The role of relationship dissatisfaction in the dyadic associations between attachment insecurity and intimate partner violence among couples seeking therapy.

For several decades, attempts have been made to understand the risk markers of intimate partner violence (IPV) to prevent its occurrence, but few studies have emphasized both partners' perspective in examining these markers. This study explored the associations between attachment insecurity (anxiety and avoidance) and the perpetration of psychological and physical IPV through relationship dissatisfaction in a sample of 88 couples seeking therapy for relationship difficulties. Results from path analyses based on the Actor-Partner Interdependence Model showed that participants' attachment avoidance was indirectly related to their own and their partner's higher perpetration of psychological IPV through higher relationship dissatisfaction. Participants' attachment avoidance was also associated with their own higher perpetration of physical IPV through their higher relationship dissatisfaction. The findings highlight that attachment insecurities and relationship dissatisfaction can both contribute to establishing dysfunctional and violent interaction patterns in couples.

INTRODUCTION

The study of intimate partner violence (IPV) and its risk markers has interested a number of researchers, primarily due to its detrimental effects on the well‐being of individuals and families (Hellemans et al., 2015; Katz, 2016). However, the examination of IPV using a dyadic perspective, which allows us to consider both partners' processes and predisposing factors as reciprocally influencing one another (Bartholomew & Cobb, 2011; Bates, 2016) is recent. Indeed, most studies examining IPV continue to adopt an individual standpoint, either emphasizing risk factors in perpetrators, or repercussions in victims (Johnson et al., 2020). Unfortunately, this individual approach does not help practitioners and researchers understand cases of IPV resulting from dysfunctional dynamics within the couple such as the escalation of conflicts (Aaron & Beaulaurier, 2017; Hamel, 2007). In addition, many individuals seeking therapy for relationship difficulties report that the violence within their relationship is bidirectional (Hamel, 2012; Madsen et al., 2012), which cannot be fully understood using an individual perspective. More recently, results from a few dyadic studies have supported the relevance of considering both partners when examining IPV (e.g., Dugal et al., 2019; Sommer et al., 2017). As it allows to account for the influence of both partners, attachment theory offers a comprehensive framework to understand how dysfunctional couple dynamics sometimes lead to IPV. This theoretical framework might therefore be useful for examining psychological and physical violence within couples (Godbout et al., 2017; Spencer et al., 2020). This study relies on attachment theory to examine the role of relationship dissatisfaction in the associations between attachment insecurity and IPV, using a dyadic approach.

An attachment perspective on intimate partner violence

Attachment theory (Bowlby, 1969/1982) posits that children form internal models of self and others based on how their attachment figures (e.g., a parent) responded to their needs for protection, emotional connection, and security during childhood. These internal models remain relatively stable over time and across personal relationships, including in adult romantic relationships where the partner becomes the primary attachment figure (Feeney, 2016; Hazan & Shaver, 1987). Adult attachment can be conceptualized as including two continuous dimensions: attachment anxiety and avoidance (Brennan et al., 1998). Attachment anxiety is characterized by a negative model of the self as unworthy of love. Individuals who report a high level of attachment anxiety tend to be hypervigilant to any signs of rejection and potential abandonment from their romantic partner. When they perceive that their relationship is potentially threatened, their attachment system is activated, they are more prone to use hyperactivation strategies, such as constantly seeking reassurance and closeness from their partner. Attachment avoidance is defined by a negative model of others. Individuals who report high levels of attachment avoidance tend to have an excessive need for independence and are uncomfortable with emotional intimacy. When they perceive that their need for independence is threatened, they tend to use deactivation strategies to reduce their sense of vulnerability, such as repressing their feelings and maintaining a distance from their partner (Bartholomew & Horowitz, 1991; Brennan et al., 1998; Mikulincer & Shaver, 2016).

From an attachment perspective, violence in relationships can be understood as an excessive form of protest behavior that occurs when a partner's availability or responsiveness is perceived as either lacking (attachment anxiety) or as smothering (attachment avoidance; Mikulincer & Shaver, 2016). As a result, individuals may use aggression towards their partner as a way to either get their attention, seek closeness, and increase an emotional connection, or to push them away, establish emotional distance, and increase their independence (Brassard et al., 2014; Godbout et al., 2017). Individuals with a greater level of attachment anxiety may prefer the negative attention received through conflict to a perceived emotional distance leading them to tolerate violence in order to maintain a level of closeness (Mikulincer & Shaver, 2016). Detachment and lack of nurturance of individuals who report high levels of attachment avoidance could provoke their partner, especially if they are demanding or dependent, to use violence to gain attention or love (Mikulincer & Shaver, 2016). Several studies have supported these claims by highlighting that attachment anxiety and avoidance are positively related to psychological (the use of verbal and non‐verbal communication intended to hurt or exert control over the partner; Breiding et al., 2015) and physical (the intentional use of brutal or excessive physical acts that harm the physical and psychological integrity of the romantic partner; Breiding et al., 2015) IPV perpetration and victimization (for reviews, see Mikulincer & Shaver, 2016; Spencer et al., 2020; Velotti et al., 2020). Yet, it remains unclear how exactly attachment insecurities increase a couple's risk of experiencing psychological and physical IPV. As suggested by Fournier and colleagues (2011), relationship dissatisfaction could help explain how attachment insecurities are related to IPV perpetration.

The role of relationship dissatisfaction

Relationship satisfaction is defined by the subjective evaluation of the quality of one's relationship at a given time (Bradbury et al., 2000; Sabourin et al., 2005). As put forward by Bartholomew and Cobb (2011), both partners' personal characteristics or vulnerabilities (e.g., attachment insecurities) tend to interact in ways that foster specific dyadic contexts (e.g., relationship dissatisfaction) in which dysfunctional relational patterns emerge and can lead to increasingly aggressive interactions. For instance, insecure individuals' expectations of others as potentially rejecting or smothering might lead to pessimistic expectations about their partner and their relationship, which may increase their own and their partner relationship dissatisfaction. Indeed, attachment insecurities have been associated with lower relationship satisfaction—or higher relationship dissatisfaction—in both partners, with attachment avoidance being more strongly associated with higher relationship dissatisfaction than attachment anxiety (for reviews, see Candel & Turliuc, 2019; Feeney, 2016; Mikulincer & Shaver, 2016). In turn, dyadic contexts in which attachment insecurities are activated and in which couples are dissatisfied, are typically marked by more hostility, criticism, contempt as well as poor communication and problem‐solving skills (Bartholomew & Cobb, 2011; Johnson et al., 2005). These dysfunctional behaviors, in addition to being potentially harmful to their partner's relationship satisfaction, are known to increase the risk of perpetrating and sustaining IPV (Mikulincer & Shaver, 2016). Studies have indeed documented higher relationship dissatisfaction in couples who report IPV (Gewirtz & Finzi, 2020; Panuzio & DiLillo, 2010; Spencer et al., 2020).

Although several studies have found that attachment insecurities and relationship dissatisfaction are related to the experience of IPV, very few studies have examined these associations using a dyadic design (Bartholomew & Cobb, 2011; Fournier et al., 2011). To our knowledge, only two studies have explored the role of relationship satisfaction in the associations between attachment insecurity and IPV (Fournier et al., 2011; Gou & Woodin, 2017). Fournier and colleagues (2011) found that men's lower relationship satisfaction explained the link between men's attachment anxiety and their perpetration of psychological IPV, whereas Gou and Woodin (2017) showed that relationship satisfaction explained the association between men's attachment insecurity (anxiety and avoidance) and their psychological IPV perpetration. However, Fournier and colleagues (2011) used a small sample of 55 men in treatment for relationship difficulties, including aggression, and did not consider both partners' perspectives, and Gou and Woodin (2017) assessed psychological violence among a very specific sample of couples expecting their first child, but without using a dyadic analytic strategy. In addition, many studies that have examined the links between attachment and IPV have only focused on one form of violence (e.g., Gou & Woodin, 2017) without comparing whether risk markers vary according to the form of IPV experienced by couples. Indeed, authors have stressed the need to distinguish between the forms of violence to account for different factors involved in psychological and physical violence (e.g., Sommer et al., 2017; Velotti et al., 2018). To date, however, no study has examined the associations between attachment and psychological and physical IPV perpetration among couples seeking therapy for relationship difficulties. Yet, IPV prevalence rates for couples in clinical samples are higher than those of couples from the community. Indeed, recent studies have found that psychological IPV is reported by 57.0% of women and 64.0% of men from the general population (Dugal et al., 2019), and by 80.5% of women and 74.8% of men seeking couple therapy (Tougas et al., 2016), and that physical IPV is reported by 33.7% of women and 30.2% of men from the general population (Panuzio & DiLillo, 2010) and 73% of women and 78% of men seeking therapy for IPV (Madsen et al., 2012). Considering that clinical samples of couples who experience significant relationship distress tend to also experience high instances of IPV (Tougas et al., 2016), these shortcomings in the literature need to be addressed.

Objectives and hypotheses

This study aims to explore whether attachment insecurity (anxiety and avoidance) is associated with the perpetration of psychological and physical IPV by both partners through higher levels of relationship dissatisfaction among couples seeking therapy. Based on previous studies, it was hypothesized that attachment insecurity (anxiety, avoidance) would be associated with higher (a) psychological and (b) physical IPV perpetration directly (actor and partner direct effects) and indirectly through both partners' relationship dissatisfaction (actor and partner indirect effects).

METHOD

Participants and procedure

The sample consisted of 88 couples seeking therapy for relationship difficulties who had been in their relationship for an average of 8.7 years (SD = 1.00) and were either cohabiting (64.0%), living separately (14.0%), or married and living together (21.0%). On average, women were aged 34.7 years (SD = 10.06) whereas men were aged 38.5 years (SD = 10.39), and women had 1.7 children (SD = .14) whereas men had 1.6 children (SD = .15). Women had completed elementary school (13.4%), high school (68.3%), or held a postsecondary diploma (18.3%). Men had completed elementary school (23.7%), high school (67.1%), or held a postsecondary diploma (9.2%).

Couples were recruited through two community‐based clinics providing couple and partner violence services. Before their first therapy session, partners were invited by their practitioner to complete a series of paper‐pencil questionnaires as part of the assessment procedure in both organizations. To take part in this study, participants had to be 18 years or older, be in a relationship for at least 12 months, and understand French. Participants were living in the province of Quebec, Canada. This research was approved by the research ethics committee of the researchers' institution.

Measures

Measures for this study were selected based on their psychometric properties and were presented to participants in their French validated version. Alpha coefficients for all measures are shown in Table 1. Participants completed a sociodemographic questionnaire assessing personal and relationship information (e.g., age, gender, relational status, duration of the relationship, number of children, education).

TABLE 1

Descriptive statistics, Cronbach's alpha coefficients, and Pearson correlations for the main variables (N = 88 couples)

	1	2	3	4	5	6	7	8	9	10
1. Attachment anxiety W	—
2. Attachment avoidance W	.089	—
3. Attachment anxiety M	−.081	.169	—
4. Attachment avoidance M	.197	.369**	.164	—
5. Relationship dissatisfaction W	.156	.558**	.245*	.442**
6. Relationship dissatisfaction M	.171	.376**	.217*	.572**	.450**	—
7. Psychological IPV W	.211*	−.050	.132	.036	.224*	.174	—
8. Psychological IPV M	.133	−.046	.250*	−.001	.210*	.136	.761**	—
9. Physical IPV W	−.105	−.013	.008	−.063	.064	−.003	.411**	.425**	—
10. Physical IPV M	−.051	−.025	.030	−.143	.124	.002	.387**	.526**	.765**	—
M	4.18	2.74	3.66	2.65	9.98	8.51	46.90	49.52	13.85	12.18
SD	1.12	1.07	.99	1.06	3.56	3.23	30.45	35.32	35.36	28.15
Skewness	−.438	.311	−.009	.470	−.119	−.044	.446	.525	4.95	4.76
Kurtosis	.205	−.613	−.310	−.100	−.311	−.751	−.177	−.122	28.36	27.39
Cronbach's α	.82	.82	.70	.84	.75	.69	.72	.71	.84	.82

Abbreviations: W, women. M, men. IPV, intimate partner violence.

p < .05.

p ˂ .01.

Attachment insecurity

Attachment‐related anxiety and avoidance were measured by the 12‐item short version of the Experiences in Close Relationships scale (ECR‐12: Lafontaine et al., 2016). Participants indicated their level of agreement with each item using a 7‐point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Two scores were computed by averaging the items included in each subscale; high scores indicated a higher level of anxiety (e.g., “I worry about being abandoned”) or avoidance (e.g., “I don't feel comfortable opening up to romantic partners”), respectively. The ECR‐12 was validated using five samples, including French‐Canadian couples seeking help for relationship difficulties, and confirmatory factor analyses supported its bi‐dimensional structure, which evidenced good internal consistency for anxiety (α = .78–.87) and avoidance (α = .74–.83; Lafontaine et al., 2016).

Relationship dissatisfaction

Relationship dissatisfaction was measured using the short four‐item validated version of the Dyadic Adjustment Scale (DAS‐4; Sabourin et al., 2005). The items (e.g., “In general, how often do you think that things between you and your partner are going well?”), measured on a 6 or 7‐point scale, were reversed and summed to indicate the extent to which participants reported relationship dissatisfaction. Reversed scores ranged from 0 to 21, with a higher score indicating greater relationship dissatisfaction. The DAS‐4 is correlated with the 32‐item original version of the DAS (r = .94) and its predictive validity is supported by a longitudinal study of marital dissolution (Sabourin et al., 2005).

Intimate partner violence

To assess the frequency of perpetrated psychological and physical violence within the last year, the Revised Conflict Tactics Scales (CTS2; Straus et al., 1996) was administered. Participants reported the frequency of violent behaviors that they and their partner used in the previous year on an 8‐point scale (“This has never happened,” “Once,” “Twice,” “3–5 times,” “6–10 times,” “11–20 times,” “21 or more times,” and “Not in the past year, but this has happened before”). A total of 8 items assessed psychological violence (e.g., “I insulted or swore at my partner”), and 12 items assessed physical violence (e.g., “I punched or hit my partner with something that could hurt”). As directed by Straus and Douglas (2004), the midpoints of the rating categories (e.g., for “3–5 times” the number 4 was used, for “21 or more times” the number 25 was used) were summed to create the participants' scores for psychological and physical IPV, which indicates the annual frequency of behaviors each partner reported having used (perpetration score) and sustained (victimization score) in the past year. To create our final perpetration scores for each partner, we used the method of the maximum dyadic report, whereby the higher of the two self‐report scores provided was used as the measure of psychological and physical IPV perpetration (e.g., the highest score among self‐reported perpetration by men and their female partner's report of victimization; Cuenca et al., 2015; Straus et al., 1996). In our sample, interobserver reliability (i.e., the agreement between behaviors that a partner reported perpetrating and behaviors that the other partner reported sustaining) varied from .78 to .97 for items pertaining to psychological violence and from .80 to .97 for items pertaining to physical violence.

Data analysis

Data were first screened for missing values, normality, and extreme values. Single imputation in SPSS was used to estimate missing values for those who failed to answer certain questions (less than 5% of data set, missing at random) allowing to maximize statistical power (McKnight et al., 2007). According to Ackerman et al., (2015), a sample of 88 couples allows the estimation of medium (β = .20) actor and partner direct effects with a statistical power of .79 at an alpha level of .05. To maximize power, two separate dyadic path analyses based on the Actor‐Partner Interdependence Mediation Model (APIMeM, Ledermann et al., 2011) were conducted with Mplus using the robust maximum likelihood estimator to examine the associations among attachment insecurity (anxiety, avoidance) and psychological IPV perpetration (followed by physical IPV) through relationship dissatisfaction. As suggested by Hayes (2009), a one‐step bootstrapping approach was used to test the direct and indirect effects simultaneously, as it allows for models in which intervening variables explain direct associations that are not necessarily statistically significant. The specified models assessed direct actor effects (e.g., the link between one's attachment avoidance and one's perpetrated physical IPV), partner effects (e.g., the link between one's attachment avoidance and the partner's relationship dissatisfaction), and indirect effects through relationship dissatisfaction (with 95% confidence intervals estimated on 10,000 bootstrapping samples), while considering the non‐independence of the dyadic data. As recommended by Kenny and colleagues (2006), an omnibus within‐dyad test of distinguishability was performed to determine whether women and men should be treated as indistinguishable dyad members in the path analysis. To perform this test, actor and partner effects were constrained to be equal for women and men. A chi‐square difference test was then conducted to compare the constrained model to a non‐constrained model. Based on Kline's (2016) recommendations, the adjustment of the model to the data was assessed with three indices: a non‐significant chi‐square, a Comparative Fit Index (CFI) higher than .95, and a Root Mean Square Error of Approximation (RMSEA) lower than .08. According to Monte Carlo simulations ran on the fully constrained indistinguishable dyads models model with 1000 random samples, observed power estimates varied from 7% to 66.5% (indirect effects) and from 10% to 99% (direct effects).

Preliminary analyses

Table 1 presents the descriptive statistics, Cronbach's alpha coefficients, and bivariate correlations between the main variables for women and men. Over the past year, 95.5% of women and 92.0% of men reported using psychological IPV towards their partner at least once, whereas 53.4% of women and 51.1% of men reported using physical IPV at least once. The estimated frequency of psychological IPV perpetrated by women and men varied from 0 to 144 in the 12 months preceding the completion of the questionnaire. The estimated frequency of physical IPV perpetrated by women varied from 0 to 254 in the last year, while among men it varied from 0 to 204. As expected, we observed departure from normality in IPV variables, supporting our choice of using a robust maximum likelihood estimator in the main analyses.

Main analyses

Psychological IPV

The first model tested used psychological IPV as the outcome variable. Prior to analyzing the model, distinguishability of dyads was tested and revealed the absence of gender differences in actor and partner effects (Δχ2[15] = 8.73, p = .891). The model constraining all associations to be equal for women and men presented a better fit to the data (χ2[16] = 11.75, p = .761, CFI = 1.000, RMSEA = 0.000, 90% CI [0.000; 0.070]) than the model with no constraints (χ2[1] = 2.96, p = .085; CFI = 0.988; RMSEA = 0.149, 90% CI [0.000; 0.359]). The constrained model allowed higher statistical power and parsimony, so it was retained for the analyses.

As shown in Figure 1, significant direct positive associations were found between participants' attachment anxiety scores and their own and their partner's perpetration of psychological IPV. However, significant negative direct links were found between participants' attachment avoidance scores and their own and their partner's perpetration of psychological IPV. Results of indirect actor effects (see Table 2) revealed that participants' attachment avoidance was positively associated with their and their partner's higher relationship dissatisfaction, which in turn was positively associated with their perpetration of psychological IPV. The indirect actor effect of participants' attachment avoidance on their perpetration of psychological IPV accounted for 68% of the total actor effect of attachment avoidance on the perpetration of psychological IPV. The indirect effect through participants' own relationship dissatisfaction accounted for 66% of the indirect effect, while the effect through their partner's relationship dissatisfaction accounted for 34% of the indirect effect.

FIGURE 1

Associations among attachment insecurity, relationship dissatisfaction, and psychological intimate partner violence. Notes All paths between variables were tested. Only significant standardized path coefficients are shown. P1 = Partner 1. P2 = Partner 2. IPV = Intimate partner violence. *p < .05. **p < .01. ***p < .001

TABLE 2

Actor and partner total, direct, and indirect effects in dyadic models of psychological and physical IPV (N = 88 couples)

Effect	Psychological IPV				Physical IPV
	B	SE	p	95% CI	B	SE	p	95% CI
Avoidance
Actor total	−2.120	2.233	.342	[−6.794; 1.997]	−1.710	1.254	.173	[−4.237; 0.734]
Actor direct	−6.597	2.344	.005	[−11.402; −2.155]	−4.169	1.759	.018	[−7.592; −0.555]
Actor indirect	4.477	1.569	.004	[1.373; 7.500]	2.458	1.067	.021	[0.473; 4.733]
via own dissatisfaction	2.955	1.176	.012	[0.823; 5.467]	1.582	0.768	.039	[0.236; 3.285]
via partner's dissatisfaction	1.522	0.791	.054	[0.330; 3.514]	0.876	0.747	.241	[−0,122; 2.945]
Partner total	−1.602	2.274	.481	[−6.128; 2.853]	−0.473	1.219	.698	[−2.942; 1.861]
Partner direct	−6.259	2.564	.015	[−11.238; −1.251]	−3.099	2.245	.167	[−7.809; 1.069]
Partner indirect	4.657	1.710	.006	[1.293; 8.005]	2.625	1.488	.078	[−0.026; 5.872]
via one's dissatisfaction	3.289	1.391	.018	[0.602; 6.109]	1.893	1.396	.175	[−0.423; 5.133]
via partner's dissatisfaction	1.368	0.632	.030	[0.389; 2.925]	0.732	0.405	.070	[0.119; 1.753]
Anxiety
Actor total	8.147	2.088	<.001	[4.109; 12.394]	−1.205	1.959	.539	[−5.785; 1.918]
Actor direct	6.744	2.176	.002	[2.617; 11.163]	−1.987	1.872	.289	[−6.462; 0.906]
Actor indirect	1.403	0.910	.123	[0.032; 3.779]	0.782	0.617	.205	[−0.074; 2.466]
via own dissatisfaction	0.635	0.563	.260	[−0.194; 2.171]	0.340	0.363	.352	[−0.091; 1.440]
via partner's dissatisfaction	0.768	0.592	.195	[−0.019; 2.488]	0.442	0.475	.349	[−0.045; 2.177]
Partner total	5.389	2.064	.009	[1.511; 9.647]	−0.323	1.795	.857	[−4.490; 2.508]
Partner direct	3.993	2.149	.063	[0.037; 8.510]	−1.099	1.818	.545	[−5.227; 1.899]
Partner indirect	1.397	0.913	.126	[0.029; 3.782]	0.776	0.610	.203	[−0.021; 2.655]
via one's dissatisfaction	0.707	0.652	.278	[−0.213; 2.543]	0.407	0.475	.392	[−0.098; 2.076]
via partner's dissatisfaction	0.690	0.491	.160	[−0.006; 2.056]	0.369	0.280	.186	[−0.004; 1.202]

Unstandardized estimates are shown. Significant indirect effects are bolded.

Abbreviation: CI, confidences intervals (on 10,000 samples).

Results of indirect partner effects showed that participants' attachment avoidance was indirectly related to their partners' higher perpetration of psychological IPV through their own and their partner's higher relationship dissatisfaction. The indirect partner effect of participants' attachment avoidance on their partner's perpetration of psychological IPV accounted for 74% of the total partner effect of participants' attachment avoidance on the perpetration of the partner's psychological IPV. The indirect effect through participants' own relationship dissatisfaction accounted for 71% of the indirect effect, while the effect through their partner's relationship dissatisfaction accounted for 29% of the indirect effect.

Physical IPV

The second model tested included physical IPV as the outcome variable. Prior to analyzing the model, the distinguishability of dyads test revealed the absence of gender differences in actor and partner effects (Δχ2[15] = 15.56, p = .412). The model constraining all associations to be equal presented a better fit to the data (χ2[16] = 18.08, p = .319, CFI = 0.978, RMSEA = 0.038, 90% CI [0.000; 0.109]) than the model with no constraints (χ2[1] = 2.96, p = .085; CFI = 0.988; RMSEA = 0.149, 90% CI [0.000; 0.359]). The constrained model allowed higher statistical power and parsimony, so it was retained for the analyses.

As shown in Figure 2, a direct negative association was found between participants' attachment avoidance scores and their own perpetration of physical IPV. The direct link between attachment anxiety and their own perpetration of physical IPV was not significant. Results of indirect actor effects (see Table 2) revealed that participants' attachment avoidance was indirectly related to their higher perpetration of physical IPV through their higher relationship dissatisfaction. The indirect actor effect of participants' attachment avoidance on their perpetration of physical IPV accounted for 38% of the total actor effect of attachment avoidance on the perpetration of physical IPV. Results based on confidence intervals computed on 10,000 random samples also revealed significant associations between participants' attachment avoidance and physical IPV perpetrated by their partner through their partners' higher relationship dissatisfaction (i.e., partner effects), which accounted for 24% of the total partner effect of attachment avoidance on the perpetration of physical IPV by the partner.

FIGURE 2

Associations among attachment insecurity, relationship dissatisfaction, and physical intimate partner violence. Notes All paths between variables were tested. Only significant standardized path coefficients are shown. P1 = Partner 1. P2 = Partner 2. IPV = Intimate partner violence. *p < .05. **p < .01. ***p < .001

DISCUSSION

This study adds to the literature on IPV in couples seeking treatment for relationship difficulties by exploring the role of relationship dissatisfaction in the associations among attachment‐related anxiety and avoidance and the perpetration of psychological and physical IPV. In comparison with other clinical samples, the participants in this study reported a higher occurrence and frequency of psychological and physical IPV perpetrated by both partners (e.g., Bélanger et al., 2015; Tougas et al., 2016), higher attachment anxiety, lower attachment avoidance (Mondor et al., 2011), and similar relationship dissatisfaction (Sabourin et al., 2005). In line with previous studies using clinical samples of couples (e.g., Crane et al., 2014; Madsen et al., 2012), results showed that physical and psychological IPV perpetrated by participants were associated with their partner's perpetration of IPV, suggesting the presence of bidirectional IPV in many of the participating couples. Results emphasized that participants' attachment avoidance was associated with their higher perpetration of psychological IPV through their own, and their partner's, higher levels of relationship dissatisfaction. Findings also revealed that participants' attachment avoidance was indirectly related to their higher perpetration of physical IPV through their higher relationship dissatisfaction, and to higher perpetration of physical IPV by their partner through the partner's higher relationship dissatisfaction. Differences were observed between the psychological and physical IPV models, supporting the need to examine risk factors for the different forms of IPV separately. Taken together, the results support how attachment insecurities, especially attachment avoidance, might alter partners' relational dynamics in ways that may significantly impede their satisfaction with the relationship as well as their tendency to use different forms of violence towards each other.

Our results are consistent with previous research suggesting that attachment avoidance can be harmful to one's relationship satisfaction (e.g., Mikulincer & Shaver, 2016; Mondor et al., 2011). Indeed, individuals with greater attachment avoidance are prone to hold negative expectations and beliefs about their romantic relationship and their partner, to perceive others as unreliable and to act in ways that maintain the distance with their partner, all of which may be detrimental for their perception of satisfaction within their relationship (Mikulincer & Shaver, 2016). Relationally dissatisfied individuals are prone to respond with negativity (e.g., hostile anger, criticism, contempt) and to perceived negativity from their partner, which can contribute to establish dysfunctional patterns of interactions escalating into violence (Bartholomew & Cobb, 2011; Johnson & Whiffen, 2003; Mikulincer & Shaver, 2016). In turn, it is possible that when these individuals experience greater relationship dissatisfaction, their tendency to withdraw and suppress negative emotions pushes them, when it becomes unbearable, to use psychological violence towards their partner as a way to create distance with the partner (Brassard et al., 2014; Godbout et al., 2017). In addition, it is possible that when avoidantly attached individuals perceive that their partner is dissatisfied with the relationship, or when they receive demands for emotional closeness and reassurance by a dissatisfied partner, they might resort to aggression to keep their partner at a distance. By doing so, those with high avoidance feel they are better able to maintain their emotional independence from the partner which, they believe, might alleviate both partners' distress (for reviews, see Candel & Turliuc, 2019; Mikulincer & Shaver, 2016). Yet, our results show this strategy might rather enhance the risk of violence by both partners.

Indeed, our findings also highlighted that participants' attachment avoidance is associated with greater use of psychological and physical IPV by their partner through both partners' higher relationship dissatisfaction. As noted above, avoidant individuals are prone to distance themselves from their partners, especially when they experience relationship dissatisfaction, in order to reduce their discomfort with emotional intimacy or negative emotions (Mikulincer & Shaver, 2016). In a relational context, the use of deactivation strategies (e.g., denial, withdrawal strategies, disengagement during conflicts) by an avoidant individual may lead their partner to experience frustration, discontentment, and relationship dissatisfaction, which might be expressed through the use of violence (for reviews, see Mikulincer & Shaver, 2016; Velotti et al., 2018). Thus, faced with this distance, their partner may seek satisfaction to their emotional needs (e.g., attention, closeness, emotional support) through the use of increasingly violent behaviors (Bartholomew & Allison, 2006; Mikulincer & Shaver, 2016).

Beyond these indirect effects, a negative direct association was also found between participants' attachment avoidance and their perpetration of psychological and physical IPV. This result may appear to contradict previous studies (for reviews, see Mikulincer & Shaver, 2016; Spencer et al., 2020; Velotti et al., 2018), although they are complementing this study's indirect effects, revealing that attachment avoidance is positively linked to psychological and physical IPV perpetration through relationship dissatisfaction. Perhaps, avoidant partners commonly use withdrawing strategies to avoid conflicts and aggression with their partner, which prevents them from experiencing conflictual situations that can escalate into physical violence (Mikulincer & Shaver, 2016). Yet, when they do experience relationship dissatisfaction and their deactivation strategies are not sufficient to maintain a comfortable distance with their partner, their initial tendency to suppress their negative emotions during conflicts might become overwhelming and trigger negative patterns of interactions that lead to the use of physical or psychological violence. Thus, the use of these deactivation strategies to avoid intimacy or feelings of distress within the relationship might, on the one hand, be reinforced as they seem to directly help avoiding the escalation of conflicts yet, on the other hand, they might undermine their relationship satisfaction and put them at greater risk of using relationship violence (Mikulincer & Shaver, 2016).

Finally, results revealed a significant direct association between participants' attachment anxiety and their own and their partner's perpetration of psychological IPV. Consistent with previous studies (for reviews, see Mikulincer & Shaver, 2016; Spencer et al., 2020), this result suggests that individuals with greater attachment anxiety tend to use psychological violence (e.g., threats, insults) as an inadequate attempt to get their partner's attention, care, and support, or as a way to make their partner respond to their unmet attachment needs (Mikulincer & Shaver, 2016). These individuals' hypervigilance and concerns regarding the loss of their partner could also increase their perception of situations as threatening and decrease their ability to regulate negative emotions, which could put them at a greater risk of using violence when faced with the perception of unavailability from their partner (Mikulincer & Shaver, 2016; Stevens, 2014). In line with the results from previous studies (Péloquin et al., 2011; Sommer et al., 2017), attachment anxiety is also associated with greater use of psychological IPV by the partner. It is thus possible that the pursuit strategies (e.g., attention‐seeking, attempts to communicate, verbal abuse) used by anxious individuals, may contribute to their partners' violent response when they have failed to respond non‐violently (Bartholomew & Allison, 2006; Feeney, 2016). Yet, our findings only revealed a smaller role of attachment anxiety in the perpetration of violence in couples seeking help, which could be partly explained by the small sample size and the possibility that attachment avoidance may exert a stronger role on IPV, thus accounting for more variance in the statistical model. It is also possible that attachment anxiety alone is not associated with IPV perpetration but rather, IPV is precipitated by a partner's undesirable behavior such as rejection and inattentiveness, which are behaviors that avoidant individuals are prone to adopt (Mikulincer & Shaver, 2016). As such, a dyadic model that would account for attachment pairings might help us better understand the role played by attachment anxiety on IPV. Nonetheless, other studies that have recruited distressed couples in therapy have also highlighted that attachment avoidance, compared to attachment anxiety, seems to exert a stronger negative impact on relationship dissatisfaction (Mondor et al., 2011). Indeed, anxious individuals are known to perceive threats to the relationships quite easily which pushes them to seek reassurance or security even in relationships that are satisfactory and well‐functioning. However, avoidant individuals' sensitivity to perceived threats to their independence is weaker, which can lead them to use withdrawing strategies only in highly distressing situations (Mikulincer & Shaver, 2016). As such, the smaller role of attachment anxiety on relationship dissatisfaction and on IPV perpetration, compared to the role of avoidance, could thus be due to the highly distressing nature of the relationship in couples seeking treatment (Mondor et al., 2011).

Limitations

This study had several strengths such as the use of a dyadic design, validated measures, rigorous dyadic statistical analyses, and the distinction between different forms of violence as well as the specificity of the sample. Yet, some limitations need to be considered. First, the research design was cross‐sectional, which requires caution in the interpretation of the directionality of the results. Data were also collected through self‐report questionnaires, which can enhance recall bias and social desirability that were not controlled for in this study. Also, the statistical power of our analyses was limited by the small sample size but constraining the effects between men and women has helped to reduce this bias. Yet, the models allowed us to detect some significant indirect effects, suggesting that a larger sample size may have allowed detecting additional significant indirect effects. Considering the difficulties inherent to recruiting couples seeking couple therapy for violent behaviors, our results on this rare population reveal important findings. Moreover, the percentage of explained variance in IPV, whether physical or psychological, that relationship dissatisfaction offers remains small, which suggests that other variables might explain the link between attachment insecurities and the perpetration of physical and psychological IPV. Future studies should aim to use larger clinical samples and replicate the proposed models by including both forms of IPV in one comprehensive model with more predictors, such as empathy (Péloquin et al., 2011), impulsivity (Dugal et al., 2019), and dysfunctional communication patterns (Fournier et al., 2011). In addition, this study used the highest score of the partners to calculate IPV perpetration in order to account for the possible under‐reporting of IPV, but that could lead to an overestimation of perpetrated IPV. Previous studies have used this method (e.g.,Cuenca et al., 2015; Dugal et al., 2019; Gou & Woodin, 2017) and it has been judged less problematic than using self‐report scores, without considering the partner's data.

Finally, the generalizability of our findings is limited by the composition of the sample which only included cross‐gender French‐Canadian couples seeking therapy at two community‐based clinics providing specialized services for couple and partner violence difficulties. Couples who consult in private settings might report a larger array of difficulties, such as sexual issues or co‐parenting difficulties (e.g., Gurman, 2008; Péloquin et al., 2019), which might differ from couples who seek therapy in a specialized center for partner violence. Also, the socioeconomic status of couples who seek treatment in community‐based clinics is generally lower than the status of couples who consult in private settings (e.g., Callaci et al., 2020; Tougas et al., 2016), which limits generalizability. In addition, some couples in the study lived separately, which may have limited the frequency of violent behaviors that the participants used toward their partner compared to couples living together. Future research could replicate this study with larger samples of participants from various sexual and gender diversity and cultural backgrounds. This would also allow to consider attachment pairings/mispairings or the interaction between attachment anxiety and avoidance within participants, to better understand how both partners' attachment insecurities interact in a way that might increase a couple's risk of experiencing IPV. Also, future studies should use longitudinal research designs to investigate the directionality of the associations among these study variables. Finally, studies may benefit from assessing the circumstances under which individuals use violence against their partners (e.g., using daily diaries) to further unveil the links between attachment, relationship dissatisfaction, and IPV.

Implications

Considering the very high prevalence of IPV reported by our participants, findings underline the importance of assessing both partners' perpetrated acts of violence when starting couples' therapy. By doing so, clinicians might be better able to tailor their interventions to the specific dynamics of the couple, whether there are no violence or violent acts perpetrated by one or both partners. Results also support the clinical implications offered by Spencer and colleagues (2020) in their review on attachment and IPV, which promotes the use of an attachment‐based framework when working with couples who report IPV resulting in the escalation of conflicts (Johnson, 2008) to prevent or reduce its occurrence. It is important to note, however, that in the presence of severe unidirectional violence or intimate terrorism, which arises when one partner is controlling and threatening the other, couple therapy is contraindicated and individual therapy is recommended (Stith et al., 2011). Couples experiencing relationship difficulties, such as situational IPV that might result from unmet attachment needs and relationship dissatisfaction, may benefit from therapeutic approaches such as Emotionally Focused Couple Therapy (EFCT; Johnson, 2020), a model that is gaining empirical support and can promote both secure attachment and relationship satisfaction (Brassard & Johnson, 2016). In EFCT, therapists work with couples to change the negative rigid interactional patterns that partners are consistently engaging in by promoting their ability to share their attachment needs and reach for connection (co‐regulation) rather than entering the conflict in a blaming/critical (hyperactivation) or withdrawn/shut down way (deactivation; Johnson, 2020). Results emphasize that by helping avoidant partners to reengage in the relationship, all while guiding anxious partners to diminish their pursuing and blaming behaviors (Johnson, 2020), EFCT might directly enhance relationship satisfaction, and indirectly prevent or reduce the perpetration of psychological and physical IPV among couples who desire to remain with their romantic partner and who want to change their behavior.