Differentiation of Self and Dyadic Adjustment in Couple Relationships: A Dyadic Analysis Using the Actor-Partner Interdependence Model

Abstract

Bowen’s multigenerational theory provides an account of how the internalization of experiences within the family of origin promotes development of the ability to maintain a distinct self whilst also making intimate connections with others. Differentiated people can maintain their I-position in intimate relationships. They can remain calm in conflictual relationships, resolve relational problems effectively, and reach compromises. Fusion with others, emotional cut-off, and emotional reactivity instead are common reactions to relational stress in undifferentiated people. Emotional reactivity is the tendency to react to stressors with irrational and intense emotional arousal. Fusion with others is an excessive emotional involvement in significant relationships, whilst emotional cut-off is the tendency to manage relationship anxiety through physical and emotional distance. This study is based on Bowen’s theory, starting from the assumption that dyadic adjustment can be affected both by a member’s differentiation of self (actor effect) and by his or her partner’s differentiation of self (partner effect). We used the Actor-Partner Interdependence Model to study the relationship between differentiation of self and dyadic adjustment in a convenience sample of 137 heterosexual Italian couples (nonindependent, dyadic data). The couples completed the Differentiation of Self Inventory and the Dyadic Adjustment Scale. Men’s dyadic adjustment depended only on their personal I-position, whereas women’s dyadic adjustment was affected by their personal I-position and emotional cut-off as well as by their partner’s I-position and emotional cut-off. The empirical and clinical implications of the results are discussed.

INTRODUCTION

Bowen’s multigenerational theory (1978) emphasizes the importance of autonomy and interdependence processes to individual and family development. It assumes that patterns of social interaction and adjustment are transmitted across generations—in other words, that children’s social, emotional, and cognitive development is influenced by their parents’ relationship pattern—and that this enables individuals to develop an age-appropriate degree of autonomy and differentiation of self and to create functional relationships with others (Bowen, 1978; Kerr & Bowen, 1988).

Differentiation is central to Bowen’s account of systemic and individual developmental trajectories, and it is a fundamental property of family relations and the organization of self. The family differentiation construct describes emotional, relational, and behavioral patterns of distance regulation within families. These patterns reflect the degree to which the family system encourages intimacy between members and individuality (Cohen, Vasey, & Gavazzi, 2003; Kerr & Bowen, 1988) and offer a way to conceptualize family functioning. Optimal family differentiation combines tolerance or expectation of high levels of both individuality and intimacy, whereas low family differentiation is linked with low tolerance or expectation of individuality and intimacy.

Family differentiation is linked to a variety of individual adjustment variables and has a big impact on personal development and differentiation of self. Differentiation of self is a universal requirement: All individuals must learn to maintain a distinct self whilst also making intimate connections with others (Knerr & Bartle-Haring, 2010). In Bowen’s model, differentiation of self is defined in terms of intrapsychic and interpersonal traits. On the intrapsychic level, differentiation of self is the capacity to balance emotional and logical processes; on the interpersonal level, it is the ability to create intimate, emotional ties with others whilst remaining on some level independent from them (Bowen, 1978; Kerr & Bowen, 1988).

Kerr and Bowen (1988) described two levels of differentiation of self: basic and functional. The basic level is relatively fixed and non-negotiable. It is determined during a child’s development in his or her family of origin and includes I-position action statements. It consists of integrated beliefs, convictions, and ideas that are invariant across relationships and environments. In contrast, the functional level of differentiation (or pseudo-self) can change according to environment and relationships. It is negotiable and fluctuates according to the relationship system in which one is involved.

Kerr and Bowen (1988) argued that individuals with high differentiation of self tend to have more effective social and relational coping skills, whereas those with low differentiation of self experience greater relational anxiety, function less effectively in stressful situations, and suffer more from physiological and psychological symptoms (Peleg & Zoabi, 2014). Four indicators of differentiation of self have been developed: ability to take an I-position, emotional reactivity, emotional cut-off, and fusion with others (Kerr & Bowen, 1988; Skowron & Friedlander, 1998).

Differentiated people are flexible and can maintain their I-position in intense interactions and relationships. They can remain calm in conflicted relationships, resolve relational problems effectively, and reach compromises. Less differentiated individuals, on the other hand, tend to be overwhelmed by emotions and to feel stress, anxiety, and discomfort in intimate relationships. Fusion with others, emotional cut-off, and emotional reactivity are common reactions to relational stress in undifferentiated people. Fusion with others is defined as excessive emotional involvement in significant relationships, whilst emotional cut-off is the tendency to manage relationship anxiety through physical and emotional distance by rejecting emotional attachments to family and romantic partners. Individuals who score high on an emotional cut-off display exaggerate autonomy and independence to create the illusion of real emotional separation. Emotional reactivity is a person’s tendency to respond to external of internal stressors with intense, irrational emotional activation (Bowen, 1978; Peleg, 2002; Peleg & Zoabi, 2014; Skowron & Friedlander, 1998; Skowron & Schmitt, 2003).

Bowen (1978) argued that more differentiated individuals tend to have better psychological adjustment than less differentiated people, who experience more relational, psychological, and physical problems and symptoms. In recent years this assertion has been the focus of a great deal of theoretical, clinical, and empirical research, most of which has confirmed that differentiated individuals enjoy good physical and psychological health (Skowron, 2000; Skowron, Stanley, & Shapiro, 2009), are less anxious (Peleg & Yitzhak, 2010; Skowron & Dendy, 2004), and reveal high levels of marital satisfaction (Peleg & Yitzhak, 2010) and marital adjustment (Aryamanesh, Fallahchai, Zarei, & Haghighi, 2012; Gubbins, Perosa, & Bartle-Haring, 2010; Lampis, 2016; Rodríguez-González, Skowron, Cagigal de Gregorio, & Muñoz San Roque, 2016; Skowron, 2000).

DIFFERENTIATION OF SELF AND COUPLE RELATIONSHIPS

Our relationships with our first significant caregivers have a big impact on our emotional responses and our sense of our self and others in intimate relationships during adult life. When selecting an intimate partner, individuals often seek someone who resembles a significant parental figure or responds to unsatisfied infantile needs (Solomon, 2003, 2009). Neuroscientific research has confirmed that this primary tendency is imprinted in the brain (e.g., Schore, 2002; Siegel, 2001). Individuals’ early primary relationships and the style of affective regulation they develop through interaction with their primary caregivers influence their synaptic connectivity and thus affect their sense of self, interpersonal relationship style, and perception and conduct of future relationships (Tatkin & Solomon, 2011).

More than any other kind of relationship, intimate relationships with romantic partners tend to recreate the good and bad features of our early experiences of dependency (Solomon, 2003; Tatkin & Solomon, 2011). The stress, pain, and dyadic conflict experienced by romantic partners generally do not depend on issues they are aware of and discuss (finance, children’s education, parents, etc.); rather, they tend to revolve around insecurity in the bond, with at least one partner perceiving the other as either inaccessible or emotionally distant (Atkinson, 2005; Gottman & DeClaire, 1997; Johnson, 2007; Johnson, 2013). Atkinson (2005) maintained that togetherness-first and independence-first tendencies (analogous to Bowen’s emotional fusion and emotional cut-off styles, respectively) are not simply preferences but strategies for maintaining the emotional stability of an intimate relationship.

Differentiation of self seems to be fundamental to an individual’s capacity to achieve intimacy and mutuality in marriage. In recent years, many studies have proposed that differentiation of self is an important predictor of the quality of romantic relationships, and there are many reports that individuals and couples who report good self-differentiation are more satisfied with their intimate relationships and experience less relational conflict than those who report fusion with others, emotional reactivity, or emotional cut-off (Anderson & Sabatelli, 1992; Aryamanesh et al., 2012; Knauth & Skowron, 2004; Lampis, 2016; Lampis, Cataudella, Busonera, & Skowron, 2017; Mohsenian, Karamlo, & Ganjavi, 2008; Peleg, 2008; Rodríguez-González et al., 2016; Sabatelli & Bartle-Haring, 2003; Skowron, 2000). These findings also are consistent with research (Gubbins et al., 2010; Schwartz, Thigpen, & Montgomery, 2006) trying to connect Bowen’s theory self-differentiation with Gottman’s model of marital interactions (Gottman, 2011; Gottman & DeClaire, 1997; Gottman & Silver, 2012) and with research showing that spouses’ differentiation from their birth family was associated with emotional flooding during arguments and marital satisfaction (Gubbins et al., 2010) and with failure to create a stable sense of self (Schwartz, Thigpen, & Montgomery, 2006).

Research on the relationship between couple functioning and differentiation of self has revealed that anxiety about being alone leads individuals with the fusion with other style to blur or dissolve the boundaries between “I” and “we”. The emotional cut-off style is characterized by feeling anxious about losing one’s sense of self and managing this threat to autonomy through physical or emotional avoidance or by creating conflict in one’s intimate relationship. People with an emotionally reactive style tend to react to external and internal stressors with irrational and intense emotional arousal (Harrison, 2003; Lampis, 2016; Lampis, Busonera, Cataudella, & Skowron, 2017; Lampis, Cataudella et al., 2017; Yousefi et al., 2009). What is not clear is how I-position, fusion with others, emotional reactivity, and emotional cut-off affect dyadic adjustment. Some studies suggest that emotional cut-off is the only predictor of couple dissatisfaction (Lampis, 2016; Skowron, 2000; Skowron & Friedlander, 1998), but others show that all components of differentiation predict marital satisfaction (Aryamanesh et al., 2012; Yousefi et al., 2009) and some marital adjustment is influenced by I-position, emotional cut-off, and emotional reactivity, but not fusion with others (Peleg, 2008).

In addition, few studies have considered gender differences in the relationship between differentiation of self and couple’s lifestyle, and the results have been mixed. One study reported that, consistent with Bowen’s (1978) theory, there are no gender differences (Aryamanesh et al., 2012); another found that men’s emotional cut-off predicted marital discord (Skowron, 2000); whilst another connected marital satisfaction to emotional reactivity, emotional cut-off, and I-position in men, but only emotional cut-off in women (Peleg, 2008).

Studies on differentiation of self and couple relationships have partially confirmed Bowen’s assertion that the differentiation of self-construct represents a human universal (Bowen, 1978; Kerr & Bowen, 1988). Studies have been conducted in Western countries, such as the United States (e.g., Skowron & Dendy, 2004), Spain (Rodríguez-González et al., 2016), and Italy (Lampis, 2016; Lampis, Cataudella et al., 2017); in Asian countries, such as Korea (e.g., Kang & Park, 2005); and in Middle Eastern countries, such as Iran (e.g., Yousefi et al., 2009) and Israel (e.g., Peleg, 2008; Rizkalla & Rahav, 2016). Those results suggest that only the fusion with others dimension seems to be affected by cultural context. Fusion with others appears to be sensitive to the differences among Confucian, collectivist Asian cultures, and individualistic Western cultures (Kang & Park, 2005; Lam & Chan-So, 2013).

The Present Study

Many of the studies mentioned previously suggest that undifferentiated interpersonal styles—fusion with others, emotional reactivity, and emotional cut-off—are used to regulate emotional closeness and manage the risk of being overwhelmed by emotion (Bowen, 1978; Kerr & Bowen, 1988) and that they can have a big impact on the life of couples. Bowen’s (1978) theory assumes that individuals’ level of differentiation of self is fundamental to their capacity to achieve intimacy and mutuality in a couple relationships and that individuals with a higher I-position tend to be more satisfied with their intimate relationships and experience less relational conflict than individuals characterized by fusion with others, emotional reactivity, or emotional cut-off (Anderson & Sabatelli, 1992; Aryamanesh et al., 2012; Gubbins et al., 2010; Knauth & Skowron, 2004; Lampis, 2016; Mohsenian et al., 2008; Sabatelli & Bartle-Haring, 2003; Skowron, 2000; Skowron & Dendy, 2004).

We suggest that, in accordance with general systemic principles and with Bowen’s (1978) theory, partners’ interdependence also can contribute to their dyadic adjustment. Indeed, it is possible to develop new relational regulation patterns through relational routines with a significant romantic partner (Fishbane, 2007; Johnson, 2013; Siegel, 2001; Tatkin & Solomon, 2011). When two people interact in a romantic relationship, in fact, their individual outcomes are affected by their partner’s cognitions, emotions, and behavior, and by the global functioning of couple relationship. This implies that an individual’s dyadic adjustment is affected by his or her differentiation of self but also by his or her partner’s differentiation of self (Prest, Benson, & Protinsky, 1998). Taking this logic as our starting point, and given the importance assigned to observation and analysis of interdependence processes in clinical practice and systems research, we chose the actor-partner interdependence model (APIM; Cook & Kenny, 2005; Kashy & Kenny, 2000; Kenny, Kashy, & Cook, 2006) to provide a framework for our analysis of how dyadic adjustment is influenced by the differentiation of self of each dyad member.

In recent years, the APIM has been used to study the couple relationship as a system of interdependent individuals. Attachment theory is the starting point for some of this research, for example, analysis of how adult attachments are related to partner violence (Sommer, Babcock, & Sharp, 2017), partners’ mental health (Gallagher et al., 2017), relationship satisfaction, and sexual satisfaction (Conradi, Noordhof, Dingemanse, Barelds, & Kamphuis, 2017), and marital quality (Sandberg, Bradford, & Brown, 2017). Other studies have adopted a clinical perspective, investigating, for example, the associations between relationship functioning and post-traumatic stress disorder (PTSD) and anger (Roberge, Allen, Taylor, & Bryan, 2016). Yet other studies have looked at the effect of the partner personalities (Brock, Dindo, Simms, & Clark, 2016), alexithymia (Eid & Boucher, 2012; Hesse, Pauley, & Frye-Cox, 2015), and conflict style or work-family balance (Yucel, 2017) on marital quality or dyadic adjustment.

We are aware of only one published study that has applied the APIM to the relationship between differentiation of self and dyadic adjustment (analyzing intimacy as a mediator) in nonindependent dyadic data. This was a study of Palestinian and Jewish married couples (Niveen & Rahav, 2016) that found that regardless of the indicator used, differentiation of self was positively associated with intimacy and intimacy was positively associated with level of dyadic adjustment.

Considering the lack of studies that used APIM to analyze the relationship between differentiation of self and dyadic adjustment, and given the results of earlier research on the contributions of specific dimensions of self-differentiation on couple relationship, the aim of our study was to analyze, in a sample of couples, how individuals’ dyadic adjustment was affected by their own differentiation of self (actor effect) and their partner’s differentiation of self (partner effect).

We hypothesized:

• H1: Both personal and partner I-position have a positive effect on an individual’s dyadic adjustment.

• H2: Personal and partner emotional cut-off, emotional reactivity, and fusion with others have negative effects on an individual’s dyadic adjustment.

METHOD

Participants

A convenience sample of 137 Italian heterosexual couples participated in this study. The men were 20–83 years old (M = 40.80 years; SD = 13.65), and the women were 19–79 years old (M = 37.85 years; SD = 13.11). The modal number of children born in the couple relationship was 1 (M = 0.7; SD = 0.7; range = 0–3) (see Table 1).

Table 1.

Participant Characteristics

Variable	Men	Women
Education
Attended primary school	9.5%	13.2%
Attended secondary school	31.4%	31.6%
Graduated from secondary school	40.9%	44.8%
Employed	67.2%	67.6%
Relationship status
Married	36.5%
Cohabiting	19.0%
Engaged but not cohabiting	36.5%
Length of relationship in years—Mean (±SD)	10.801 (±12.592)

Participants had to be age 18 years or older and involved in a romantic relationship for at least 6 months. Participants were recruited from the population of those attending a series of psychological seminars conducted by the researchers. The seminars were aimed at psychology students, workers, teachers, counselors, and members of local nonprofit, voluntary associations. Participation was voluntary, and participants were told the information provided would be anonymous and remain confidential and that only group data would be reported. Data only from couples in which both partners agreed to participate in research were included. Couples were interviewed in their own homes by the researchers.

The couples were informed about the objectives of the research and given detailed oral and written instructions on how to complete the questionnaires. Written, informed consent was obtained from all participants. The questionnaires were administered to 220 couples but data from 83 couples (37.7%) were excluded because we decided to consider only protocols completed by both partners simultaneously in two different rooms under the supervision of a trained researcher during the home visit lasting about 1 hour. Eighty-three pairs were excluded due to practical and logistical limits that did not allow the two partners to complete the protocol simultaneously in separate rooms (due to the presence in other rooms of family members and/or children). It was established, however, that these 83 pairs did not show statistically significant differences compared to the 137 pairs included in the research and compared to the variables examined in the study [age (t = 1.411; df = 439; p = .159); length of relationship (t = 1.431; df = 439; p = .153); number of children (t = 0.933; df = 439; p = .351); education (χ² = 2.527; df = 2; p = .283); employment (χ² = 1.804; df = 1; p = .179); relationship status (χ² = 3.678; df = 2; p = .159); I-position (t = 0.167; df = 439; p = .868); emotional cut-off (t = 0.673; df = 439; p = .502); fusion with other (t = −.388; df = 439; p = .698); emotional reactivity (t = 1.487; df = 439; p = .138); and dyadic adjustment (t = −1.039; df = 439; p = .299)].

Measures

The Differentiation of Self Inventory (DSI–R; Skowron & Friedlander, 1998; Skowron & Schmitt, 2003) consists of 46 questions; we used the validated Italian version (Lampis, Busonera et al., 2017). The DSI–R assesses four aspects of differentiation of self: emotional reactivity (ER, e.g., When someone close to me disappoints me), I-position (IP, e.g., I am able to say no to others even when I feel pressured by them), emotional cut-off (EC, e.g., I tend to distance myself when people get too close to me), and fusion with others (FO, e.g., I worry about people close to me getting sick, hurt, or upset). The ER and IP subscales are intended to assess the intra-psychic aspects of differentiation, whereas the EC and FO subscales are intended to assess the interpersonal aspects (Skowron, Holmes, & Sabatelli, 2003).

Responses were given on a 6-point Likert scale ranging from 1 (totally disagree) to 6 (totally agree). Subscale scores were the average of responses to the relevant items (range: 1–6). The DSI–R has shown good reliability (Skowron & Schmitt, 2003), and all four subscales of the Italian version have shown good internal consistency (ER, α = .78; EC, α = .85; IP, α = .77; FO, α = .72) as has the scale as a whole (α = .84; Lampis, Busonera et al., 2017). In this sample, the Italian DSI–R also showed good internal consistency (DSI–R: α = .767; ER, α = .602; EC, α = .887; IP, α = .738; FO, α = .634).

The Dyadic Adjustment Scale (DAS; Gentili, Contreras, Cassaniti, & D’Arista, 2002; Spanier, 1976) is a 32-item self-report scale measuring each partner’s adaptation within the relationship and their perception of the quality of the relationship in terms of four subscales: dyadic consensus (CON, e.g., Please indicate below the approximate extent of agreement or disagreement between you and your partner for philosophy of life), dyadic satisfaction (SAT, e.g., How often do you discuss or have you considered divorce, separation or terminating your relationship?), dyadic cohesion (COH, e.g., How often you and your mate calmly discuss something?), and affective expression (AE, e.g., Please indicate below the approximate extent of agreement or disagreement between you and your partner for sex relations). The following internal consistencies have been reported: total DAS score, α = .96; CON, α = .90; SAT, α = .94; COH, α = .86; and AE, α = .73 (Spanier, 1976). The DAS also showed adequate consistency in our sample (CON, α = .781; SAT, α = .696; COH, α = .608; AE, α = .601; total DAS, α = .802). We used only total DAS scores (range: 0–151) in our analyses.

Procedure

The questionnaires were administered to couples during a home visit lasting about 1 hour. Participants completed the DSI–R and DAS and provided basic demographic information in their native language (Italian). The questionnaires were administered separately to each member of the couple by a trained undergraduate psychology researcher. Each member of the couple completed the questionnaires at the same time. Couples did not receive any reward for participating and were informed that the information they provided would remain anonymous and confidential. The study was approved by the Ethics Committee of the University of Cagliari (Italy).

Data Analysis

We first calculated descriptive statistics, then we calculated pairwise correlations between variables (Pearson’s r) and used multivariate analysis of variance (MANOVA) to test for gender differences in the observed variables. Next, to consider the specific nature of the couple data, the dyadic analyses were carried out using the software MPlus 7.0 (Muthén & Muthén, 2012).

After this, we used structural equation modeling (SEM; Kenny et al., 2006), with bootstrapping (5,000 draws) to apply the APIM to the dyadic data. Resampling procedures were used to assess the strength and stability of parameter estimates (Gana, Saada, & Untas, 2013).

The models treated both of the two subjects nested within the dyad and the dyad as the essential unit of analysis (Cook & Kenny, 2005; Kenny et al., 2006; Ledermann, Macho, & Kenny, 2011). We assumed that one person’s score on the independent variables might affect their partner’s score on the dependent variable (e.g., partner effect, between-subjects effect) as well as their own score in the dependent variable (actor effect, within-subjects effect) (Ledermann et al., 2011). The model comprised four measured variables (two for the male partner, partner_1; two for the female partner, partner_2) and two covarying latent error terms (E1 and E2). Specifically, the measured variable X1 (X_male) was defined as a predictor of the measured dependent variables Y1 (DAS_male) and Y2 (DAS_female). Similarly, the measured variable X2 (X_female) was defined as a predictor of the dependent variables Y2 (DAS_female) and Y1 (DAS_male). Thus, the model was characterized by two actor effects (actor effect_male and actor effect_female) and two partner effects (partner effect_male and partner effect_female). The models were used to estimate the influence of each dyad member on the outcome for both them and their partner.

RESULTS

Preliminary Analyses

We first calculated basic descriptive statistics to evaluate gender differences in the study variables (Table 1). Next, we calculated Pearson’s r for the bivariate correlations between dyadic pairs of values for variables (Table 2), which confirmed their nonindependence and thus provided support for the decision to use a dyadic approach. The correlations between DSI–R subscale scores of pairs of participants were less than .70, indicating that they are sufficiently distinct. Linear associations between the variables at individual level and between dyadic partners also were indicated (Table 2).

Table 2.

Bivariate Correlations Between Variables (Values for Men and Women are Displayed Above and Below the Diagonal Respectively)

		1	2	3	4	5	6	7
1	Age	—	.465**	−.120	−.030	.028	−.024	.030
2	Length of relationship	.421**	—	.016	.096	.100	.159	.024
3	I Position	.039	.041	—	−.191*	.215*	−.089	.358*
4	Emotional cut-off	−.062	.108	−.260**	—	−.025	.591**	−.048
5	Fusion with other	.113	.002	.398**	−.387**	—	.366**	.049
6	Emotional reactivity	.345**	.241**	.066	.126	.429**	—	.003
7	Dyadic adjustment	−.177*	−.131	.247**	.007	.146	−.110	—

Note.

^**p < .01;

^*p < .05.

MANOVA was used to assess gender differences in the dependent variables (IP, EC FO, ER, and DAS). There was a multivariate effect (Wilks’s lambda = .904; df = 5,268; p = .005) and a univariate effect of IP (Table 3), such that men had higher IP scores than women.

Table 3.

MANOVA to Assess Gender Differences in Study Variables

	Men		Women		F df = 1,274
	Mean	SD	Mean	SD
I-position	4.501	0.686	4.155	0.724	16.430**
Emotional cut-off	3.777	1.218	3.877	1.314	0.429
Fusion with other	3.818	0.583	3.743	0.801	0.780
Emotional reactivity	3.465	0.721	3.447	0.702	0.043
Dyadic adjustment	120.474	9.970	119.700	10.856	0.253

Note.

^**p < .01;

^*p < .05.

We used multiple linear regression to evaluate the associations between age and length of relationship and dyadic adjustment (DAS score). The analyses did not reveal relationships between dyadic adjustment and participant’s age or length of relationship (R² = .006; β_age = −.059, t = −0.859, p = .391; β_{length_of_relationship} = −.030, t = −0.433, p = .665), so these variables were not included in the structural models. The independent variables for the structural models were IP, FO, EC, and ER (coded separately for men and women), and the dependent variable was the DAS score (also coded separately for men and women).

Structural Models

We used structural equation modeling (SEM) with maximum likelihood estimation and 5,000 bootstrap draws to test the APIM parameters (Cook & Kenny, 2005; Kenny et al., 2006; Ledermann et al., 2011). A different model was run in relation to each DSI–R scale, measured in relation to two partners; as previously stated, in each model the independent variables were the DSI–R dimensions (in turn, emotional cut off, I-position, fusion, emotional reactivity), and the dependent variable was the DAS total score. Table 4 shows the results for each model.

Table 4.

Results of Structural Models (N = 137 dyads)

	Men			Women
	β	p	95% CI	β	p	95% CI
I-Position on dyadic adjustment
Within-subject effect	.318**	.0001	0.189; 0.447	.180*	.030	0.043; 0.316
Between-subject effect	.124	.130	−0.011; 0.258	.221**	.007	0.086; 0.357
R2 for DAS	.138*	.012		.103*	.037
Emotional cut-off on dyadic adjustment
Within-subject effect	.108	.363	−0.088; 0.304	.331*	.004	0.140; 0.521
Between-subject effect	.018	.877	−0.178; 0.215	−.289*	.013	−0.481; −0.098
R2 for DAS	.015	.360		.036	.152
Fusion with others on dyadic adjustment
Within-subject effect	.021	.803	−0.120; 0.162	.137	.103	−0.001; 0.275
Between-subject effect	.044	.611	−0.097; 0.184	.058	.495	−0.082; 0.197
R2 for DAS	.003	.767		.024	.356
Emotional reactivity on dyadic adjustment
Within-subject effect	−.122	.148	−0.260; 0.017	−.005	.956	−0.145; 0.136
Between-subject effect	−.007	.938	−0.146; 0.133	−.046	.586	−0.187; 0.094
R2 for DAS	.015	.470		.002	.785

Note.

^**p < .01;

^*p < .05.

At the first step in the analyses, the structural models were fully saturated and a perfect fit to the data (χ² = 0; df = 0; CFI = 1.00; TLI = 1.00; RMSEA = 0.00). Subsequently, the models were rerun with nonsignificant paths set to zero; all models were a good fit to the data (I-position, χ² = 2.242; df = 1; p = .1343; CFI = 0.987; TLI = 0.936; RMSEA [95% CI] = 0.095 [0.000; 0.268]; emotional cut-off, χ² = .381; df = 1; p = .5371; CFI = 1.000; TLI = 1.034; RMSEA [95% CI] = 0.000 [0.000; 0.192]; fusion with others χ² = .770; df = 2; p = .6805; CFI = 1.000; TLI = 1.039; RMSEA [95% CI] = 0.000 [0.000; 0.128]; and emotional reactivity, χ² = .300; df = 2; p = .8605; CFI = 1.000; TLI = 1.056; RMSEA [95% CI] = 0.000 [0.000; 0.090]).

Although all four models had satisfactory fit indices, only IP and EC affected actor and partner DAS score. FO and ER did not affect the partner’s DAS score.

I-Position

Male IP was positively related to personal DAS score (β = .318; p = .0001) and partner’s DAS score (β = .221; p = .007) whereas female IP was positively related to personal DAS score (β = .180; p = .030) but not partner’s DAS score (Table 4). This model explained 13% of the variance in male partners’ dyadic adjustment and 10% of the variance of female partners’ dyadic adjustment (Figure 1).

Figure 1.

Structural Model for I-Position.

Note. Standardized coefficients are shown; *p < .05; **p < .01.

Emotional Cut-Off

Male EC was not related to personal DAS score (β = .108; p = .363), but was negatively related to partner’s DAS score (β = −.289; p = .013). Female EC was positively related to personal DAS score (β = .331; p = .004) but not to partner’s DAS score (β = .018; p = .887) (Table 4). This model explained very little of the variance in personal and partner’s dyadic adjustment (Figure 2).

Figure 2.

Structural Model for Emotional Cut-Off.

Note. Standardized coefficients are shown; *p < .05; **p < .01.

DISCUSSION

The main objective of our study was to use the APIM (Cook & Kenny, 2005; Kashy & Kenny, 2000; Kenny et al., 2006) to study the effect of romantic partners’ levels of differentiation of self on their own (actor effect) and their partner’s (partner effect) dyadic adjustment.

Our results indicated that in men personal I-position was the strongest predictor of dyadic adjustment, whilst in women both personal and partner’s I-position and emotional cut-off predicted dyadic adjustment.

Like other studies (e.g., Anderson & Sabatelli, 1992; Aryamanesh et al., 2012; Gubbins et al., 2010; Lampis, 2016; Lampis, Busonera et al., 2017; Lampis, Cataudella et al., 2017; Peleg, 2008; Skowron, 2000; Skowron & Dendy, 2004), our research provides support for Bowen’s theory and for the notion that individuals with a greater capacity to take an I-position in their relationships with others experience better dyadic adjustment, perhaps because they are better at creating functional affective relationships and responding empathically to their partners. In both men and women, I-position had a positive effect on personal dyadic adjustment (actor effect). Self-reported personal dyadic adjustment is higher in people who are comfortable with their feelings and able to access them freely, successful in maintaining a measure of autonomy in their close relationships, capable of staying calm in conflicted relationships, resolving relational problems effectively, and reaching compromises. In women, EC also was related to partner’s dyadic adjustment (partner effect). In our sample, women who perceived intimacy as threatening, preferred to distance themselves from their feelings and emotions and to isolate themselves from others, denied the importance of family and romantic relationships, and displayed a facade of independence reported better dyadic adjustment. This finding is quite surprising and is not in line with other empirical research, but it may make more sense in the context of the rest of our results.

In our sample, women with a tendency to cut themselves off emotionally showed good adaptation to their couple relationship, but if their partner shared their tendency to emotional detachment, these women showed lower dyadic adjustment. The women in our sample may have actively and strongly avoided reporting negative feelings about their relationship and this probably was reflected in how they rated it. If we had assessed the dyadic adjustment via objective measures (e.g., behavioral observation tasks), perhaps we would have found an association between greater emotional cut-off and critical levels of dyadic adjustment. Incorporating some of these objective measures into future analyses would provide a better picture of these associations.

We also speculate that the cultural characteristics of our sample may have influenced our results. For specific historical, political, economic, and religious reasons, there is a gap between traditional and modern gender norms in Italy, and the transformation in gender norms is taking place more slowly and laboriously than in other Western countries (Ruggiero, Prandin, & Mantero, 2003; Tager & Good, 2005). The traditional notion that men are the stronger sex and women the weaker sex persists (Paladino, Zaniboni, Fasoli, Vaes, & Volpato, 2014).

Given this context, it is possible that social desirability bias may have influenced our results. The women in our sample may have exaggerated their independence from their family (by assigning high values to EC items) to emphasize a trait that, in fact, did not negatively affect the perceived couple’s adjustment level. It appears that women are less likely to achieve good dyadic adjustment within a couple relationship if their partner tends to emotional detachment and denial of the importance of family and romantic relationships, and displays a facade of exaggerated independence.

These findings, however, should be studied in depth because our models explained only very low proportions of variance in dyadic adjustment. Furthermore, our sample was quite small, which may have influenced the results. Finally, other unmeasured variables may be responsible for the observed variation in dyadic adjustment in our sample.

It is interesting that in women emotional cut-off was associated with personal dyadic adjustment, but only when controlling for partner’s emotional cut-off. In fact, this appears to be a suppression effect related to the absence of bivariate correlation between female emotional cut-off and her dyadic adjustment (r = .007). These data might be explained considering the specificity of interdependent dyadic data and, in our opinion, it further reinforces the choice to use a simultaneous system of regressions as the APIM models to enlighten these relationships. This suppression effect might be better explored in further research.

The fusion with others and emotional reactivity scores were not associated with any of the dyadic adjustment variables. This is not surprising in the case of the former and replicates other research (e.g., Lampis, 2016; Lampis, Busonera et al., 2017; Lampis, Cataudella et al., 2017; Peleg, 2008). Our sample was recruited in Italy, a Western country in which the individualistic values common to Western countries coexist with the strong collectivistic family values common to Eastern countries (Lampis, 2016). It seems that the fusion with others scale may have tapped normal processes of regulation of self and interaction experiences rather than the over-involvement with significant others that characterizes individuals with poor differentiation (Lam & Chan-so, 2013; Lampis, Busonera et al., 2017; Lampis, Cataudella et al., 2017). The emotional reactivity results run counter to the results of other studies (e.g., Lampis, 2016; Peleg, 2008; Skowron, 2000; Skowron et al., 2009) but they may have been influenced by the nature of our non-clinical sample, which—as a whole—showed good differentiation of self and dyadic adjustment. Our findings on the relationships between dyadic adjustment and emotional cut-off and fusion with others will be explored in more depth in future research. We speculate that fusion with others—and in particular emotional reactivity—may prove important predictors in a larger sample or in clinical samples of couples that seeking a therapy for high levels of conflict or emotional regulation problems. In future research, we also may analyze the role of potential moderators of the relationship between dyadic adjustment and differentiation of self, such as romantic attachment style, conflict communication style, and psychological distress.

Our data-connected partner effect revealed that, consistent with general systemic principles and with Bowen’s theory, dyadic adjustment is affected by individual’s differentiation of self and that of their dyadic partner and hence by specific interdependence processes (Prest et al., 1998). As Kenny et al. (2006, p. 150) argued:

If there were partner effect, then there would be evidence that the two persons are part of an interdependent system. It follows logically that the greater the interdependence, the greater the partner effect. … Therefore, the more important the partner is to someone (consciously or unconsciously), the more he or she would be affected by the partner’s characteristics.

A systemic interpretation of our results revealed that globally, interdependence with respect to I-position and emotional cut-off appear to be most important when explaining dyadic adjustment for women than for men (i.e., higher explained variance, more significant direct actor and partner effect). Women’s dyadic adjustment is related to their personal I-position and emotional cut-off but also to their partner’s I-position and emotional cut-off; as a consequence, the global functioning of their relationship is affected in terms of dimensions of differentiation of self. Women whose partners report a good balance between personal autonomy and affective involvement in the couple relationship tend to show good dyadic adjustment. In contrast, women with partners who find intimacy threatening, are emotionally detached, isolate themselves, and display exaggerated independence tend to show lower dyadic adjustment.

Our data also revealed that men’s dyadic adjustment depends only on personal I-position; it appears that their perceptions of cohesion, affective expression, satisfaction, and consensus in their couple relationships are based solely on intrapsychic factors.

These findings suggest that the women in our sample were more relationship oriented than the men, and that the specific ways in which they live their couple relationship are affected by different relational fonts and are more influenced by interdependency processes.

Our results can be interpreted in terms of the relational-interdependent, self-construal construct (Cross, Bacon, & Morris, 2000), which describes individuals’ tendency to define themselves in terms of their close relationships and to include their significant relationships in their self-representation alongside other self-defining characteristics (Cross, S. E., Gore, J. S., & Morris, M. L. (2003). In contrast, the self-definition of a person with an independent self-construct is based largely on his or her personal attributes and the emphasis is on maintaining a sense of autonomy (Markus & Kitayama, 1994).

Some empirical research has shown that compared with men, women tend to move in their relational contexts, and especially in intimate relationships, with a higher level of relational self-construct. Women describe themselves in more relational terms, consider relationship-oriented aspects of their identity more important, are more careful with others, talk more about their relationships, and have a better memory for close others and relationship events (Cross & Madson, 1997; Impett & Peplau, 2006; Linardatos & Lydon, 2011). Our data are consistent with this pattern, but further research is needed to explore what is a preliminary finding, especially in view of the low proportions of variance in dyadic adjustment explained by our models.

Limitations and Future Research

Although our study provides the first empirical evidence that individuals’ dyadic adjustment is related to their partner’s differentiation of self, it has several limitations that should be addressed in the future. First, the sample consisted exclusively of couples who volunteered to participate in a research project. The participating couples tended to be highly educated, financially well-off, and report good dyadic adjustment; generalizing our findings to samples that do not share these characteristics is problematic. Replicating our findings in other samples (e.g., high-conflict couples, couples with a low socio-economic status, couples who have sought therapy) would address this limitation. A related point is that our research was carried out in the West, and although recent studies have not found differences between individualistic and collective cultures with respect to the relationship between differentiation of self and dyadic functioning (Rizkalla & Rahav, 2016), some caution is necessary in generalizing our findings to other cultures.

A further limitation is that we relied solely on self-report data. It is thought that responses to questions about sensitive issues, such as the quality of intimate relationships, can be influenced by cognitive biases such as denial, idealization, and social desirability bias. This could be addressed through qualitative research assessing couple functioning and differentiation processes, such as genogram interviews.

In addition, when examining how the interaction between personal and partner differentiation of self affected dyadic adjustment, we did not consider potential moderation or mediation factors whether individual (e.g., personality, self-esteem), dyadic (e.g., conflict, resilience), or contextual (e.g., social support, socio-economic status). Incorporating some of these variables into future analyses would provide a better picture of the complex network of variables contributing to dyadic adjustment and perhaps enable us to explain a higher proportion of variance.

The cross-sectional nature of the study is also an important limitation. In fact, it does not enable us to evaluate any inverse and/or reciprocal associations between the variables investigated. To overcome this limitation, these dimensions could be analyzed in a longitudinal perspective. Finally, we did not consider the circularity of relationships between studied variables. In the Introduction, we noted that Bowen and Kerr (1988) described two levels of differentiation of self: basic and functional. The functional level of differentiation reflects the component of self that can change with the environmental and relationship context and is negotiable. Accepting this theoretical model of differentiation of self implies that the relationship between differentiation of self and dyadic adjustment is bidirectional. In future research it would be worth investigating if there is a causal relationship between dyadic adjustment and differentiation of self, and, if so, what the direction of influence is.

Practical Implications

In view of these methodological limitations it is rather difficult to draw firm conclusions from our results, but we believe this study constitutes a step toward more complex models of the nature, function, and relationship between couple functioning and trigenerational processes of differentiation. We believe that our findings have some implications for systems research and therapy.

The couple is a complex relational system, characterized by many levels of functioning that affect each other over time: interactive and communicative patterns, emotional bonds, symbolic dimensions, etc. The behavior, emotions, feelings, and personality of an individual also are influenced by the behavior, emotions, feelings, and personality of their intimate partners throughout the lifespan. This assumption, due to their innumerable methodological implications, constitutes one of the main problems those conducting studies in this field must manage and represents a challenge for systemic researchers.

Our results have revealed the importance of studying the interdependence of partners’ differentiation of self to arrive at a more detailed picture of their relationship.

Because our data were collected from an opportunity sample in a nonclinical setting, we must be cautious in generalizing our findings to clinical populations; nevertheless, we think that our results have some clinical implications and represent a challenge to systems and family therapists. Our results provide support for the proposal that differentiation of self can be a central aspect of couple-based interventions.

Our results suggest that a preliminary assessment of the differentiation of self of both members of a couple could help clinicians and counselors to identify specific maladaptive patterns in the couple relationship. The information obtained from such an assessment could be used in several ways to facilitate couple counseling or therapy (Lampis, Busonera et al., 2017; Lampis, Cataudella et al., 2017). Information about the specific nature of a couple’s pattern of differentiation of self might enable the clinician to tailor the intervention to the parties’ levels of differentiation of self, to choose effective psychotherapeutic interventions, and to concentrate on clinical work that will help the couple understand how their individual levels of self-differentiation contribute to their current marital problems.

Psychological interventions could be used to increase the I-position status of members of a couple or to decrease their tendency to emotional cut-off. Partners also could be helped to reflect on the systemic functioning of their couple relationship and on the fact that they are part of an interdependent system in which their emotions, feelings, behaviors, and cognitions influence their partner and are influenced circularly by his/her partner as well. This might improve the partners’ self-regulation and regulation of the relationship, and thus create a new emotional climate for interactions within the relationship.