Sex is a crucial component of romantic relationships (Busby et al., 2020), and serves as an important protective factor in marriage for both females and males (Birnbaum, 2015; Butzer & Campbell, 2008; Peters & Meltzer, 2021; Sprecher & Cate, 2004). It also has a significant impact on partners’ psychological and physiological health throughout the course of a relationship (Mitchell et al., 2011). While higher sexual satisfaction leads to higher relationship quality and lower relationship instability (Fallis et al., 2016; Yeh et al., 2006), the lack of sexual satisfaction undermines both partners’ overall functioning (Anderson, 2013), perceived intimacy (Heiman et al., 2011), and relationship happiness (Fisher et al., 2015). Moreover, higher sexual dissatisfaction has strong associations with a variety of problems, including deteriorated relationship quality (Byers, 2005), insecure pair-bonding (Meltzer et al., 2017), lower long-term commitment (Seiter et al., 2020), higher divorce rate (Shakerian et al., 2014), less effective communication and problem-solving (Duba et al., 2012), lower self-esteem (Terauchi et al., 2017), and decreased physical health (Diamond Huebner, 2012). Therefore, sexual dissatisfaction has been purposed as one of the most important treatment targets for relieving relational distress in couple therapy (Rothman et al., 2020). In this study, we attempt to understand how two specific types of relational aggression (i.e., love withdrawal and social sabotage) might be directly and indirectly associated with sexual dissatisfaction through attachment insecurity.

Relational aggression is a covert form of aggressive behaviors for handling conflicts or differences in a relationship. It is defined as purposeful actions for hurting or harming others by manipulation of relationships, and/or by creating social rejection and exclusion (Crick & Grotpeter, 1995). Examples of relational aggression include spreading rumors, creating rejection from social groups, and withdrawing affection. Some researchers use the closely related concepts of relational aggression, nonverbal aggression, and psychological aggression interchangeably. Yet, these three constructs are distinct. Unlike nonverbal aggression, relational aggression does not include any physically aggressive behaviors (Bowie, 2007). Psychological and relational aggression are both nonphysical in nature. But relational aggression is generally more subtle (Crick & Grotpeter, 1995), and does not manifest through overt behaviors (i.e., ridiculing, verbal threats) as psychological aggression commonly does (Lawrence et al., 2009). Above all, the most definable characteristic of relational aggression is that it aims at damaging the victim’s significant relationships, instead of directly harming the victim physically or psychologically.

The detrimental effects of relational aggression within peer contexts, especially among children and adolescents, have been studied and supported by many scholars (Casas & Bower, 2018). Even though relational aggression is more common in couples than in peers (Goldstein, 2011), it was not until the most recent decade that researchers started to investigate the influences of relational aggression in romantic relationships.

Relational Aggression in Couples

While the majority of current studies on couple aggressive behaviors have focused on more overt forms (e.g., intimate partner violence, and reciprocity of negative interactions), findings on covert patterns of aggression are glaringly lacking. Noticing the pervasiveness of non-overt forms of aggression (Archer & Coyne, 2005), researchers have recently expanded the conceptualization of couple aggression to include more covert, indirect, and subtle forms (Goldstein, 2011; Linder et al., 2002; Nelson & Carroll, 2006). These studies indicate that both females and males use relational aggression in romantic relationships frequently. The resulting damaging effects include but are not limited to significant mental maladjustment (i.e., social anxiety, depressive symptoms, alcohol abuse; Bagner et al., 2007), attachment insecurity (Oka et al., 2016; Sandberg et al., 2018), and worsened relational and personal wellbeing (Burk & Seiffge-Krenke, 2015; Goldstein, 2011).

Carroll et al. (2010) suggested two distinct forms of relational aggression in couples: love withdrawal and social sabotage. Love withdrawal, a direct form of relational aggression, is characterized by a partner’s consistent and direct inattentiveness, such as withholding affection and sex, and silent treatment. Love withdrawal does not involve a third party, although people outside of a relationship might notice the aggressive behaviors. Examples of love withdrawal include “my partner ignores me when she/he is angry with me,” and “my partner intentionally ignores me until I give in to his/her way about something.” Social sabotage, an indirect form of relational aggression, is manifested by a partner’s circuitous behaviors in recruiting third parties to attack the victim, such as sharing rumors and gossip with friends. Because social sabotage significantly violates the private and exclusive nature of a couple dyad, it is more extreme and damaging, while love withdrawal is more common (Carroll et al., 2010). Examples of social sabotage include “my partner tries to embarrass me or make me look stupid in front of others,” and “my partner has shared negative information about me to be mean.” Studies have shown that both love withdrawal and social sabotage are predictive of poorer marital quality and lower marital stability across time (Coyne et al., 2017). Yet, only social sabotage has a longitudinal association with partners’ physical health (Martin et al., 2015). Additionally, albeit females in general use both love withdrawal and social sabotage more often and at higher levels (Carroll et al., 2010), males’ use of both types is more destructing (Coyne et al., 2017).

The body of literature on relational aggression in couples is growing, but there are three noteworthy limitations to the existing studies. First, most of these studies defined relational aggression as a broad construct that involves all nonphysical aggression, such as verbal attacks and name-calling (Burk & Seiffge-Krenke, 2015). As discussed earlier, relational aggression differs from other types of aggression (e.g., nonverbal and psychological aggression). A broad conceptualization of relational aggression might limit the specific contexts to which the findings are applicable. Besides, a broad definition also makes it harder to distinguish the unique consequence of relational aggression from other aggressive behaviors (e.g., intimate partner violence). Second, there are at least two distinct types of relational aggression in couples (Carroll et al., 2010), love withdrawal and social sabotage. While love withdrawal does not involve other people, the use of social sabotage breaks the dyadic boundary to recruit third parties. Therefore, social sabotage could certainly inflict harm that is not associated with love withdrawal, such as defamation of the victim’s public reputation (Coyne et al., 2017). Yet, many existing studies only assessed relational aggression as a whole without considering its subtypes, which could obscure the corresponding impacts of a particular subtype, and make it harder to generate specific interventions and practical guidance. Lastly, without tests of measurement equivalence (also known as factorial invariance; Little, 2013) on relational aggression across gender groups, the found gender differences in relational aggression might be owing to female and male different understandings (Dyer, 2015), instead of real different impacts from relational aggression.

In sum, for relational aggression in couples, existing studies have indicated its prevalence and negative outcomes. Yet, with the limitations discussed above, more studies are needed for delineating its ramifications in couple relationships, such as sexual dissatisfaction.

Relational Aggression and Sexual Dissatisfaction

Compared to other interpersonal contexts, a unique aspect of relational aggression in couples is that it usually includes withholding sexual, emotional, and physical intimacy from the victim (Nelson & Carroll, 2006). This could directly hinder partners’ expression of love, lower sexual desire, frequency, and enjoyment, and ultimately contribute to a higher level of sexual dissatisfaction for both partners (Sandberg et al., 2018). Moreover, relational aggression is linked with poorer physical health (Martin et al., 2015), and significant psychological maladjustment (e.g., anxiety and depression; Bagner et al., 2007), through which relational aggression could further decrease partners’ sexual function and quality (McCabe et al., 2010).

Relational Aggression and Attachment Insecurity

As negative tactics, relational aggression would weaken partners’ feelings of connection and acceptance (Goldstein, 2011), which would unavoidably lead to the experience of insecure feelings in the relationship (Burk & Seiffge-Krenke, 2015). For example, love withdrawal can directly create a sense of disconnection for the victimized partner through behaviors such as consistently withholding affection and sex (Carroll et al., 2010). What’s more, social sabotage could be perceived as a form of betrayal by the victim (Coyne et al., 2017) that the victim would feel less capable of trusting or depending on the perpetrator (Linder et al., 2002), which would further contribute to the experience of insecurity in the relationship (Oka et al., 2016). Besides, social sabotage, by recruiting third parties to handle differences and conflicts, is essentially a form of attachment avoidance that the perpetrator does not communicate with his/her partner directly or seek acceptance from the relationship (Hoover & Jackson, 2019).

Please note, the current study assessed partners’ experience/feelings of attachment insecurity (i.e., attachment anxiety and avoidance), instead of attachment styles in the current study. While attachment styles are developed with caregivers at a very young age and are commonly thought to be consistent across life (Bowlby, 1958), attachment experience and behaviors could change in given relational contexts throughout an individual’s lifespan (Birnbaum, 2015; Birnbaum & Reis, 2019; Burk & Seiffge-Krenke, 2015; Novak et al., 2017; Sandberg et al., 2018).

Attachment Insecurity and Sexual Dissatisfaction

“Adult romantic relationships are typically sexual in nature” (Zayas et al., 2015, p. 69), and attachment security plays an irreplaceable role in enhancing emotional bonds (Birnbaum, 2015), and advancing proximity and sexual intimacy (Johnson, 2019). In romantic relationships, the attachment and sexual systems reciprocally affect each other (Zayas et al., 2015), and jointly operate to impact romantic partners’ overall well-being and relationship quality (Birnbaum & Reis, 2019). Attachment guides partners’ development and perceptions of sexual interactions, while sexual behaviors serve as crucial motivations and forces for the formation of a strong romantic attachment (Birnbaum, 2015). From a developmental perspective, sex not only attracts partners to each other at the beginning stages of a relationship, but also powerfully fosters emotional bonds (Johnson, 2019) and promotes relationship quality during almost the entire course of the relationship (Butzer & Campbell, 2008). At the same time, attachment security, by providing a sense of safety and acceptance, could increase partners’ sexual desire and satisfaction (Mark et al., 2018). On the contrary, attachment insecurity would significantly increase sexual anxiety, inhibit sexual communication, and sabotage the overall physical satisfaction of both partners (Busby et al., 2020; Davis et al., 2016).

Relational Aggression, Attachment Insecurity, and Sexual Dissatisfaction

Attachment insecurity has been proposed as a significant mediating mechanism between relational predictors and outcomes in many extant studies (Alder et al., 2018; Ben-Naim et al., 2013; Novak et al., 2017). Particularly, both types of relational aggression could lead to experience/feelings of attachment insecurities in the relationship (Burk & Seiffge-Krenke, 2015), which might further create sexual anxiety, and disrupt sexual connection and satisfaction (Davis et al., 2016). The study by Sandberg et al. (2018) found that, for both females and males, attachment insecurity (i.e., anxiety and avoidance) significantly mediated the relationship between relational aggression and sexual satisfaction with a few exceptions. Their findings suggest, relational aggression directly compromises emotional security and safety for both partners in a relationship, and further undermines couples’ sexual satisfaction through the insecure feelings. However, their study examined relational aggression as a whole without knowing the unique influences of love withdrawal and social sabotage on sexual satisfaction.

To sum up, findings from previous studies support attachment insecurity as a potential mediator between relational aggression and sexual dissatisfaction. Because relational aggression would increase feelings of insecurity in the relationship, and insecurities feelings run counter to sexual satisfaction. However, existing findings are very limited. Particularly, no research thus far has investigated how two dimensions of attachment insecurity (i.e., attachment anxiety and avoidance) uniquely mediate the relationship between relational aggression and sexual dissatisfaction. Therefore, by including love withdrawal and social sabotage as two distinct subtypes of relational aggression, the current study aims at evaluating the mediating effects of attachment anxiety and attachment avoidance in the connection between relational aggression and sexual dissatisfaction.

Current Study

This study explores the associations between relational aggression (i.e., love withdrawal and social sabotage) and sexual dissatisfaction in marriage, with attachment insecurity (i.e., attachment anxiety and attachment avoidance) as the mediators. We used Actor-Partner Interdependence Model (APIM; Cook & Snyder, 2005) and Actor-Partner Interdependence Mediation Model (APIMeM; Ledermann et al., 2011) to answer the following questions:

  1. RQ1:

    Do each partner’s perceived love withdrawal and social sabotage have direct associations with their own (actor effects) and their partner’s (partner effects) sexual dissatisfaction?

  2. RQ2:

    Using attachment anxiety as the mediator, do each partner’s perceived love withdrawal and social sabotage have indirect associations with their own (indirect actor effects) and their partner’s (indirect partner effects) sexual dissatisfaction?

  3. RQ3:

    Using attachment avoidance as the mediator, do each partner’s perceived love withdrawal and social sabotage have indirect associations with their own (indirect actor effects) and their partner’s (indirect partner effects) sexual dissatisfaction?

Method

Participants

The sample for this study was taken from Wave 3 of the Flourishing Families Project (FFP; please see Day and Padilla-Walker (2009) for detailed information). FFP is a longitudinal study for examining various inner-family processes, and FFP received annual approval from the Institutional Review Board. As the focus of the current study was on marriage, single-parent families were excluded, resulting in an initial sample of 327 couples at Wave 3. There were very few participants who did not identify themselves as heterosexual and/or were not married at Wave 3 (less than 1%, n = 3), so they were not included in this study. The final sample for the current study included 324 heterosexual married couples, who were from a large northwestern city in the United States.

In our study, there were very few missing values (< 1%) on the main variables. The missingness on demographic information were relatively low: no missing values on age and marital status, 0.31% (n = 1) missing on male race/ethnicity, 0.31% (n = 1) missing on male education, 4.32% (n = 14) of female and 1.54% (n = 5) of male missing on income, and 1.54% (n = 5) of female and 1.85% (n = 6) of male missing on length of marriage.

Participants’ average age was 45.95 years (SD = 5.26) for females, and 47.66 years (SD = 5.79) for males. The majority of females (84.26%, n = 273) and males (88.58%, n = 287) were European American, with small percentages of Asian American (4.01%, n = 13 for females; 1.86%, n = 6 for males), Multi-Ethnic (3.7%, n = 12 for females; 2.17%, n = 7 for males), Hispanic (3.4%, n = 11 for females; 0.62%, n = 2 for males), African American (3.09%, n = 10 for females; 4.02%, n = 13 for males), and other (1.54%, n = 5 for females; 2.17%, n = 7 for males). In terms of religion, almost a half of females (47.64%, n = 121) and males (43.65%, n = 110) reported themselves as Protestant, with 18.90% (n = 48) of females and 16.67% (n = 42) of males reported as Catholic. As for education, 72.22% (n = 234) of females and 71.3% (n = 231) of males had a bachelor’s degree or higher. For combined annual income, 1.29% (n = 4) of females and 0.94% (n = 3) of males were less than $20,000, 15.48% (n = 48) of females and 15.99% (n = 51) of males were between $20,000 and $59,999, 37.42% (n = 116) of females and 33.23% (n = 106) of males were between $60,000 and $99,999, and 45.81% (n = 142) of females and 49.84% (n = 159) of males had more than $100,000. The average relationship length was 19.79 years (SD = 4.53). Besides, 79.32% (n = 257) of females reported having child/children and the median number of children was 3.

Measures

Sexual Dissatisfaction

A four-item scale was created by FFP with responses ranging on a 5-point Likert-type scale from 1 (never) to 5 (all of the time), and higher scores indicate greater sexual dissatisfaction. Items include “I would like my partner to express more tenderness during intercourse,” and “I feel our sexual activity is just routine.” Cronbach’s alpha was 0.72 for females and 0.71 for males. Measurement invariance tests showed factor loadings (i.e., weak invariance; Dyer, 2015) for sexual dissatisfaction were invariant (ΔCFI = 0) across partners, meaning that the scale items tap into a similar concept across gender groups in our sample (Dyer, 2015). Equality constraints of factor loadings were imposed across partners for this measure.

Relational Aggression

The Couples Relational Aggression and Victimization Scale (Nelson & Carroll, 2006) has 11 items with responses ranging on a 7-point Likert-type scale from 1 (not at all true) to 7 (very true), and higher scores indicate greater perceived relational aggression. The love withdrawal subscale has five items evaluating participants’ perception that their partner withholds affection and intimacy to handle conflicts or differences. Items include “My partner gives me the ‘cold shoulder’ when he/she is angry with me.” Cronbach’s alpha was 0.91 for females and 0.87 for males. Measurement invariance tests showed factor loadings for love withdrawal were invariant (i.e., weak invariance, ΔCFI = 0.004) across partners. Equality constraints of factor loadings were imposed across partners for the love withdrawal subscale. The social sabotage subscale has six items assessing participants’ perceptions that their partner uses socially aggressive behaviors to handle conflicts or differences. Items include “When my partner has been mad at me, he/she recruits other people to ‘take sides’ with him/her and gets them upset with me too.” Cronbach’s alpha was 0.84 for females and 0.90 for males. Measurement invariance tests showed patterns of fixed and free factor loadings (i.e., configural invariance, CFI = 0.934; Dyer, 2015) for this subscale were invariant across partners, indicating that same scale items are appropriate for resenting the conceptual information across genders. Equality constraints for configural invariance were imposed across partners for the social sabotage subscale.

Attachment Insecurity

The Revised Experiences in Close Relationships Questionnaire (Chris Fraley et al., 2006) is an 8-item measure with responses ranging on a 7-point Likert-type scale from 1 (strongly disagree) to 7 (strongly agree), and higher scores indicate greater experience and feelings of attachment insecurity. To help participants focus on the current relationship, the wording was changed from “a/romantic partner” to “my partner.” The attachment anxiety subscale has four items, including “I am afraid that I will lose my partner’s love.” Cronbach’s alpha was 0.89 for both partners. Measurement invariance tests showed factor loadings, intercepts, and error variances (i.e., strict invariance; Dyer, 2015) for attachment anxiety were invariant (ΔCFI = 0.004) across partners, meaning that all the scale’s properties are the same across gender groups. Equality constraints for strict invariance were imposed across partners for the attachment anxiety subscale. The attachment avoidance subscale has four items, including “I find it difficult to allow myself to depend on my partner.” Cronbach’s alpha was.81 for females and 0.79 for males. Measurement invariance tests showed patterns of fixed and free factor loadings (i.e., configural invariance, CFI = 0.934) for attachment avoidance were invariant across partners. Corresponding equality constraints were imposed across partners for this subscale.

Analysis

Considering the natural links between partner scores (Cook & Snyder, 2005), it is essential to use dyadic methods to account for the non-independent components. We used APIM and APIMeM with structural equation modeling (SEM) in Mplus Version 8.7 (Muthén & Muthén, 2017) to study the actor and partner effects of relational aggression. SEM takes into account measurement errors and different weightings of a scale’s items, which gives more robust estimations (Kline, 2016). Further, in order to compensate for the potential issues with low Cronbach’s alpha reliability, adjust for bias from measurement error, and estimate more reliable results (Cole & Maxwell, 2003), we used all latent variables in this study. Because item responses in the current sample were highly skewed to one or two categories, all items were analyzed as categorical variables with the weighted least square mean and variance adjusted (WLSMV) estimator in Mplus 8.7. WLSMV does not assume a normal distribution, and is a preferred option for estimating models with categorical or ordered data (Brown, 2015).

To test attachment insecurity’s function as a mediator, we first used an APIM to assess the direct effects of relational aggression on sexual dissatisfaction. Next, we utilized two APIMeMs to assess the indirect effects of relational aggression via the mediating mechanism of attachment insecurity. Please note, attachment anxiety and attachment avoidance were estimated as mediators in two separate APIMeMs because the participants’ attachment anxiety and attachment avoidance were highly correlated (r = 0.71, p < 0.001 for female; r = 0.74, p < 0.001 for male), that when they were both included in one APIMeM, we noticed issues of collinearity (e.g., significantly flipped or theoretically inconsistent signs of coefficients).

Indirect effects are not normally distributed (Ledermann et al., 2011). Thus, we used bootstrapping (n = 5,000) for the bias-corrected bootstrap confidence intervals in the two APIMeMs (Alfons et al., 2021). Moreover, as this study is essentially a comparison across groups (i.e., female and male), to ensure the results are valid, we performed measurement invariance tests on all scales. The decrease in the comparative fit index (CFI; ΔCFI > 0.01) was the reference for determining whether the model fit significantly worsened (Dyer, 2015).

To determine the inclusion of control variables, we first performed multi-collinearity diagnoses with all the demographic variables (i.e., age, race/ethnicity, income, religion, education level of both partners). Though we did not have a variance inflation factor that was over 5, we excluded males’ age and income in the following analyses because they were highly correlated with females (r = 0.75, p < 0.0001 for age; r = 0.92, p < 0.0001 for income). Besides, with high percentages of Protestants (47.64% of female; 43.65% of male), the original religion variables were binary coded that Protestants coded as 1 and all others as 0. Then, we conducted bivariate correlation analyses for continuous demographic variables, bivariate Spearman correlation analyses for ordinal demographic variables, and one-way analyses of variance (ANOVAs) for categorical demographic variables. These analyses did not provide evidence for including any demographic variables as controls. Previous studies included demographic variables such as relationship length, income, age, and so on as controls (Coyne et al., 2017; Martin et al., 2015). However, the inclusion of those demographic variables not only significantly decreased the model fit (e.g., ΔCFI > 0.01), but also did not show statistical significance. Therefore, no demographic variables were controlled for in the current study.

Results

All the main variables were positively (p < 0.01) correlated with each other. Please see Table 1 for correlations, means, and standard deviations of all main study variables.

Table 1 Correlations for main study variables
Full size table

Direct Path Effects

Model fit indices suggest that the initial APIM excellently fits the data: χ2 (382, N = 324) = 590.69, p < 0.001; RMSEA = 0.041; CFI = 0.980; TLI = 0.977; SRMR = 0.057. The APIM accounted for 33.70% for females and 38.30% for males of the variance in sexual dissatisfaction. Please see Fig. 1 as the conceptual APIM, and see Table 2 for detailed results.

Fig. 1
figure 1

APIM for Relational Aggression and Sexual Dissatisfaction. N = 324 couples. Solid lines depict significant paths and dashed lines depict nonsignificant paths. Path and covariance coefficients are unstandardized (b). F female data; M male data; SEX sexual dissatisfaction; LW perceived love withdrawal; SS perceived social sabotage. *p < .05, **p < .01, ***p < .001

Full size image
Table 2 APIM for relational aggression and sexual dissatisfaction
Full size table

Direct Actor Effects

Female perceived love withdrawal was directly associated (b = 0.39, p < 0.001) with female sexual dissatisfaction. Both male perceived love withdrawal (b = 0.25, p < 0.001) and social sabotage (b = 0.16, p = 0.041) were directly associated with male sexual dissatisfaction. These positive associations mean that the higher a partner perceived love withdrawal or social sabotage, the higher the partner’s sexual dissatisfaction. The actor path of perceived social sabotage was not significant for females, indicating that female perceived social sabotage was not predictive of their own sexual dissatisfaction.

Direct Partner Effects

Male perceived love withdrawal was positively associated (b = 0.18, p = 0.018) with female sexual dissatisfaction. Yet, male perceived social sabotage was negatively associated (b = -0.20, p = 0.033) with female sexual dissatisfaction, indicating that the greater the male perceived social sabotage, the lower the female sexual dissatisfaction. The partner paths of both types of relational aggression were not significant for males, meaning that female perceived relational aggression was not predictive of male sexual dissatisfaction.

Indirect Path Effects Mediated by Attachment Anxiety

Model fit indices suggest that this APIMeM excellently fits the data: χ2 (628, N = 324) = 905.838, p < 0.0001; RMSEA = 0.037; CFI = 0.983; TLI = 0.981; SRMR = 0.054. This APIMeM accounted for 46.10% for females and 45.70% for males of the variance in sexual dissatisfaction, and 51.60% for females and 50.10% for males of the variance in attachment anxiety. Please see Fig. 2 as the conceptual APIMeM, and see Table 3 for detailed results.

Fig. 2
figure 2

APIMeM with Attachment Anxiety as Mediator. N = 324 couples. Solid lines depict significant paths and dashed lines depict nonsignificant paths. Path and covariance coefficients are unstandardized (b). F female data; M male data; SEX sexual dissatisfaction; LW perceived love withdrawal; SS perceived social sabotage; ANX attachment anxiety. *p < .05, **p < .01, ***p < .001

Full size image
Table 3 APIMeM with Attachment Anxiety as Mediator
Full size table

Indirect Actor Effects

With female attachment anxiety as the mediator, female perceived love withdrawal (b = 0.09, p = 0.013) and social sabotage (b = 0.18, p = 0.002) were both indirectly associated with their own sexual dissatisfaction. With male attachment anxiety as the mediator, male perceived love withdrawal was indirectly associated (b = 0.13, p = 0.003) with their own sexual dissatisfaction. These found positive and indirect actor effects were through an actor-actor mediation mechanism (Ledermann et al., 2011).

Indirect Partner Effects

With male attachment anxiety as the mediator, indirect partner effects from perceived love withdrawal were found for females (indirect partner effects through actor-partner mediation), and indirect partner effects from perceived social sabotage were found for males (indirect partner effects through partner-actor mediation; Ledermann et al., 2011). Specifically, male perceived love withdrawal was indirectly associated (b = 0.08, p = 0.044) with female sexual dissatisfaction, and female perceived social sabotage was indirectly associated (b = 0.08, p = 0.021) with male sexual dissatisfaction. Female attachment anxiety did not mediate any indirect paths between relational aggression and sexual dissatisfaction.

Direct Actor Effects

Female perceived love withdrawal was directly associated (b = 0.30, p < 0.001) with their own sexual dissatisfaction. Female (b = 0.30, p < 0.001) and male (b = 0.27, p < 0.001) attachment anxiety were both directly associated with their own sexual dissatisfaction.

Direct Partner Effects

Male perceived social sabotage was directly and negatively (b = -0.21, p = 0.019) associated with female sexual dissatisfaction, which was consistent with the results from the initial APIM. Male attachment anxiety was directly associated (b = 0.16, p = 0.029) with female sexual dissatisfaction.

Relational Aggression on Attachment Anxiety

Female perceived love withdrawal (b = 0.29, p < 0.001) and social sabotage (b = 0.60, p < 0.001) were both directly associated with female attachment anxiety. Male perceived love withdrawal (b = 0.48, p < 0.001), and female perceived social sabotage (b = 0.31, p = 0.005) were directly associated with male attachment anxiety.

Indirect Path Effects Mediated by Attachment Avoidance

Model fit indices suggest that this APIMeM excellently fits the data: χ2 (625, N = 324) = 974.804, p < 0.0001; RMSEA = 0.042; CFI = 0.969; TLI = 0.965; SRMR = 0.058. This APIMeM accounted for 53.40% for females and 53.60% for males of the variance in sexual dissatisfaction, and 53.20% for females and 41.40% for males of the variance in attachment avoidance. Please see Fig. 3 as the conceptual APIMeM, and see Table 4 for detailed results.

Fig. 3
figure 3

APIMeM with Attachment Avoidance as Mediator. N = 324 couples. Solid lines depict significant paths and dashed lines depict nonsignificant paths. Path and covariance coefficients are unstandardized (b). F female data; M male data; SEX sexual dissatisfaction; LW perceived love withdrawal; SS perceived social sabotage; AVO attachment avoidance. *p < .05, **p < .01, ***p < .001

Full size image
Table 4 APIMeM with Attachment Avoidance as Mediator
Full size table

Indirect Actor Effects

Female perceived social sabotage was indirectly associated (b = 0.31, p = 0.007) with female sexual dissatisfaction through female attachment avoidance. Male perceived social sabotage was indirectly associated (b = 0.13, p = 0.022) with male sexual dissatisfaction via male attachment avoidance. Attachment avoidance did not mediate the indirect actor effects between partners’ perceived love withdrawal and sexual dissatisfaction.

Direct Actor Effects

Female (b = 0.28, p < 0.001) and male (b = 0.17, p = 0.027) perceived love withdrawal were both directly associated with their own sexual dissatisfaction. Moreover, female (b = 0.52, p < 0.001) and male (b = 0.04, p < 0.001) attachment avoidance were also directly associated with their own sexual dissatisfaction.

Relational Aggression on Attachment Avoidance

Both female perceived love withdrawal (b = 0.17, p = 0.037) and social sabotage (b = 0.59, p < 0.001) were directly associated with female attachment avoidance. Only male perceived social sabotage was associated (b = 0.33, p = 0.001) with male attachment avoidance.

Discussion

The current study investigated the direct and indirect associations between two types of relational aggression and sexual dissatisfaction. Results indicated not only direct path effects between relational aggression and sexual dissatisfaction, but also indirect actor and partner effects via attachment anxiety and attachment avoidance. Moreover, attachment anxiety and avoidance were predictive of sexual dissatisfaction for both partners in our study.

Direct Path Effects

Results of the direct effects model (i.e., the initial APIM) are mostly consistent with previous studies that both love withdrawal and social sabotage have detrimental effects in marriages (Carroll et al., 2010; Coyne et al., 2017; Martin et al., 2015). The current study specifically shows that female perceived love withdrawal has a positive association with female sexual dissatisfaction, while male perceived love withdrawal has a positive association with sexual dissatisfaction for both self and the female partner. As love withdrawal is commonly manifested through withholding sexual intimacy, physical affection, and emotional support (Nelson & Carroll, 2006), it is not surprising to find these positive associations between love withdrawal and sexual dissatisfaction. As for female perceived love withdrawal not being predictive of male sexual dissatisfaction, a potential explanation may be that, with self-reported measures, the love withdrawal perceived by females could just be the female partners’ individual experience and feelings, instead of males’ actual use of love withdrawal in the marriage (Coyne et al., 2017). This potential gap between female perception and male actual use of love withdrawal might explain the finding that, female perceived love withdrawal is only directly linked to their own, but not male’s, sexual dissatisfaction. However, as one of our reviewers pointed out, there was a significant association between male perceived love withdrawal and female sexual dissatisfaction. Therefore, we need further study that includes both perceived and actual use of love withdrawal to understand these associations.

The direct effects model also indicate that male perceived social sabotage has a positive association with male sexual dissatisfaction, but a negative association with female sexual dissatisfaction. There are three possible explanations for these contradictory associations. First, compared to males, females are more likely to use relational aggression from a very young age (Archer & Coyne, 2005), and regard social sabotage to be more acceptable (Nelson & Carroll, 2006) and less detrimental (Casas & Bower, 2018; Crick & Grotpeter, 1995). Second, behaviors such as “talking about husband with friends” might be perceived as social sabotage by males. Yet, for females, those behaviors could be common ways of relieving marital distress or enhancing social bonds with friends (Carroll et al., 2010). Third, relational aggression has positive links with social anxiety and loneliness (Bagner et al., 2007). By connecting with third parties, females’ use of social sabotage could greatly relieve their anxiety and loneliness, at the same time, distract them from their marital issues (Bowen, 1978). But these potential positive effects of social sabotage are a possibility for females and not for their partners. Thus, the explanations above may be the rationale that, male perceived use of social sabotage by females is predictive of lower female sexual dissatisfaction, but higher male sexual dissatisfaction.

Indirect Path Effects Mediated by Attachment Anxiety

Results of the first APIMeM support attachment anxiety’s mediating role between both subtypes of relational aggression and sexual dissatisfaction, and all the found indirect effects are positive. Particularly, mediated by female attachment anxiety, female perceived love withdrawal and social sabotage both have an indirect association with female sexual dissatisfaction. Meanwhile, male attachment anxiety mediated three paths: (1) the indirect link between male perceived love withdrawal and male sexual dissatisfaction; (2) the indirect connection between male perceived love withdrawal and female sexual dissatisfaction; (3) the indirect association between female perceived social sabotage and male sexual dissatisfaction. An explanation for these findings is that both love withdrawal and social sabotage threaten partners’ sense of safety and acceptance in the relationship (Goldstein, 2011; Linder et al., 2002). When partners’ needs for security are threatened, they experience more anxiety and have less capability in engaging and responding to each other’s needs (Johnson, 2019), including sexual intimacy and activities.

Indirect Path Effects Mediated by Attachment Avoidance

Results of the second APIMeM only support attachment avoidance’s mediating role between perceived social sabotage and sexual dissatisfaction. Specifically, female attachment avoidance mediated the indirect association between female perceived social sabotage and female sexual dissatisfaction. At the same time, male attachment avoidance mediated the indirect association between male perceived social sabotage and male sexual dissatisfaction. These findings could be potentially explained by the nature of social sabotage and attachment avoidance. Social sabotage, unlike love withdrawal, involves third parties in relational conflicts or differences, thus the perpetrator does not need to confront the relationship distress or depend on the victimized partner for intimacy and comfort (Coyne et al., 2017, 2017). This is commonly what individuals with attachment avoidance would do (Burk & Seiffge-Krenke, 2015). Avoidant individuals are generally unable to turn to their partners for support or closeness due to a lack of trust or fear of rejection (Chris Fraley et al., 2006). This could be the potential explanation that, for both females and males, their own attachment avoidance only mediated the indirect association between social sabotage and sexual dissatisfaction for the self, not for the partner.

Additionally, the results of this study are consistent with existing findings that both attachment anxiety and attachment avoidance are predictive of sexual dissatisfaction (Busby et al., 2020; Mark et al., 2018). Again, please note, the current study assesses partners’ experience/feelings of attachment insecurity (i.e., attachment anxiety and avoidance), instead of attachment styles. Furthermore, our findings also support a significant link between couple relational aggression and attachment insecurity (Allison et al., 2008; Ben-Naim et al., 2013).

Clinical Implications

The current study provides evidence that perceived relational aggression, both love withdrawal and social sabotage, has direct and indirect associations with sexual dissatisfaction for both partners in marriage. Moreover, partners’ attachment experience (i.e., attachment anxiety and avoidance) has mediating effects on the indirect paths between relational aggression and sexual dissatisfaction. Our findings have important implications for clinical practices.

First, it is important to assess relational aggression in couples. While most of the studies, as well as couple therapists in general, have an emphasis on more overt forms of aggressive behaviors in romantic relationships (Archer & Coyne, 2005; Carroll et al., 2010), such as intimate partner violence, our study demonstrates significant path effects of perceived relational aggression for both partners in marriage. Therefore, we encourage therapists working with couples to evaluate both partners’ level of relational aggression (Sandberg et al., 2018) at the early stage of therapy, so as to better construct treatment plans. Specifically, our study found associations between relational aggression and sexual dissatisfaction. For couples seeking therapy for decreasing sexual dissatisfaction or improving sexual functioning, couple therapists particularly need to deliberately incorporate the assessments and interventions on relational aggression into the therapy process. However, it is worth noting that, while the Couples Relational Aggression and Victimization Scale (Nelson & Carroll, 2006) could be used to measure relational aggression in clinical couples, it was designed for assessing relational aggression in general. Thus, new measures that specifically target relational aggression in couples seeking therapy are needed.

Second, both love withdrawal and social sabotage are essentially dysfunctional strategies for handling conflicts or differences in a relationship (Crick & Grotpeter, 1995). Thus, in addition to direct clinical interventions on those aggressive behaviors, couple therapists should also help both partners gain healthier methods in handling conflicts or differences, such as effective communication skills. Besides, our results show that, love withdrawal and social sabotage have different path effects in the relationship. Therefore, it is necessary for couple therapists to differentiate these two forms of relational aggression, and target its effects in the relationship with corresponding interventions. For example, partners with a high level of social sabotage might be better off with a direct negotiation on the boundaries of relational privacy.

Third, our study indicates that, attachment experience in a romantic relationship, both attachment anxiety and attachment avoidance, serves as significant mediating mechanisms between relational aggression and sexual dissatisfaction. This result shows the necessity of investigating partners’ attachment experience in a particular relationship (Chris Fraley et al., 2006) when they seek therapy. At the same time, therapists should pay attention to equipping couples with more adaptive attachment behaviors, and strategies for increasing a sense of security and emotional connection in the relationship, such as specific interventions from emotionally focused couple therapy models (Johnson, 2019). The study by (Giard & Woolley, 2017) suggests, through emotionally processing the behaviors and experience that created attachment insecurities in the relationship, emotionally focused couple therapy could significantly help couples experience increased attachment security and sexual intimacy.

Limitations and Future Directions

Several limitations of the current study are worth noting. First, the participants in the study were predominantly middle-class and highly educated European American couples who were in their mid-forties with a marriage length of 19.79 years. This group of participants is very stable in terms of their relational predictors and outcomes (Coyne et al., 2017). Therefore, the generalizability of the results to other sociodemographic groups may be limited. Future studies are recommended to use samples that are more ethnically diverse and representative of the general population. Additionally, the sample included only heterosexual married couples, thereby the results of this study may not be representative of non-heterosexual or cohabitating heterosexual couples. We need further research on understanding relational aggression in non-heterosexual couples, among whom perceptions, experiences, and judgments about relational aggression might vary. Second, as a community sample, it is reasonable to believe that the reported relational aggression, attachment insecurity, and sexual dissatisfaction could be significantly lower than couples who are clinically distressed. Clinicians should be judicious in using the results to inform treatment plans and procedures. Researchers who are interested in drawing further clinical implications should collect data from clinical contexts. Third, we did not include data on couple overt aggression (e.g., intimate partner violence), so we could not determine whether the covert relational aggression we measured was an extension /signal of or different from overt aggression (Bagner et al., 2007). It would be helpful to measure both covert and overt forms of couple aggression to understand their potential associations. Fourth, both the love withdrawal and social sabotage subscales measured the victim’s perceptions of their partner’s aggressive behaviors. It is important for future studies to measure perpetrators’ reports on the actual use of relational aggression. By comparing the consequences of perceived and actual use of relational aggression, we could better comprehend the adverse effects of relational aggression. Fifth, mediation is a claim of change across time. For examining the real change, longitudinal data are needed to control for prior levels (Little, 2013). Hence, the revealed mediation effects of attachment insecurity in this study could be explained by simple cross-sectional associations (Bagner et al., 2007; Sandberg et al., 2018). Again, one should interpret the revealed mediation mechanism with some caution. Future studies that aim at establishing causality should employ advanced designs such as a longitudinal mediation model (Little, 2013). Lastly, the reliability alpha of sexual dissatisfaction was relatively low (i.e., around 0.70). But considering they each had only four items and were modeled with latent variables, the reliability and results were acceptable (Bagner et al., 2007; Kline, 2016; Little, 2013).

Conclusion

This is the first study to empirically investigate the unique influences of two types of relational aggression (i.e., love withdrawal and social sabotage) on sexual dissatisfaction in marriage, with each dimension of attachment insecurity (i.e., attachment anxiety and attachment avoidance) as mediators. Findings from this study demonstrate the significant direct and indirect associations between two types of relational aggression and sexual dissatisfaction in marriage. The results also indicate that both attachment anxiety and attachment avoidance play important yet different roles in mediating the link between relational aggression and sexual dissatisfaction. We highlight the need for future studies to understand the longitudinal associations between relational aggression and sexual dissatisfaction, with attachment insecurity as the mediator.