Family Functioning

We next tested the mediating role of males and females’ marital satisfaction in the link between his or her attachment avoidance and anxiety and family functioning. Although we hypothesised that males’ and females’ marital satisfaction would mediate the link between romantic attachment avoidance and anxiety and family functioning, no significant associations were found between males’ anxiety and males’ marital satisfaction, and between males’ marital satisfaction and family cohesion and adaptability. Figure 2 shows the final model testing for indirect effects.

In what concerns family cohesion and adaptability, indirect effects were found between females’ anxiety with abandonment, and cohesion and adaptability (β = −.05 and β = −.04, p < .01), as well as between females’ attachment avoidance, and cohesion and adaptability (β = −.28 and β = −.26, p < .001), through females’ marital satisfaction. Contrary to our hypotheses, no indirect effects were found via males’ marital satisfaction between spouses’ romantic attachment and family cohesion and adaptability. Regarding triangulation, and for females, indirect effects were found between anxiety and triangulation (β = .03, p < .001), via females’ marital satisfaction, as well as between avoidance and triangulation (β = .18, p < .001), via males and females’ marital satisfaction. For males, a single indirect effect was found between avoidance and triangulation (β = .13, p < .05), via males’ marital satisfaction. In order to identify the distinct contribution of males’ and females’ marital satisfaction in the link between females’ avoidance and triangulation, we followed up this analysis with the following procedure (Bollen 1989): to test the mediating role of males’ marital satisfaction, the path between females’ avoidance and females’ marital satisfaction was deleted and substituted by a correlation; to test the mediating role of females’ marital satisfaction, the path between females’ avoidance and males’ marital satisfaction was deleted and substituted by a correlation. Results showed that the indirect effect for females’ marital satisfaction (β = .20, p < .01) appeared to be stronger than the indirect effect for males’ MS (β = .06, p < .05).

Discussion

Our study extends previous research investigating the association between spouses’ romantic attachment and family functioning by examining the mediating role of males’ and females’ marital satisfaction. Support was found for the hypothesis that marital satisfaction mediates the link between spouses’ romantic attachment and family cohesion, family adaptability, and triangulation of the child. In addition, the proposed model showed a better fit than the alternative models, suggesting that low levels of spouses’ anxiety over abandonment and avoidance of intimacy are associated with higher levels of marital satisfaction, which in turn relates to better family cohesion and adaptability and lower levels of triangulation of the child. The findings are consistent with the view that romantic attachment facilitates warm, close, and harmonious interactions among family members (Olson 2000). In addition, romantic attachment may also make it less likely that children will become triangulated in the context of conflicted or discordant marital relationships. More generally, the findings support previous work showing that happy marriages are associated with better family functioning (e.g., Cowan et al. 1994; Lindahl et al. 1997; Lindahl and Malik 2011; McHale 1997).

More importantly, these findings mark a step forward in empirical work regarding the mechanisms that may explain the relationship between romantic attachment and family relationships. The present investigation extends previous work by Mikulincer and Florian (1999), showing that marital satisfaction mediates the link between romantic attachment dimensions and family cohesion, adaptability, and triangulation of the child. This represents an important contribution of this study for two main reasons. First, most approaches to couple and family therapy take little account of an attachment framework, though efforts have been made to incorporate adult attachment constructs in some family therapy models (for an example of these models, please see Davila 2003 and Johnson 2008). Second, the precise factors that may contribute to change adult attachment patterns are still not clear, and the stability of attachment patterns is yet a debated issue among attachment researchers (Fraley et al. 2011). Taken together, these two aspects may compromise family therapy practitioners’ efforts to intervene at the level of spouses’ romantic attachment to promote and support healthier family dynamics and prevent or decrease child triangulation. Our data suggest that the most effective place to intervene with respect to family dimensions like cohesion, adaptability or triangulation is at the level of marital satisfaction, given that it is related to healthy family relationships. More precisely, marital satisfaction seems to strengthen the emotional bond between family members (or how family systems balance separateness with togetherness), to enhance family flexibility to change and adapt to contextual and developmental demands (or how family systems balance stability versus change).

Our findings also support previous studies associating spouses’ attachment security (low levels of anxiety and avoidance) with marital satisfaction (e.g., Davila et al. 1998; Dickstein et al. 2009; Feeney 2002; Shaver et al. 2005). Consistent with Mikulincer and Shaver (2007) conclusions, females’ attachment anxiety and avoidance was associated with their marital satisfaction, whereas for males, only avoidance was associated with their level of marital satisfaction. This finding concurs with the view that avoidant men may be at greater risk for marital dissatisfaction than anxious men, possibly because women are particularly unhappy with avoidant partners, which may increase conflict and undermine avoidant males’ marital satisfaction (Mikulincer and Shaver 2007). Avoidance includes distancing oneself from and criticizing the spouse as well as keeping options open for other relationships (Shaver et al. 2005).

Results regarding inter-spouse effects were unexpected. Females’ avoidance was related to males’ marital satisfaction, contradicting previous research showing that females’ anxiety is a stronger predictor of male marital satisfaction than females’ avoidance (Collins and Read 1990; Simpson 1990). Couples with preadolescent children will tend to be older and marriages will tend to be longer, characteristics that may help to explain the pattern of results. Empirical evidence suggests that attachment anxiety may play a more critical role in early stages of couples’ relationship, when individuals are not certain about the future of the relationship (e.g., Eastwick and Finkel 2008), and that couples in long-term relationships, as well as middle age adults, report lower attachment anxiety (Feeney 1994; Stanley et al. 2010; Segal et al. 2009; Zhang and Labouvie-Vief 2004). Feeney (1994) suggested that a developmental process may be involved, whereby the security afforded by marriage allows spouses who were initially anxious about their marital relationships to revise their negative internal working models. In other words, once the commitment about the relationship becomes clear (as usually happens in long-term married relationships) attachment anxiety regarding the loss of the partner should dissipate (Feeney 1994; Stanley et al. 2010). Also, marriages with anxious partners are less likely to endure, given that anxiety is associated with marital conflict (Feeney 1994). Therefore, it is possible that the women in this study showed lower levels of anxiety, which attenuated the strength of the relationship between women’s anxiety and men’s satisfaction in this study. This is a direction for future research. Also, studies have shown that avoidance in attachment may be more important for marital satisfaction than attachment anxiety, in clinically distressed couples (Mondor et al. 2011). Although this study was conducted in a community sample of couples, we may speculate that, in challenging periods like early adolescence, attachment avoidance may also play a more central role for marital satisfaction than attachment anxiety. Low levels of intimacy avoidance may be needed to accomplish the levels of closeness and proximity required for couples to successfully achieve child rearing tasks. Namely, the characteristic deactivation of avoidant individuals’ attachment system includes distancing and detachment from one’s partner, which may prevent the resolution of problems related to preadolescent period, especially for men with an avoidant partner. Therefore, a unique aspect of this study involved examining the links between romantic attachment, marital satisfaction, and family functioning, in the context of family transition to adolescence, extending previous studies that tended to investigate the romantic attachment–family functioning link during the transition to parenting.

Although the link between romantic attachment and family functioning seem to be partially accounted for by marital satisfaction, females’ and males’ avoidance continue to be an independent predictor of family adaptability and triangulation of the child, respectively. These findings suggest that spouses’ attachment avoidance may play a more important role in family functioning than spouses’ attachment anxiety. Women’s tendency to distance themselves from significant others, along with cognitive closure regarding the incorporation of new information, appears to damage the capacity of the family system to be flexible and adjust to changes, over and above marital satisfaction. Females’ avoidance is inconsistent with gender role stereotypes (Feeney 1999), so we can speculate that when females’ have an inconsistent attachment, this hampers the family’s capacity to be adaptive.

Likewise, males’ attachment avoidance was associated with triangulation of the child. This finding concurs with evidence of a compensatory process, in which a stronger parent–child relationship is created to compensate for the lack of proximity in the couple’s relationship (e.g., Belsky et al. 1991; Brody et al. 1986). Also, this study sheds light on the differential impact that males’ and females’ romantic attachment may have on triangulation of the child, suggesting that males’ attachment avoidance may work as a risk factor for triangulation, over and above marital satisfaction. Based on attachment theory (Bowlby 1982), it can be postulated that an avoidant spouse, by tending to withdraw and distance from the others, may deprive the other spouse of support and affection. The distanced parent may turn to a child for their attachment needs in an effort to compensate for psychological needs not being met in the marital relationship. Males’ avoidance (but not females’ avoidance), may disrupt their ability to function as a secure base for their spouses. In turn, this may give rise to triangulation of the child or, in the extreme, interactions reflecting enmeshment of the mother–child relationship (Nichols and Schwarz 1998). Bowen’s theory also offers some direction when considering why males’ attachment avoidance was related to triangulation (Bowen 1978; Kerr and Bowen 1988) suggested that couples’ anxiety and difficulties with balancing intimacy and autonomy needs (as it probably happens in couples with avoidant husbands) produces marital tension and conflict. Accordingly, Bowen proposed that a primary mechanism for addressing this marital tension is to involve a child in the conflict so as to reduce or relocate personal anxiety and relational tension. Our findings are consistent with Bowen’s (1978) theorizing regarding the process of triangulation, suggesting that women with avoidant husbands may turn to children as a way of reducing the anxiety caused by fears of rejection and abandonment by the distant spouse, seeking increased togetherness with offspring to fulfill needs for proximity and closeness that are not being meet in the couple’s relationship. An alternative hypothesis is that women with avoidant spouses’ who triangulate their children are attempting to compensate their children for the males’ withdrawal (Belsky et al. 1991; Brody et al. 1986). These findings represent another novel contribution of this study, in that they advance knowledge regarding the associations between romantic attachment and triangulation of the child.

Finally, consistent with previous work showing that women’s marital satisfaction (but not men’s marital satisfaction) was related to overall family functioning (Feldman et al. 1990), our study indicated that females’ marital satisfaction was related to all three dimensions of family functioning, whereas males’ marital satisfaction was only associated with triangulation. Collectively, these findings highlight the central role of women’s marital satisfaction in family functioning.

Limitations and Implications

Although this study advances our understanding about the mediating role of marital satisfaction in the associations between spouses’ romantic attachment and family cohesion, adaptability, and triangulation, some caveats must be noted. First, the data is cross-sectional, which makes it impossible to draw conclusions about the direction of effects. Despite the fact that alternative models were tested and these had a poorer fit than the proposed model, the possibility of bidirectional effects must be acknowledged. Given that romantic attachment classifications are current assessments of an individual’s perspective and construction of intimate relationship experiences, it is equally plausible to consider that attachment may either impact or be impacted by quality of family functioning. Therefore, although this study can help to highlight links between the constructs it was not designed to determine causal processes, and thus more research using a prospective, longitudinal design is required. Second, our sample comprised families with children aged 9–13 years old so caution should be taken when generalizing these findings to families in other stages of the family life cycle. Third, the reliance on mailed questionnaires is another limitation of the present study because we were unable to control the circumstances in which participants completed the questionnaires.

Notwithstanding these limitations, this study has implications for practitioners. When working with inflexible and disconnected families and with avoidant partners, our findings underscore the need to improve marital satisfaction, in order to create more close and flexible family relationships. However, implications may vary according to spouses’ gender. For avoidant men, interventions should aim to increase men’s availability to the spouse. For example, clinicians should help couples to develop dyadic coping strategies so that spouses can support one another more efficiently (e.g., Bodenmann 1995, 2005). In turn, this would enable parents to reduce their reliance on children to meet their attachment needs. Also, clinicians could aim to increase awareness among couples of the vicious cycle formed by enmeshed mother–disengaged father patterns of interaction, reframing it in the context of attachment avoidance. For avoidant women, therapists should attempt to support women to limit the spillover of marital dissatisfaction to family cohesion and adaptability. For example, intervention should aim to improve coping skills that are thought to be related to marital satisfaction (e.g., Ptacek and Dodge 1995) and that may help to avoid the transference of negative emotions from the marital relationship to family functioning (e.g., positive self-verbalization; reframing of the situation).

To conclude, the major contribution of this study is the examination of the mediating role of marital satisfaction in the associations between spouses’ romantic attachment and family functioning. Evidence was provided to suggest that females’ attachment dimensions and marital satisfaction play a more central role in explaining family cohesion and adaptability, while males’ attachment avoidance seems to be more relevant to the presence of triangulation dynamics within the family.