Trait Emotional Intelligence and Relationship Satisfaction: The Mediating Role of Dyadic Coping

Abstract

Trait emotional intelligence (TEI) is an important individual difference variable that is related to the quality of romantic relationships. The present study investigated the associations between TEI, dyadic coping, and relationship satisfaction. A convenience sample of N = 136 heterosexual couples was recruited online. When the actor-partner interdependence model was applied to the data, TEI showed a positive actor effect and a positive partner effect on relationship satisfaction. The actor effect and partner effect of TEI on relationship satisfaction were partially mediated through positive dyadic coping and common dyadic coping, respectively. A small total indirect actor effect was also found for negative dyadic coping. Controlling for potential content overlap between TEI and relationship satisfaction did not alter the results. However, removing variance from the TEI score that was shared with the Big Five trait factors attenuated TEI’s actor and partner effects on relationship satisfaction and rendered all but one actor effect for TEI on dyadic coping and all but one indirect effect nonsignificant. The results underline the importance of TEI for the quality of romantic relationships and they shed light on underlying mechanisms. Implications for theory, research, and applications in counseling contexts will be discussed.

Keywords:

• Actor-partner mediator model
• Big Five
• dyadic coping
• trait emotional intelligence
• relationship satisfaction
• interpersonal relations
• personality
• quality of life

Forming and maintaining a long-term romantic relationship is an important developmental task in early adulthood (Shulman & Connolly, 2013), and being in a stable, high-quality romantic relationship likely promotes one’s well-being and health (Proulx, Helms, & Buehler, 2007; Robles, Slatcher, Trombello, & McGinn, 2014). Thus, it seems important to identify the mechanisms that contribute to the quality and stability of romantic relationships. Both partners’ personality traits such as neuroticism may affect the quality and trajectory of their partnership (Heller, Watson, & Ilies, 2004; Malouff, Thorsteinsson, Schutte, Bhullar, & Rooke, 2010; Solomon & Jackson, 2014). In their vulnerability-stress-adaptation (VSA) model of marital development, Karney and Bradbury (1995) proposed that traits reflect enduring vulnerabilities that contribute to dyadic stress and to adaptive dyadic processes that couples deploy in order to deal with their stress. Inadequate adaptive processes, in turn, lead to poorer relationship quality, which promotes instability. The present study focuses on three elements of the VSA-model: personality traits, adaptive processes, and relationship quality. Recently, Malouff, Schutte, and Thorsteinsson (2014) showed that trait emotional intelligence (TEI) is positively related to relationship quality. However, little research has been devoted to adaptive mechanisms that may link TEI and relationship quality. Drawing on a dyadic design, the current study aims to provide further evidence that one’s own TEI is related to both partners’ relationship satisfaction and that these associations are mediated via dyadic coping.

Trait Emotional Intelligence and Relationship Quality

Trait emotional intelligence denotes a constellation of emotion-related dispositions or trait emotional self-efficacies that forms a separate factor in personality space, that is partially determined by several personality dimensions, and that is located at the lower levels of established trait hierarchies (Petrides, Pita, & Kokkinaki, 2007). The TEI sampling domain comprises 15 facets (e.g., emotion perception, emotion regulation, trait happiness, emotion management), of which 13 facets fall on four factors (emotionality, sociability, well-being, self-control), and two facets along with the four TEI factors are markers of a global TEI factor (Petrides, 2009). Defined in this way, TEI concerns how a person perceives his or her own socio-emotional effectiveness, which is assumed to, at least partially, reflect his or her actual socio-emotional effectiveness. In line with this hypothesis, greater TEI is related to a more frequent use of adaptive and less frequent use of maladaptive coping strategies (Petrides, Pita, & Kokkinaki, 2007), better health outcomes (Martins, Ramalho, & Morin, 2010) and subjective well-being (Sánchez-Álvarez, Extremera, & Fernández-Berrocal, 2016), better peer-relations, more peer nominations for adaptive social behaviors, cooperation, and social competence in adolescence (Frederickson, Petrides, & Simmonds, 2012; Mavroveli, Petrides, Rieffe, & Bakker, 2007).

As proposed by the VSA-model (Karney & Bradbury, 1995), personality traits such as neuroticism or agreeableness of both partners affect the quality and the trajectory of a romantic relationship (Heller et al., 2004; Malouff et al., 2010; Solomon & Jackson, 2014). Low TEI might accordingly represent an enduring vulnerability that interferes with adequate dyadic processes and impairs the relationship quality. Conversely, high TEI might foster adequate dyadic processes and help to develop and sustain high-quality relationships. Partially consistent with this hypothesis, a recent meta-analysis on correlations between TEI and relationship quality found a moderate within-person effect, overall r = .33, p < .001, and a weak between-person effect, overall r = .22, p < .001 (Malouff et al., 2014). However, these effects were based on only 10 effect sizes (within-person) and eight effect sizes (between-person), which might impair their precision and generalizability. More studies are therefore needed to expand the literature on the associations between TEI and relationship quality.

Significant within-person or actor effects for TEI on relationship quality have emerged consistently in previous research (Malouff et al., 2014). However, corresponding between-person or partner effects were neither consistently investigated (e.g., Joshi & Thingujam, 2009; Schutte et al., 2001) nor were significant partner effects consistently found in prior research (e.g., Smith, Heaven, & Ciarrochi, 2008; Zeidner, Kloda, & Matthews, 2013). The latter inconsistency could be partially due to the fact that the TEI inventories used differ in the degree to which they cover the TEI sampling domain. Inventories that provide a full coverage of the TEI sampling domain (e.g., Trait Emotional Intelligence Questionnaire, TEIQue; Petrides, 2009) are likely superior predictors of psychological criteria when compared with less comprehensive TEI inventories (Gardner & Qualter, 2010), and thus may also yield stronger partner effects. This implies a need for more research on TEI’s partner effects on relationship quality using more comprehensive TEI inventories. Using a dyadic design and the TEIQue-SF (Petrides, 2009), the current study aims to investigate TEI’s actor and partner effects on relationship satisfaction in a sample of German speaking heterosexual couples.

By definition, TEI overlaps with the Big Five trait factors (i.e., openness to experience, conscientiousness, extraversion, agreeableness, and neuroticism; e.g., Goldberg, 1993), which usually account for 50% to 70% of the variance in TEI (e.g., Petrides et al., 2010). Despite this overlap, a meta-analysis showed that TEI explains unique variance in various criteria (e.g., subjective well-being, individual coping) beyond the Big Five traits (overall ΔR² = .06; Andrei, Siegling, Aloe, Baldaro, & Petrides, 2016). This incremental validity of TEI might extend to relationship satisfaction (Joshi & Thingujam, 2009) and to dyadic coping, which was also investigated in this study.

Finally, TEI’s well-being factor includes facets such as trait happiness (Petrides, 2009). This gives rise to a potential content overlap between total TEI and relationship satisfaction, which might bias the association between both constructs (Petrides et al., 2007). Therefore, a further objective of this study was to control for the effect of this potential content overlap.

Trait Emotional Intelligence and Adaptive Processes

In Karney and Bradbury (1995) VSA-model, traits affect the relationship quality indirectly through their associations with adaptive processes and dyadic stress. To date, only a few studies examined the adaptive processes that may underlie the associations between TEI and relationship quality: For example, Smith et al. (2008) showed that TEI is related to conflict communication styles, which are in turn associated with relationship satisfaction. In Schröder-Abé and Schütz (2011), perspective taking fully mediated TEI’s actor effect and partially mediated TEI’s partner effect on relationship satisfaction. In Zeidner et al. (2013), total dyadic coping mediated TEI’s actor effects on marital quality. However, Zeidner et al. (2013) neither found partner effects for TEI and total dyadic coping on marital quality nor did they consider specific dyadic coping strategies. Given that the role of dyadic coping in the TEI-adaptive processes-relationship satisfaction pathway is poorly understood, the current study draws on three different self-reported dyadic coping strategies, which provides a more fine-grained picture of the underlying adaptive processes than a single total dyadic coping index.

In his theory of dyadic coping, Bodenmann (2008) conceives coping in relationships as a dyadic phenomenon: Individual stress and dyadic stress affect both partners and have to be managed by both partners. He distinguishes three dyadic coping strategies: Positive dyadic coping involves the provision of problem- and emotion-focused support and reducing the partner’s stress by a new division of responsibilities and contributions to the coping process. Common dyadic coping (i.e., joint dyadic coping) includes strategies in which both partners jointly engage to reduce stress (e.g., exchange tenderness, joint problem solving). Negative dyadic coping comprises insufficient support and ambivalent or hostile intervention attempts (e.g., reluctant provision of support while believing that the partner should solve the problem alone). The latter strategy reflects maladaptive dyadic coping, while both the former strategies reflect adaptive dyadic coping.

Preliminary evidence provides direct and indirect support for the supposed TEI-dyadic coping relationships: Zeidner et al. (2013) found a positive actor effect for TEI on total dyadic coping. Moreover, adaptive individual coping is positively related to positive dyadic coping and to common dyadic coping, and negatively related to negative dyadic coping, while maladaptive individual coping is negatively related to common dyadic coping (Herzberg, 2012; Papp & Witt, 2010). Adaptive individual coping is, in turn, positively correlated with TEI, whereas maladaptive individual coping is negatively correlated with TEI (Mavroveli et al., 2007; Petrides et al., 2007). These lines of evidence are consistent with the hypotheses that TEI might be positively associated with positive dyadic coping and common dyadic coping and negatively associated with negative dyadic coping.

Also consistent with the VSA-model, relationship quality correlates positively with positive dyadic coping and common dyadic coping, and negatively with negative dyadic coping (Ledermann et al., 2010). Herzberg (2012) showed positive actor effects for common dyadic coping on relationship satisfaction and a positive partner effect for females’ common dyadic coping on males’ satisfaction. Similarly, Papp and Witt (2010) showed positive actor effects for positive dyadic coping and negative actor effects for negative dyadic coping on relationship satisfaction; only females’ positive and negative dyadic coping had significant partner effects on males’ relationship satisfaction, respectively. Finally, a recent meta-analysis found positive associations between relationship satisfaction and total dyadic coping (strong overall association) and overall dyadic coping by self (moderate association; Falconier, Jackson, Hilpert, & Bodenmann, 2015). In subgroup analyses, associations between total dyadic coping and relationship satisfaction were stronger for within-person data than for between-person data. For within-person data, Falconier et al. (2015) also showed that aggregated positive forms of dyadic coping are more strongly related to relationship satisfaction than aggregated negative forms of dyadic coping. These results are consistent with our hypothesis that dyadic coping might mediate portions of the TEI-relationship satisfaction association for both between-person and within-person data.

Research Questions and Hypotheses

Based on theory (e.g., Karney & Bradbury, 1995) and prior evidence on associations between TEI and relationship quality (e.g., Malouff et al., 2014), the first hypothesis stated: TEI would have a positive actor effect on relationship satisfaction (H1a) and a positive partner effect on relationship satisfaction (H1b). In line with theory (Bodenmann, 2008; Karney & Bradbury, 1995) and with previous findings (e.g., Zeidner et al., 2013), the second hypothesis stated: TEI would be positively related to one’s own positive dyadic coping (H2a) and common dyadic coping (H2b) and negatively to one’s own negative dyadic coping (H2c). As partner effects for individual difference variables on dyadic coping have been rarely found (e.g., Herzberg, 2012; Papp & Witt, 2010; Zeidner et al., 2013), TEI-dyadic coping partner effects were investigated in an exploratory way (EQ1). In previous studies, total dyadic coping mediated TEI’s actor effects on relationship satisfaction (Zeidner et al., 2013), and common dyadic coping mediated both the actor effects of individual coping on relationship satisfaction and the partner effect of females’ individual coping on males’ satisfaction (Herzberg, 2012). Thus, we hypothesized that dyadic coping might partially mediate the TEI-relationship satisfaction effects. Accordingly, the third hypothesis stated: Portions of the TEI-relationship satisfaction actor and partner effects would be mediated by positive dyadic coping (H3a), common dyadic coping (H3b), and negative dyadic coping (H3c). Little is known about the utility of TEI to account for unique variance in relationship satisfaction (Joshi & Thingujam, 2009) and dyadic coping beyond the Big Five traits. The present study thus investigated in an explorative way, whether the effects of TEI remain significant when overlap with the Big Five traits has been removed from the total TEI score (EQ2). Finally, to control for the potential content overlap between TEI and relationship satisfaction (Petrides et al., 2007), the present study also addressed the explorative research question whether the observed effects for TEI remain significant when items pertaining to the well-being factor have been removed from the total TEI score (EQ3).

Method

Participants and Procedure

Participants were recruited during autumn 2011 by the last author by sampling in her network of acquaintances, by public notices (e.g., supermarket, kindergarten), and postings on social media (e.g., Facebook, studiVZ, gofeminin). Participants were invited to fill out an online questionnaire which took about 20 minutes to complete (inclusion criteria: age between 18 to 80 years, German as first language, being in a heterosexual relationship for at least 6 months). After providing informed consent, participants completed an online questionnaire containing several scales presented in a random order. No material incentive was given for participation. Research in non-clinical settings, that only entails self-reports, ensures anonymity, and is not expected to cause harm to the adult study subjects does not usually require institutional review board approval under German regulations.

The sample consisted of N = 136 heterosexual couples (N = 272 participants), with an average age of M = 33.10 years (SD = 12.45; range: 20-69 years) for females and M = 35.63 years (SD = 13.09; range: 19–73 years) for males. About two third of the couples (67.6%) were unmarried. Couples were either cohabitating (70.6%), had a long-distance relationship (16.2%) or lived apart (13.2%). About one third of the couples (36.0%) had children. The average relationship duration was M = 9.44 years (SD = 10.88, range: 1–45 years).

Measures

Trait Emotional Intelligence Questionnaire-Short Form (TEIQue-SF; Petrides, 2009). The TEIQue-SF provides a global assessment of TEI using two items per TEI facet. The 30 items (e.g. “On the whole, I’m able to deal with stress.”) are answered on a 7-point scale (1 = ‘completely disagree’ to 7 = ‘completely agree’). Prior research documented the reliability and validity of the German TEIQue-SF (e.g., Jacobs, Sim, & Zimmermann, 2015; Jacobs, Wollny, Sim, & Horsch, 2016). The total TEI score was derived by applying the a-priori TEIQue scoring key. Further, a reduced TEI score consisting of 24 items was created (i.e., six items belonging to the well-being factor were removed). In this study, Cronbach’s α of total TEI and reduced TEI were α = .87 and α = .82, respectively (see Table 1 for descriptive statistics).

Table 1. Descriptive Statistics and Correlations of Study Variables Within and Between Females and Males.

		1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21
1.	Duration	–
Females
2.	TEI	−.09	(.85)
3.	PDC	−.17	.30	(.77)
4.	NDC	.11	−.16	−.34	(.72)
5.	CDC	−.29	.30	.52	−.22	(.75)
6.	SAT	−.32	.30	.39	−.30	.51	(.90)
7.	O	−.18	.25	.23	−.09	.23	.00	(.61)
8.	C	.01	.36	.01	−.12	−.01	.06	.02	(.72)
9.	E	−.02	.36	.07	−.04	.11	.13	.24	−.02	(.82)
10.	A	.15	.33	.31	−.21	.20	.07	.15	.14	−.02	(.40)
11.	N	−.07	−.51	−.15	−.04	−.05	−.05	−.09	−.11	−.29	−.20	(.80)
Males
12.	TEI	−.12	.17	.02	−.11	.20	.33	−.02	.10	.12	−.04	.08	(.88)
13.	PDC	−.35	.07	.25	−.19	.21	.32	.06	.03	.06	.03	.14	.36	(.76)
14.	NDC	.15	−.10	−.26	.27	−.19	−.31	−.07	.04	.10	−.12	.12	−.18	−.36	(.72)
15.	CDC	−.15	.13	.37	−.02	.43	.33	.19	−.07	.16	.21	−.02	.27	.42	−.25	(.74)
16.	SAT	−.29	.17	.40	−.24	.45	.64	.08	.05	−.00	.09	.02	.39	.42	−.33	.53	(.91)
17.	O	−.15	.08	−.07	.03	.17	.24	.06	.01	.01	.07	.07	.37	.33	−.09	.16	.16	(.73)
18.	C	.13	−.01	−.13	−.02	−.07	−.09	−.08	.04	.00	−.08	.06	.23	−.07	−.02	−.15	−.12	.07	(.62)
19.	E	−.13	−.00	.06	−.02	.14	.06	.03	.07	−.06	−.01	−.06	.24	.25	−.16	.17	.21	.26	−.06	(.75)
20.	A	−.00	−.02	−.01	−.08	.07	.08	.03	.02	−.09	−.16	−.02	.42	.29	−.25	.23	.20	.05	.14	−.05	(.51)
21.	N	.10	−.03	−.06	.17	−.15	−.20	.07	−.03	−.10	.09	−.13	−.54	−.22	.16	−.23	−.38	−.06	−.01	−.07	−.14	(.70)
	M	9.44	5.38	3.90	1.52	3.47	4.11	5.36	5.42	5.04	5.64	4.26	5.24	3.93	1.76	3.43	4.11	5.29	5.25	4.63	5.23	3.49
	SD	10.88	0.62	0.66	0.59	0.76	0.75	1.11	1.17	1.40	0.89	1.47	0.70	0.49	0.64	0.71	0.75	1.24	1.05	1.30	1.00	1.35

Note. Duration = relationship duration in years, TEI = trait emotional intelligence, PDC = positive dyadic coping, NDC = negative dyadic coping, CDC = common dyadic coping, SAT = relationship satisfaction; O = openness to experience; C = conscientiousness; E = extraversion; A = agreeableness; N = neuroticisms; M = mean; SD = standard deviation; Cronbach’s alpha is shown in brackets. Correlations in italics are significant at p < .05, correlations in bold face are significant at p < .01, and underlined correlations are significant at p < .001 (2-tailed).

Dyadic Coping Inventory (DCI; Bodenmann, 2008). The DCI consists of 37 items that assess dyadic coping perceived either by self (6 subscales) or by the partner (4 subscales). All items are answered on a 5-point scale (1 = ‘very rarely’ to 5 = ‘very often’). In prior research, the DCI subscales demonstrated good reliabilities and satisfying convergent, discriminant, criterion, and prognostic validity (e.g., Ledermann et al., 2010). In the current study, only self-perceptions were used: The positive dyadic coping subscale (7 items; e.g., “I take on tasks and activities that he usually does to relieve him.”) aggregates supportive dyadic coping (i.e., providing problem-focused and emotion-focused support) and delegated dyadic coping (i.e., taking on responsibilities and reducing the partner’s stress by a new division of contributions to the coping process).¹ The negative dyadic coping subscale (4 items; e.g., “I don't take his stress seriously.”) assesses hostile dyadic coping, reluctant, insufficient, or inefficient support, and insincere support. The common dyadic coping subscale (4 items; e.g., “We help each other to see the problem in a new light.”) taps into processes in which both partners jointly apply pragmatic and emotion-focused coping strategies (i.e., jointly cope with stress). In the full sample, coefficients α of the three subscales were fair: α = .77 (positive dyadic coping), α = .73 (negative dyadic coping), and α = .75 (common dyadic coping).

Big Five Inventory-Short form (BFI-S; Gerlitz & Schupp, 2005). The BFI-S assesses the Big Five personality factors (Goldberg, 1993) with three items per factor: extraversion (e.g. “I see myself as someone who is outgoing, sociable.”), neuroticism (e.g. “I see myself as someone who worries a lot.”), conscientiousness (e.g. “I see myself as someone who does a thorough job.”), agreeableness (e.g. “I see myself as someone who has a forgiving nature.”) and openness to experience (e.g. “I see myself as someone who has an active imagination.”). All items were rated on a 7-point scale (1 = ‘does not apply to me at all’ to 7 = ‘applies to me perfectly’). Despite shortcomings for the agreeableness subscale, the subscales demonstrated acceptable levels of reliability, stability in the midterm, and convergent and discriminant validity (Hahn, Gottschling, & Spinath, 2012). In the present study, the coefficients α for extraversion (.79), neuroticism (.77), conscientiousness (.67), agreeableness (.48), and openness to experience (.66) were similar to Hahn et al. (2012) and acceptable for short three-item scales.

Relationship satisfaction (Grau, Mikula, & Engel, 2001). Relationship satisfaction was assessed with the global satisfaction subscale taken from the German adaptation of Rusbult’s Investment Model Scale (Grau et al., 2001). The five satisfaction items (e.g. “I feel satisfied with our relationship.”) were answered on a 5-point scale (1 = ‘completely disagree’ to 5 = ‘completely agree’). The reliability, structural and criterion validity of this scale have been shown in Grau et al. (2001). In the present full sample, Cronbach’s α was .90.

Statistical Analyses

First, we will report correlations between study variables. Second, we will test for relations between TEI and relationship satisfaction using the actor-partner-interdependence model (APIM; Kenny, Kashy, & Cook, 2006). The APIM takes into account the nonindependence of dyadic data and allows to estimate actor and partner effects (Kenny et al., 2006): An actor effect (path a in Figure 1, plot a) represents the association of one’s own predictor variable (e.g., TEI) with one’s own outcome variable (e.g., relationship satisfaction) controlling for the partner effect (path b) and the effect of relationship duration (path c). A partner effect (path b in Figure 1, plot a) represents the association of the partner’s predictor variable with one’s own outcome variable controlling for the actor effect and the effect of relationship duration.

Figure 1. Conceptual Depiction of the TEI-Relationship Satisfaction Actor-Partner Interdependence Model (Plot a) and of the TEI-Dyadic Coping-Relationship Satisfaction Actor-Partner Mediator Model (Plot b). Note. For simplicity, effects are supposed to be equal for males and females and subscripts M and F for individual paths (e.g., a_M and a_F) were therefore omitted.

Third, three actor-partner mediator models (APMeM; Ledermann & Bodenmann, 2006) were tested, which included either positive, negative, or common dyadic coping as mediator variables (for a conceptual depiction see Figure 1, plot b). In each APMeM, three actor effects (paths a, d, and g) and three partner effects (paths b, e, and h) were estimated along with the effects of relationship duration (paths f and c). As a result, each total indirect TEI-relationship satisfaction actor effect was composed of two specific indirect actor effects (an indirect actor-actor effect via the paths d and g and an indirect partner-partner effect via the paths e and h; see Figure 1, plot b). Similarly, each total indirect TEI-relationship satisfaction partner effect consisted of two specific indirect partner effects (an indirect partner-actor effect via the paths e and g and an indirect actor-partner effect via the paths d and h).

Fourth, overlap of total TEI with the Big Five factors was removed from the total TEI score by means of multiple regression analysis with the person as unit of analysis and saving the standardized TEI residuals.² Next, all models were re-estimated using the residualized TEI score as the predictor variable. Sixth, all models were tested again using the reduced TEI score based on 24 TEIQue-SF items.

All APIMs and APMeMs were tested in Mplus 7.0 (Muthén & Muthén, 2012) using a path analytical approach and the robust Satorra-Bentler (2001)-scaled SB-χ² statistic. Prior to the analyses, all variables were z-scored using the sample grand mean (all mixed variables) or the dyad mean (relationship duration). In the first step, actor and partner effects, and effects of relationship duration were freely estimated, resulting in just identified models. Next, actor effects, partner effects, and effects of relationship duration were stepwise forced to be equal across both sexes. If the model fit did not deteriorate (i.e., a non-significant SB-χ²-difference test), the more restricted model was preferred. This procedure consistently led to APIMs with three degrees of freedom and to APMeMs with eight degrees of freedom. The bootstrapped 95% bias-corrected confidence intervals of indirect effects were based on 10,000 resamples.

Results

Correlations of Study Variables

Within-person and between-person correlations of study variables are shown in Table 1. Two findings are noteworthy: First, the level of homogamy for TEI was low, r = .17, p = .051, which is consistent with prior research (e.g., Zeidner et al., 2013; Smith et al., 2008). Similar levels of both partners’ TEI seem to bear little relevance for the formation of romantic couples. Second, relationship duration was negatively related to dyadic coping variables and relationship satisfaction which warrants its inclusion as covariate in the following analyses.

Main Analyses

First, the actor effects and partner effects for TEI on relationship satisfaction, while controlling for the effects of relationship duration were estimated. The restricted APIM (i.e., all effects were set equal across both sexes) fitted perfectly to the data, SB-χ²(3) = 1.72, p = .63 (for results see Table 2, upper section). Together, TEI and relationship duration explained 22.1% of females’ and 22.9% of males’ variance in relationship satisfaction. The actor effect (β = .29, p < .001) and the partner effect (β = .18, p < .001) were both significant. Setting the actor effect and partner effect equal worsened the model fit, SB-Δχ²(df = 1) = 3.97, p = .046, implying that the actor effect was significantly stronger than the partner effect.

Table 2. Path Coefficients and Model-Fit for the Trait Emotional Intelligence-Relationship Satisfaction APIMs and for the Trait Emotional Intelligence-Dyadic Coping-Relationship Satisfaction APMeMs Depicted in Figure 1.

		Correlated uniqueness and standardized path coefficients										R^2SAT		Fit statistic
Model	Mediator	r4	r5	path a	path b	path c	path d	path e	path f	path g	path h	females	males	df	robust χ^2	p-value
Models include total TEI as exogenous variable
APIM	n.a.	.55***	–	.29***	.18***	−.26**	–	–	–	–	–	.22***	.23***	3	1.72	.63
APMeM	PDC	.49***	.23**	.23***	.12*	−.16*	.30***	−.03	−.24***	.23***	.22**	.32***	.34***	8	8.05	.43
	CDC	.48***	.36***	.19***	.10*	−.17*	.25**	.11	−.18*	.35***	.15**	.37***	.39***	8	8.11	.42
	NDC	.51***	.24**	.25***	.15*	−.22**	−.15*	−.06	.10	−.19***	−.15***	.29***	.29***	8	2.47	.96
Models include residualized TEI (i.e., overlap with the Big Five factors was partialled out prior to the analyses) as exogenous variable
APIM	n.a.	.58***	–	.19**	.16**	−.27**	–	–	–	–	–	.17**	.16**	3	2.27	.52
APMeM	PDC	.49***	.19*	.16**	.13*	−.14*	.10	.04	−.26***	.28***	.22***	.31***	.30***	8	7.00	.54
	CDC	.48***	.40***	.13**	.13**	−.15*	.14*	.01	−.20**	.39***	.16***	.37***	.37***	8	7.23	.51
	NDC	.52***	.25**	.19***	.16**	−.21**	−.01	−.02	.12	−.23***	−.18***	.28***	.26***	8	3.93	.86
Models include reduced TEI (i.e., six TEIQue-SF items organized under the well-being factor were excluded) as exogenous variable
APIM	n.a.	.55***	–	.27***	.19***	−.27**	–	–	–	–	–	.22***	.22***	3	1.09	.78
APMeM	PDC	.48***	.22**	.20***	.14*	−.16*	.29***	−.02	−.25***	.24***	.21***	.32***	.33***	8	6.63	.58
	CDC	.48***	.36***	.17**	.12*	−.17**	.24***	.11	−.19*	.36***	.14**	.37***	.39***	8	7.54	.48
	NDC	.51***	.24**	.23***	.16**	−.23**	−.15**	−.04	.11	−.19***	−.16**	.29***	.29***	8	1.85	.99

Note. APIM = actor partner interdependence model (for a conceptual depiction see Figure 1, plot a); APMeM = actor-partner mediator model (for a conceptual depiction see Figure 1, plot b); PDC = positive dyadic coping, CDC = common dyadic coping, NDC = negative dyadic coping, TEI = trait emotional intelligence, SAT = relationship satisfaction; path a = actor effect _{TEI → SAT}; path b = partner effect _{TEI → SAT}; path c = effect _{relationship duration → SAT}; path d = actor effect _{TEI → dyadic coping}; path e = partner effect _{TEI → dyadic coping}; path f = effect _{relationship duration → dyadic coping}; path g = actor effect _{dyadic coping → SAT}; path h = partner effect _{dyadic coping → SAT}; r₄ = correlated uniqueness _{SAT, SAT}; r₅ = correlated uniqueness _{dyadic coping, dyadic coping} (the correlations r₁ to r₃ between male’s TEI, female’s TEI, and relationship duration are given in Table 1). In all models, the respective actor effects, partner effects, and effects of relationship duration were fixed to be equal across both sexes. *p < .05, **p < .01, ***p < .001.

Next, three actor partner mediation models that included positive dyadic coping, common dyadic coping, or negative dyadic coping were tested (for results see Table 2, upper section). When restricted models were specified (i.e., all effects were set equal across both sexes), all three restricted APMeMs showed a perfect model fit (SB-χ²[8] ranged from 2.47 to 8.11, all ps ≥ .42). In all three models, 28.9% to 36.8% (females) and 29.4% to 39.4% (males) of the variance in relationship satisfaction was explained. Except that all significant actor and partner effects that included negative dyadic coping were reversed in sign, the pattern of effects was similar across the models: Significant actor effects and partner effects on relationship satisfaction were found for positive dyadic coping and common dyadic coping (positive effects) and for negative dyadic coping (negative effects; paths g and h in Figure 1, plot b; for results see Table 2). In all three models, TEI showed positive direct significant actor effects and partner effects on relationship satisfaction (paths a and b in Figure 1, plot b). However, these direct effects were attenuated when compared to the corresponding total effects in the APIM. Moreover, TEI’s actor effects on dyadic coping were significant in all three models as well, while TEI’s partner effects were non-significant (paths d and e in Figure 1, plot b; for results see Table 2). Interestingly, TEI’s actor effect on negative dyadic coping was rather small (β = -.15, p = .016), which impaired its utility as a mediator.

The obtained total indirect effects, specific indirect effects, and respective bootstrapped 95%-BC-CIs for the three APMeMs are shown in Table 3 (upper section). In all three models the 95%-BC-CIs of the total indirect actor effects precluded zero implying significance, while significant total indirect partner effects were found only for positive dyadic coping and common dyadic coping. Moreover, significant specific indirect effects were also found for positive dyadic coping and common dyadic coping, but not for negative dyadic coping. The latter finding is likely due to TEI’s relatively small actor effect on negative dyadic coping. All four significant specific indirect effects build upon TEI’s actor effect on either positive dyadic coping or common dyadic coping (i.e., indirect actor-actor or actor-partner effects). Trait EI’s nonsignificant partner-effects on dyadic coping prevented significant indirect partner-partner or partner-actor effects occurring. In sum, each significant yet attenuated direct actor effect and direct partner effect of TEI on relationship satisfaction and the significant total and specific indirect effects via actor-actor and actor-partner pathways imply partial mediation via positive dyadic coping or via common dyadic coping. Negative dyadic coping mediated only a small significant portion of TEI’s actor effect on relationship satisfaction when both non-significant specific indirect actor effects were summed up to form the total indirect actor effect.

Table 3. Unstandardized Total and Specific Indirect Effects and Bootstrapped 95% BC-Confidence Intervals for the Trait Emotional Intelligence -Dyadic Coping-Relationship Satisfaction Actor-Partner Mediator Models Shown in Figure 1 (Plot b) and in Table 2.

	Indirect TEI-SAT actor effects			Indirect TEI-SAT partner effects
Mediator	Total indirect effect	Actor-actor effect	Partner-partner effect	Total indirect effect	Actor-partner effect	Partner-actor effect
Models include total TEI as exogenous variable
PDC	.064 [.014, .114]	.070 [.029, .111]	−.006 [−.039, .020]	.059 [.011, .108]	.065 [.024, .107]	−.006 [−.033, .021]
CDC	.102 [.044, .160]	.087 [.038, .135]	.015 (−.006, .036]	.073 [.015, .131]	.036 [.004, .067]	.037 [−.008, .083]
NDC	.038 [.001, .075]	.028 [−.001, .057]	.010 [−.009, .028]	.035 [−.003, .073]	.023 [−.003, .048]	.012 [−.009, .033]
Models include residualized TEI (i.e., overlap with the Big Five factors was partialled out prior to the analyses) as exogenous variable
PDC	.035 [−.011, .080]	.027 [−.012, .065]	.008 [−.020, .037]	.032 [−.013, .076]	.021 [−.010, .053]	.010 [−.026, .046]
CDC	.056 [−.002, .114]	.055 [.007, .104]	.001 [−.021, .023]	.025 [−.034, .085]	.023 [−.002, .048]	.002 [−.048, .052]
NDC	.007 [−.029, .044]	.005 [−.023, .033]	.003 [−.016, .022]	.007 [−.028, .042]	.004 [−.018, .026]	.003 [−.020, .027]
Models include reduced TEI (i.e., six TEIQue-SF items organized under the well-being factor were excluded) as exogenous variable
PDC	.065 [.017, .113]	.070 [.029, .111]	−.005 [−.028, .019]	.055 [.008, .101]	.060 [.020, .100]	−.006 [−.033, .022]
CDC	.101 [.043, .159]	.085 [.037, .134]	.016 [−.006, .037]	.074 [.014, .133]	.034 [.003, .065]	.040 [−.007, .087]
NDC	.036 [.000, .072]	.030 [.001, .058]	.006 [−.011, .024]	.032 [−.005, .068]	.024 [−.001, .049]	.008 [−.012, .028]

Note. PDC = positive dyadic coping; CDC = common dyadic coping; NDC = negative dyadic coping; TEI = trait emotional intelligence; SAT = relationship satisfaction. Actor–actor effect = indirect effect via path d (actor effect _{TEI → dyadic coping}) and path g (actor effect _{dyadic coping → SAT}); Partner–partner effect = indirect effect via path e (partner effect _{TEI → dyadic coping}) and path h (partner effect _{dyadic coping → SAT}); Actor–partner effect = indirect effect via path d (actor effect _{TEI → dyadic coping}) and path h (partner effect _{dyadic coping → SAT}); Partner-actor effect = indirect effect via path e (partner effect _{TEI → dyadic coping}) and path g (actor effect _{dyadic coping → SAT}). Estimates of the upper and lower levels of the 95% BC-CI’s were based on 10,000 bootstrap resamples; significant effects (i.e., the 95%-BC-CI precludes zero) are printed in bold.

Analyses for the Exploratory Research Questions

When total TEI was regressed on the Big Five factors, all Big Five factors were specifically related to TEI: neuroticism, β = -.40, conscientiousness, β = .23, agreeableness β = .30, extraversion, β = .21, and openness to experience, β = .20 (all ps < .001). The Big Five factors jointly accounted for 50.2% of the variance in TEI, F(5,266) = 53.55, p < .001.

All APIM and APMeM analyses were repeated with the residualized TEI score (for results see Tables 2 and 3, middle sections). In the APIM analysis, residualized TEI’s actor effect (β = .19, p = .002) and partner effect (β = .16, p = .007) remained significant, but were attenuated in size. Thus, TEI accounted for incremental variance in relationship satisfaction over and above the Big Five factors. In the APMeM analyses, the only significant actor effect of the residualized TEI score on dyadic coping was found for common dyadic coping (β = .14, p = .012). As a consequence, the only significant specific indirect effect was observed via the actor-actor pathway that included common dyadic coping as mediator (see Table 3, middle section). Taken together, these predominant nil-findings imply that the previously observed effects of total TEI on dyadic coping and the mediation effects are mainly driven by variance that overlaps with the broader Big Five factors.

Finally, the models were re-run with the reduced TEI score that omitted all six well-being items. The utilization of the reduced TEI score in the APIM and in the three APMeMs revealed only negligible differences to the effects obtained for the total TEI score (see Tables 2 and 3, lower sections). These findings suggest that TEI’s effects are not severely biased by potential content overlap between relationship satisfaction and TEI’s well-being factor.

Discussion

Trait EI is an important individual difference variable that concerns how people perceive their own socio-emotional effectiveness and that is related to various psychological criteria (e.g., Andrei et al., 2016; Malouff et al., 2014; Martins et al., 2010; Sánchez-Álvarez et al., 2016). Using a cross-sectional dyadic design, this study investigated TEI’s associations with self-reported dyadic coping and relationship satisfaction in heterosexual romantic relationships. This study added to the literature by using a comprehensive measure of TEI, by considering three strategies of dyadic coping, by controlling for the overlap between TEI and the Big Five traits, and by considering potential content overlap between TEI and relationship satisfaction.

In the APIM (Kenny et al., 2006) analysis, significant and positive actor and partner effects of TEI on relationship satisfaction were found (H1a and H1b confirmed). Size and direction of both effects are consistent with meta-analytical findings (Malouff et al., 2014), which imply that emotion related dispositions bear relevance for the quality of relationships. The actor effect was significantly stronger than the respective partner effect. This difference might be due to shared method variance in self-reports, which might have inflated the actor effect. This difference might alternatively be due to substantive relational processes: As in the present study, one’s own TEI is more strongly related to one’s own adaptive processes than partners’ TEI, and one’s own adaptive processes tend to be more strongly associated with one’s own relationship goals and satisfaction than the adaptive processes deployed by the partner (e.g., Smith et al., 2008; Zeidner et al., 2013), which results in a stronger actor effect.

The partner effect supports the claim that TEI has interpersonal implications in romantic relationships. It complements prior findings that better TEI is related to better peer-relations and to more positive peer nominations in adolescents (Frederickson et al., 2012; Mavroveli et al., 2007). It might thus be advantageous to one’s own relationship satisfaction to select a trait emotional intelligent partner. However, TEI’s significant partner effect on relationship quality did not consistently show up in prior research (e.g., Smith et al., 2008; Zeidner et al., 2013), and the cause of this inconsistency is currently unclear. It might be that the TEI inventories employed have contributed to this inconsistency, as more comprehensive TEI assessments such as the TEIQue (Petrides, 2009) have higher criterion validity (Gardner & Qualter, 2010). However, the identification of the mechanisms that contribute to this inconsistency (e.g., insufficient power, different operationalization of TEI, relationship development stage, cultural differences) seems to be puzzling and more dyadic research is needed in this regard.

Significant actor effects for TEI on self-perceived positive and common dyadic coping (positive effects) and on negative dyadic coping (negative effect) were also found (H2a, H2b, and H2c confirmed), and the latter effect tended to be weaker than the former. Interestingly, the total overlap of negative dyadic coping with the Big Five factors tended to be smaller as well (for within-person correlations see Table 1), which complemented the findings for TEI and point to a comparatively larger discriminability of traits and negative dyadic coping. Nevertheless, the results are consistent with the VSA-model (Karney & Bradbury, 1995) and they add to the previous finding of a TEI-total dyadic coping actor effect (Zeidner et al., 2013). Significant partner effects did not emerge (EQ1), which is hardly surprising given that TEI and individual coping were rarely related to partner’s dyadic coping in the previous literature as well (Herzberg, 2012; Papp & Witt, 2010; Zeidner et al., 2013). However, the positive residual correlations found for dyadic coping are suggestive for mutual contagion effects that arise, when a partner utilizes a specific dyadic coping strategy more often. The findings therefore suggest that the utilization of positive, negative, and common dyadic coping is associated with one’s own emotion-related dispositions and with the dyadic coping strategy utilized by the partner.

Finally, in the respective APMeM (Ledermann & Bodenmann, 2006) analyses, positive dyadic coping and common dyadic coping partially mediated the TEI-relationship satisfaction actor effect through the specific indirect actor-actor effect; they also partially mediated the TEI-relationship satisfaction partner effect through the specific actor-partner effect (H3a and H3b confirmed). Thus, one’s own TEI was positively related to one’s own use of adaptive dyadic coping strategies, which in turn were positively related to one’s own and to the partner’s relationship satisfaction. For negative dyadic coping, only a significant total indirect actor effect emerged (H3c partially confirmed). The results refine previous results for mediation via total dyadic coping (Zeidner et al., 2013): Compared with hostile, reluctant, insufficient, and inefficient dyadic support, the supportive and delegated dyadic coping as well as the joint utilization of emotion-focused and pragmatic coping strategies appear to be more effective mediators of TEI’s actor and partner effects on relationship satisfaction. Future research could test the adaptive and maladaptive dyadic coping strategies as parallel mediators against each other, using larger samples with more power to detect smaller specific indirect effects. Moreover, in all three APMeM analyses, TEI’s direct effects on relationship satisfaction remained significant. The observed partial mediation suggests the presence of adaptive processes other than dyadic coping possibly including perspective taking (Schröder-Abé & Schütz, 2011), conflict resolution styles (Smith et al., 2008), appraisals of interactions, or reciprocity. These variables might be included in the models as parallel mediators in future research.

By definition, TEI overlaps with the Big Five personality factors (Petrides et al., 2007), and recent meta-analytical evidence suggests that TEI is incrementally valid beyond the Big Five factors for various criteria (Andrei et al., 2016). Does this also hold for relationship satisfaction and dyadic coping (EQ2)? The preliminary answer for relationship satisfaction is positive: When overlap with the Big Five factors has been removed from the total TEI score, both actor and partner effects of residualized TEI on relationship satisfaction were significant, albeit attenuated in size. This finding extends the findings reported by Joshi and Thingujam (2009) who controlled for three Big Five factors and considered a within-person correlation. However, the incremental validity of TEI tends to be higher when the Big Five traits are assessed with short or medium-sized scales (Andrei et al., 2016). In this study, the brief BFI-S scales (Gerlitz & Schupp, 2005) accounted for about 50% of the variance in total TEI, which acceptably resembles the overlap of total TEI with longer Big Five scales (e.g., Petrides et al., 2010). Nevertheless, the incremental validity of TEI might still be slightly inflated.

However, the preliminary answer for dyadic coping is mainly negative: In three APMeM analyses, only the actor effect of residualized TEI on common dyadic coping and subsequently only the indirect actor-actor effect via common dyadic coping on relationship satisfaction became significant. This implies that the TEI-dyadic coping associations were mainly driven by variance shared with the Big Five traits. Although TEI is incrementally valid beyond the Big Five traits when individual coping is taken as a criterion (e.g., Andrei et al., 2016), this seems not to generalize to self-perceived dyadic coping. In fact, the current data suggest that for self-perceived dyadic coping the incremental validity for TEI vis-à-vis the Big Five is rather low. However, one might speculate that the significant direct actor and partner effects of residualized TEI on relationship satisfaction point to the existence of other unmeasured adaptive mechanisms that might account for these incremental effects and for which TEI might thus be incrementally valid. This needs to be tested in future research.

Finally, critiques might be concerned about a potential content overlap between the TEIQue-SF (e.g., items assessing trait happiness) and items assessing relationship satisfaction (e.g., Petrides et al., 2007). However, when a reduced TEI score was created with all six TEI items belonging to TEI’s well-being factor removed, changes in the obtained effects remained negligible. Thus, the bias introduced by a potential content overlap seems to be negligible.

Implications

The current study provides further evidence that TEI is positively related to adaptive processes and to the quality of romantic relationships. As a consequence, increasing TEI might be a focal target in couple counseling. However, TEI is defined as a stable trait (Petrides, 2009), which may limit its potential for cultivation through EI training. A recent meta-analysis showed that EI trainings actually increase EI, that ability EI is less resistant to change than mixed EI or TEI, and that improvements are maintained over time (Hodzic, Scharfen, Ripoll, Holling, & Zenasni, 2018). Given that mixed EI or TEI is amenable to some change which in turn likely improves the relationship quality (Kotsou, Nelis, Grégoire, & Mikolajczak, 2011), it seems promising to implement EI training in couple counseling (for an exemplary training schedule see Kotsou et al., 2011) and to test its effects on adaptive processes, relationship quality, and relationship stability.

Moreover, TEI is reliably related to mental and physical health (Martins et al., 2010), subjective well-being (Sánchez-Álvarez et al., 2016), and being in a high-quality romantic relationship is, in turn, related to well-being and health (Proulx et al., 2007; Robles et al., 2014). It might thus be revealing to show whether portions of the associations between TEI, well-being, and health are mediated via adaptive dyadic processes and relationship quality.

Limitations

The present study has several limitations. First, the cross-sectional nature of the data prevents any causal claims. The sequence of variables in the APIM and APMeMs was guided by theory (Karney & Bradbury, 1995), as stable traits affect less stable constructs such as adaptive processes and relationship quality more likely than vice versa. However, other models might be consistent with the data as well. Moreover, unmeasured variables might alternatively account for the observed effects. A second limitation of the current study is the reliance on self-reports. Common rater effects might have biased the within-person associations (e.g., mood state, consistency motif, social desirability). However, Falconier et al. (2015) showed that dyadic coping perceived by self is usually more weakly related to relationship quality than perceptions of dyadic coping by the partner and by both partners together. Given that the current study utilized only perceptions of dyadic coping by self, the observed effects might actually underestimate the overall utility of dyadic coping as a mediating mechanism for the TEI-relationship satisfaction association. Thus, more studies on TEI, adaptive processes, and relationship quality including multiple sources of information (e.g., self- and partner-perceptions), different operationalization of the involved variables, and different designs (e.g., longitudinal, interventional) are needed to further elaborate the effects of TEI on intimate relationships. Third, the observed effects may not generalize to same-sex couples or to couples in non-western cultures. In fact, little is known about the effects of TEI in same-sex and in non-western romantic relationships (e.g., Malouff et al., 2014, included only one study with non-western couples). Fourth, it cannot be ruled out that the observed associations vary across stages of relationship development and increasing interdependence or with the severity and endurance of stressors (e.g., daily dyadic adversities, transition to parenthood, bereavement, unemployment). The present sample was too small for analyses within stratified subsamples. Thus, more research is needed with larger or more targeted samples to test for potential interactions with relationship status or kinds of stressors.

Despite these limitations, the current study advances our understanding of how TEI is associated with both partners’ relationship satisfaction and dyadic coping. The results support the status of TEI as an important individual difference variable that bears relevance for various psychological criteria in general and for the quality of romantic relationships in particular.

Additional information

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sections.

Notes on contributors

Anna Wollny

Anna Wollny, PhD, is a staff psychologist at University of Potsdam, Germany. Her research interests include trait emotional intelligence, resilience, mental health, and test construction.

Ingo Jacobs

Ingo Jacobs, PhD, is a psychologist and cognitive behavioral therapist working as lecturer at Medical School Berlin and at Sigmund Freud University Berlin, Germany. His research interests include trait emotional intelligence, mindfulness, resilience, and personality disorders.

Luise Pabel

Luise Pabel, certified clinical psychologist and psychotherapist, is a psychology doctoral student at University of Dresden, Germany and works as psychotherapist and scientific researcher at the University Hospital Dresden. Her research interests include, among others, the relation between olfaction and depressive disorders.

Notes

1 Translations of the German DCI example items were carried out by the authors.

2 Nonindependence of the trait scores of both dyad members was low (see Table 1). Such low nonindependence of mixed variables implies that the scores of both dyad members are only weakly interrelated, and such weak violation of independence of observations causes negligible bias when the person is used as the unit of analysis (Kenny, Kashy, & Bolger, 1998).