Self- and Partner-Reported Psychopathic Traits’ Relations with Couples’ Communication, Marital Satisfaction Trajectories, and Divorce in a Longitudinal Sample

Abstract

Given that psychopathy is composed in large part by an antagonistic relational approach and is associated with many troubling interpersonally relevant outcomes, its role in romantic functioning warrants greater attention. The current study used data from a community sample of 172 newlywed couples to examine spouses’ psychopathic traits in relation to their partners’ psychopathic traits, observed communication, 4-year marital satisfaction trajectories, and 10-year divorce rates. Spouses reporting greater levels of psychopathic traits were married to partners reporting greater levels of psychopathic traits. Psychopathic traits were correlated cross-sectionally with more negative affect and less positive affect during conversations regarding sources of tension in the relationship. Longitudinally, hierarchical linear modeling of spouses’ 4-year marital trajectories indicated that psychopathic traits generally predicted lower initial and sustained marital satisfaction for spouses and their partners over time. In addition, wives’ ratings of husbands’ psychopathic traits predicted declines in husbands’ satisfaction over time and elevated 10-year divorce rates. These findings highlight the relationship impairment associated with psychopathic traits, indicate that this impairment is present from the beginning of couples’ marital trajectories, and show that psychopathic traits predict divorce. Findings also suggest that partner-ratings of psychopathic traits provide substantial incremental validity in the prediction of marital functioning outcomes relative to self-ratings. Future research on the pathways by which psychopathic traits undermine relationship functioning over time would be valuable.

Psychopathic personality traits include a number of characteristics that likely pose challenges to establishing and maintaining satisfying long-term relationships, including aggressiveness, callousness, deceitfulness, manipulativeness, egocentrism, impulsivity, and irresponsibility (Cleckley, 1941; Hare, 2006). Although psychopathy is commonly studied in forensic contexts where dyadic romantic relationships may be rarer or less easily studied, there is strong evidence that psychopathy is best conceptualized as a dimensional rather than categorical construct (Edens, Marcus, Lilienfeld, & Poythress, 2006). Accordingly, there exist many individuals with some degree of psychopathic traits who are involved in dyadic romantic relationships, raising important questions about the role of psychopathic traits in these relationships. In particular, it is important to understand how these antagonistic and largely aversive relational traits affect relationship development over time, particularly among established, committed relationships such as marital relationships. In the current study we use a trait-based measure of psychopathic traits to examine the relations between husbands’ and wives’ psychopathic traits (self- and spouse- reported) in relation to couples’ communication, marital satisfaction, and divorce in a sample of couples studied over the first 10 years of marriage.

Psychopathy and Relationship Functioning

Early research on psychopathy and marital functioning focused on associations between the MMPI-2 Psychopathic Deviate (Pd) scale and marital functioning. This work indicates that elevations on this scale are associated with marital distress and maladjustment in maritally distressed, psychiatric, and nonclinical samples (e.g., Hjemboe & Butcher, 1991; Lewak, Wakefield, & Briggs, 1985; Snyder & Regts, 1990). More recently, data from Dark Triad studies involving the simultaneous study of psychopathy, narcissism, and Machiavellianism in relation to romantic functioning have similarly revealed that psychopathic traits are associated with negative interpersonal outcomes (e.g., Paulhus & Williams, 2002). For example, individuals with psychopathic traits desire and/or experience less intimacy in their relationships (Ali & Chamorro-Premuzic, 2010), and are more likely to engage in sexual infidelity (e.g., Brewer, Hurt, James, & Abell, 2015; Jones & Weiser, 2014). These studies indicate that psychopathy is negatively associated with overall romantic relationship quality (e.g., Jonason, Li, & Czarna, 2013), as well as lower relationship satisfaction and commitment (Smith et al., 2014).

Despite this research suggesting that psychopathic traits may have a pernicious effect on relationship functioning, critical gaps remain in our understanding of psychopathic traits’ association with marital functioning over time. First, much of the work examining psychopathy’s relations with romantic functioning has been done in undergraduate dating couples. Assessing married couples is important to understand how psychopathic traits manifest in more established, committed relationships. Second, all of the research to date has been cross-sectional, which cannot address how psychopathy prospectively predicts how relationships unfold over time. Prospective, longitudinal studies that assess couples beginning in the earliest stages of marriage are needed to address important developmental questions such as whether psychopathy affects couples’ initial satisfaction, changes in satisfaction over time, or some combination of the two. Cross-sectional studies also cannot address whether psychopathic traits predict relationship dissolution.

Third, limited research to date has taken a dyadic approach to understanding how psychopathy manifests in relationships. A dyadic approach provides several advantages. First, it allows for an examination of whether psychopathic individuals marry others with similar traits (i.e., homophily). Research on dating couples provides some evidence for homophily effects for psychopathic traits (Smith et al., 2014), but cannot address whether the same would be true for married couples. Research on homophily in Big Five personality traits among newlywed couples has shown mixed results, with some studies finding a modest degree of homophily in these traits (mean = .34, Luo & Klohen, 2005) and others finding little to no evidence of homophily (mean = −.03, Watson et al., 2004). Second, a dyadic approach permits inquiry into whether one’s psychopathic traits negatively affect the relationship functioning of one’s partner (e.g., whether husbands’ psychopathic traits affect wives’ satisfaction), including whether any effects of this type hold above and beyond the effect of one’s own personality traits (e.g., husbands’ psychopathic traits affect wives’ satisfaction, independent of wives’ own psychopathic traits). To our knowledge there has been no research examining actor and partner effects of psychopathic traits on marital functioning, though research on the related construct of antisocial personality traits does indicate significant actor and partner effects on marital satisfaction in the cross-section (Stroud, Durbin, Saigal, & Knobloch-Fedders, 2010).

A dyadic approach to understanding psychopathic traits in relationships also lends itself to the use of both self- and partner-ratings as a means of more fully assessing individuals’ standing on the psychopathy construct. Self- and spouse-reports of personality traits are correlated, but only to a moderate degree (Brock, Dindo, Simms, & Clark, 2016; South, Oltmanns, Johnson, & Turkheimer, 2011). Accordingly, it is important to consider the manner in which these sources of information uniquely predict important relational outcomes. To date, no research has examined the incremental predictive utility of partner-rated psychopathy in the context of relationship functioning, but we might expect that partner-ratings may be especially important in this interdependent interpersonal context. Indeed, some evidence suggests that partner-rated personality traits are more strongly associated with self- and partner-reports of relationship satisfaction than are self-rated traits (Brock et al., 2016; Furler, Gomez, & Grob, 2014). Prior research on self- and partner-reported personality disorder symptoms, including symptoms of antisocial personality disorder, also indicates that partner-reports provide incremental validity beyond self-reports (South, Turkheimer, & Oltmanns, 2008).

Finally, there is no observational data on how psychopathic traits affect couples’ functioning. Observational coding of couples’ communication is a common paradigm in the relationship literature (e.g., Gottman, 1994), and provides an objective rating of couples’ communication. Given evidence that psychopathic individuals engage in higher levels of proactive and reactive aggression (Coccaro, Lee, & McCloskey, 2014; Miller, Rausher, Hyatt, Maples, & Zeichner, 2014) and exhibit deficiencies in reading negative emotions such as fear, pain, and disgust (Blair et al., 2004; Kosson, Suchy, Mayer, & Libby, 2002; Marsh & Blair, 2008) in general, it is likely that psychopathic traits will be associated with more observed negative and less positive communication within the context of couples’ problem-solving communication.

The Present Study

The current study addresses these gaps using a community sample of 172 couples studied over the first 10 years of marriage.¹ We examine self- and partner-ratings of psychopathic traits assessed within the first six months of marriage in relation to several relationship outcomes. First, we examined the degree of homophily found for spouses with regard to psychopathic traits, followed by self-partner agreement on the presence of psychopathic traits. We hypothesized that a small degree of homophily would be found (Smith et al., 2014) and that self-informant agreement would be in the moderate range (Brock et al., 2016; Miller, Jones, & Lynam, 2011; South et al., 2011). Second, we examined how spouses’ psychopathic traits were associated with their observed positive and negative affect during discussions about conflicts in their relationship. We hypothesized that psychopathic individuals would demonstrate lower levels of positive affect and higher levels of negative affect in these conversations. Third, we examined how psychopathic traits predicted four-year marital satisfaction trajectories, including whether psychopathic traits were associated with the initial level of satisfaction (intercepts), changes in satisfaction over time (slopes), or both. We included actor (e.g., wives’ psychopathy traits in relation to wives’ satisfaction) and partner (e.g., wives’ psychopathy traits in relation to husbands’ satisfaction) effects to determine whether any effects for psychopathy traits are seen for the individual, the partner, or both members of the couple. Last, we examined whether psychopathy traits predicted divorce over 10 years. For all of these analyses, we examined self- and partner-rated psychopathic traits to determine the more robust predictor of marital outcomes.

To examine these questions, we used a trait-based measure of psychopathic traits. A large body of research suggests that personality disorders like psychopathy can be understood as a collection of specific traits (e.g., APA, 2013), most typically described via a five-factor model (FFM) approach. Expert ratings of prototypical cases of psychopathy (Miller, Lynam, Widiger, & Leukefeld, 2001) and meta-analytic reviews of the FFM correlates of psychopathy (O’Boyle, Forsyth, Banks, Story, & White, 2015) are consistent in demonstrating the primacy of traits from the domains of agreeableness/antagonism and conscientiousness/disinhibition. Although there is debate (e.g., Lilienfeld et al., 2012; Miller & Lynam, 2012) about the importance of other traits to psychopathy, such as those related to very low neuroticism and very high extraversion (titled fearless dominance or boldness when considered together), all major models of psychopathy agree on the centrality of antagonism and disinhibition (e.g., Miller & Lynam, 2015; Patrick, Fowles, & Krueger, 2009). Accordingly, we constructed a 15-item composite of psychopathic traits based on a FFM approach (for details of the scale’s development and validation, see Supplemental Materials), and analyzed this composite in relation to couples’ relationship outcomes. FFM-based psychopathy indices have shown high convergent validity with other measures of psychopathy and predictive validity with expected life outcomes in undergraduate, community, and clinical samples (Derefinko & Lynam, 2007; Lynam & Miller, 2015; Miller & Lynam, 2003; Miller et al., 2001). Assessed separately, low levels of agreeableness and conscientiousness have been negatively associated with couples’ relationship satisfaction (Malouff, Thorsteinsson, Schutte, Bhullar, & Rooke, 2010).

Method

Participants

Couples were identified from marriage licenses filed in Los Angeles County between May 1993 and January 1994. Marriage licenses were screened to identify couples who were married for the first time, married less than 6 months, and had at least 10 years of education. Both partners had to be older than 18 years and wives had to be younger than 35 years (to allow for the possibility that all couples might become parents over the course of the study). Couples who met criteria were sent a letter requesting that they return a postcard if they wanted to participate. Of the 3,606 letters that were sent, 637 couples (17.8%) expressed interest in participating. Interested couples were interviewed by telephone to ensure that they met all inclusion criteria, including the additional criteria that they had no children, were not currently expecting a child, could read and speak English, were living together, and had no plans to leave the area. Eligible couples were invited to participate. The first 172 couples who met the screening criteria and kept their scheduled laboratory appointment comprised the sample.

At the time of data collection, husbands averaged 27.6 years of age (SD = 3.9) and 15.6 years of education (SD = 2.2), with a median income between $21,000 and $30,000. Sixty-seven percent were Caucasian, 15% were Latino-Chicano, 13% were Asian American–Pacific Islanders, and 4% were African-American. Wives averaged 26.0 years of age (SD = 3.4) and 16.2 years of education (SD = 2.0), with a median income between $11,000 and $20,000. Sixty-one percent were Caucasian, 16% were Latina-Chicana, 15% were Asian American–Pacific Islanders, and 5% were African-American.

Procedures

Eligible couples participated in a 3-hour lab session within 6 months of their wedding, in which they completed questionnaires and interaction tasks (Time 1). After 6 months, spouses completed and returned questionnaires via mail (Time 2), and 6 months later participated in a second lab session (Time 3). Assessments at Times 4–8 were conducted via mail at 6-month intervals. Relationship status (intact versus divorced) was assessed in telephone calls prior to the regular assessments and assessed again after 10 years of marriage. Couples were paid $75 for lab sessions and $25 at each follow-up. Self- and partner-rated psychopathic traits and observed communication were assessed at Time 1. Marital satisfaction was assessed at Times 1–8.

Measures

Psychopathic traits

We used a trait-based composite of psychopathy constructed using 15 items from the NEO-FFI (Costa & McCrae, 1992; for details of the development and validation of this scale, see Supplemental Materials).² Self- and partner-ratings of FFI personality were assessed in order to enhance the measurement of psychopathic traits and examine the incremental utility of self- and partner-rated psychopathy. This measure can range from 15–75.

Psychopathic traits were normally distributed for all indices. In the current sample, α’s were .70 for husbands’ self-rated psychopathic traits, .79 for husbands’ partner-rated psychopathic traits (i.e., wives’ ratings of their husbands), .73 for wives’ self-rated psychopathic traits, and .75 for wives’ partner-rated psychopathic traits (i.e., husbands’ ratings of their wives). Mean scores on the psychopathic traits indices were 33.38 (SD= 5.59) for husbands’ self-rated psychopathy (range: 18–55), 29.35 (SD = 7.05) for husbands’ partner-rated psychopathy (range: 15–55), 29.12 (SD = 5.68) for wives’ self-rated psychopathy (range: 17–44), and 30.11 (SD = 6.19) for wives’ partner-rated psychopathy (range: 16–45). Mean levels of psychopathic traits in the derivation samples (see Supplemental Materials) were 34.00 (SD = 7.65, range: 15–62) and 33.74 (SD = 7.20, range: 17–51) for Samples 1 and 2, respectively.

Observed communication

In Time 1 lab sessions, each spouse identified a source of tension in the relationship to discuss with the partner for 10 minutes. Couples then discussed these two problems in two separate discussions in a randomly predetermined order for 10 min apiece. In rare instances when spouses chose the same topic for discussion, that topic was assigned to the spouse whose topic was chosen to be first and the other spouse’s second choice of topic was then used for the second discussion. At the beginning of each conversation, couples were instructed to “discuss the topic for 10 min and try to work toward a mutually satisfying solution.” Displays of specific affect during both conversations were coded using the Specific Affect (SPAFF) Coding System Version 1.0 (Gottman & Krokoff, 1989), modified such that two codes, interest/curiosity and anticipation/ surprise/excitement/enjoyment/joy, were combined a priori into one code labeled interest. Following SPAFF procedures, trained coders were instructed to emphasize facial expressions, posture, gestures, and voice tone and pitch when coding affect; verbal content was not sufficient by itself for coding a specific affect. The 10-min discussions were divided into 5-s units to allow for the possibility that multiple emotions could occur in a speaking turn. Each unit was coded as displaying one of five negative affects (anger, contempt, whining, sadness, or anxiety), one of three positive affects (humor, affection, or interest), or neutral affect (i.e., indiscernible or subthreshold). Whining was dropped from analyses because reliability for this code was relatively low. Sadness and anxiety were dropped from analyses because they were observed infrequently. Factor analysis of a rating system based on the SPAFF system indicated that all three of the positive codes load on the same factor and that anger and contempt load on the same factor (e.g., Johnson, 2002). Thus, the affect variables analyzed here are positive affect (defined as the sum of humor, affection, and interest) and negative affect (defined as the sum of anger and contempt). The analysis of interobserver reliability yielded intraclass correlation (ICC) coefficients of .66 (husbands’ negative affect), .83 (husbands’ positive affect), .91 (wives’ negative affect), and .68 (wives’ positive affect).

Marital satisfaction

The Quality of Marriage Index (QMI; Norton, 1983) asks spouses to report the extent to which they agree or disagree with general statements about their marriage. Higher scores on this 6-item scale reflect greater satisfaction. The QMI was assessed every six months for four years. Coefficient alpha was > .90 for husbands and wives across assessments. Descriptive statistics for couples’ communication and marital satisfaction are shown in Table 1.

Table 1.

Descriptive Statistics

	Husbands Mean (SD)	Wives Mean (SD)
Marital satisfaction
Time 1	41.32 (4.41)	41.60 (4.28)
Time 2	39.62 (6.36)	40.52 (5.70)
Time 3	40.67 (5.30)	40.93 (5.88)
Time 4	39.75 (6.25)	40.52 (6.13)
Time 5	39.66 (6.16)	40.01 (6.95)
Time 6	39.02 (7.27)	39.10 (7.41)
Time 7	38.57 (7.05)	39.22 (7.14)
Time 8	38.85 (6.89)	38.57 (7.46)
Observed communication
Positive affect	7.08 (8.15)	7.74 (8.48)
Negative affect	11.16 (18.72)	18.83 (24.90)

Analytic Plan

We examined how husbands’ and wives’ psychopathic traits were associated with communication behavior at Time 1 and marital satisfaction over four years by conducting a series of multilevel models using the HLM 7.0 computer program. We used the two-intercept model outlined by Raudenbush, Brennan, and Barnett (1995; see additional discussion in Atkins, 2005). This model simultaneously estimates separate effects for husbands and wives while accounting for the interdependence between them.

To examine the effects of husbands’ and wives’ psychopathic traits on communication, we analyzed the following cross-sectional multilevel model:

$$Y_{\text{ti}}(\text{Communication})={\mathrm{\pi }}_{\mathrm{f}1\mathrm{i}}{(\mathit{\text{wife}})}_{\text{ti}}+{\mathrm{\pi }}_{\mathrm{m}1\mathrm{i}}{(\mathit{\text{husband}})}_{\text{ti}}+{\mathrm{e}}_{\text{tij}}$$ Level 1

$${\mathrm{\pi }}_{\mathrm{f}1\mathrm{i}}(\text{wife})={\mathrm{\beta }}_{\mathrm{f}10}+\text{husband psychopathy}+\text{wife psychopathy}+{\mathrm{\mu }}_{\mathrm{f}1\mathrm{i}}{\mathrm{\pi }}_{\mathrm{m}1\mathrm{i}}(\text{husband})={\mathrm{\beta }}_{\mathrm{m}10}+\text{husband psychopathy}+\text{wife psychopathy}+{\mathrm{\mu }}_{\mathrm{m}1\mathrm{i}}$$ Level 2

Husbands’ and wives’ psychopathic traits were included simultaneously at Level 2, allowing us to examine how spouses’ communication behaviors were affected by their own psychopathic traits (actor effects) and by their partner’s psychopathic traits (partner effects). For example, we tested whether husbands’ positive affect was predicted by their own psychopathic traits (actor effect) and by their wives’ psychopathic traits (partner effect). Of note, in these types of models, when partner effects are significant, these effects are above and beyond actor effects.

We conducted these analyses using spouses’ self-rated psychopathic traits (self-ratings) and using spouses’ partner-rated psychopathic traits (partner-ratings).³ We initially ran four separate models, one for each of the communication behaviors of interest (positive and negative affect, at Level 1), run separately for self-ratings and partner-ratings (entered at Level 2). We then ran two additional models (one for positive affect and one for negative affect) in which we entered self-ratings and partner-ratings simultaneously at Level 2. These additional models permit inquiry into whether self- and partner-ratings serve as unique predictors of positive and negative affect (Brock et al., 2016).

To examine four-year satisfaction trajectories, we used an extension of this model adapted for growth curves. Growth curve analyses allow for a two-level process in data analysis. Level 1 allows for the estimation of within-subject trajectories of change (growth curve) for a variable, described by two parameters: an intercept (initial level of the variable) and a slope (rate of change over time). Level 2 allows for the examination of between-subjects differences in these parameters using individual-level predictors. We used the following model:

$$Y_{\text{ti}}(\text{Satisfaction})={(\mathit{\text{wife}})}_{\text{ti}}[{\mathrm{\pi }}_{\mathrm{f}0\mathrm{i}}+{\mathrm{\pi }}_{\mathrm{f}1\mathrm{i}}{(\mathit{\text{Time}})}_{\text{ti}}]+{(\mathit{\text{husband}})}_{\text{ti}}[{\mathrm{\pi }}_{\mathrm{m}0\mathrm{i}}+{\mathrm{\pi }}_{\mathrm{m}1\mathrm{i}}{(\mathit{\text{Time}})}_{\text{ti}}]+{\mathrm{e}}_{\text{tij}}$$ Level 1

$${\mathrm{\pi }}_{\mathrm{f}0\mathrm{i}}(\text{wife intercept})={\mathrm{\beta }}_{\mathrm{f}00}+\text{husband psychopathy}+\text{wife psychopathy}+{\mathrm{\mu }}_{\mathrm{f}0\mathrm{i}}{\mathrm{\pi }}_{\mathrm{f}1\mathrm{i}}(\text{wife slope})={\mathrm{\beta }}_{\mathrm{f}10}+\text{husband psychopathy}+\text{wife psychopathy}+{\mathrm{\mu }}_{\mathrm{f}1\mathrm{i}}{\mathrm{\pi }}_{\mathrm{m}0\mathrm{i}}(\text{husband intercept})={\mathrm{\beta }}_{\mathrm{m}00}+\text{husband psychopathy}+\text{wife psychopathy}+{\mathrm{\mu }}_{\mathrm{m}0\mathrm{i}}{\mathrm{\pi }}_{\mathrm{m}1\mathrm{i}}(\text{husband slope})={\mathrm{\beta }}_{\mathrm{m}10}+\text{husband psychopathy}+\text{wife psychopathy}+{\mathrm{\mu }}_{\mathrm{m}1\mathrm{i}}$$ Level 2

These equations include separate intercepts and linear slopes for men and women, and sex-specific variance components at Level 2. As with the communication analyses, husbands’ and wives’ psychopathic traits were included simultaneously at Level 2, allowing us to examine how the parameters of spouses’ marital satisfaction trajectories were affected by their own psychopathic traits and by their partner’s psychopathic traits (e.g., husbands’ marital satisfaction predicted by their own psychopathic traits and by their wives’ psychopathic traits). We analyzed three models, one with self-ratings at Level 2, a second with partner-ratings at Level 2, and a third with self- and partner-ratings simultaneously at Level 2. The third model allowed us to assess the extent to which self- and/or partner-ratings were unique predictors of four-year satisfaction trajectories.

Results

Homophily and Self-Other Agreement

We began by examining the degree of homophily (i.e., similarity) in spouses’ ratings of psychopathic traits, as well as self-other agreement (i.e., the extent to which their self-ratings of personality corresponded to their partner’s ratings of their personality). Husbands’ and wives’ self-ratings were significantly correlated, r(172) = .21, p < .01, indicating a small degree of homophily for psychopathic traits. There was also a moderate degree of self-other agreement: husbands’ self-ratings of psychopathic traits were correlated with how their wives rated them, r(170) = .46, p < .01, and wives’ self-ratings of psychopathic traits were correlated with how their husbands rated them, r(170) = .41, p < .01.

Psychopathic Traits and Couples’ Communication Behavior

Next we examined the association between spouses’ psychopathic traits and their own observed communication (actor effects) and with their partner’s communication (partner effects), analyzed separately using self- and partner-ratings. Results were nearly identical using self- and partner-ratings (Table 2). In both models, wives’ psychopathic traits were associated with lower levels of observed positive affect for themselves and for their husbands, and higher levels of observed negative affect for themselves. Wives’ psychopathic traits were also associated with higher levels of observed negative affect for their husbands, but only when using partner-ratings of personality (i.e., husbands’ ratings of their wives’ psychopathic traits). In addition, husbands’ psychopathic traits were associated with higher levels of observed negative affect for themselves and for their wives in self- and partner-rating models.

Table 2.

Husbands’ and Wives’ Self- and Partner-Rated Psychopathic Traits and Observed Communication

	Husbands’ Communication			Wives’ Communication
	Β (SE)	t ratio	Effect size r	Β (SE)	t ratio	Effect size r
Self-Ratings of Psychopathic Traits
Results for Positive Affect
Husband psychopathy	0.02 (0.17)	0.09	0.01	0.02 (0.15)	0.12	0.01
Wife psychopathy	−0.25 (0.13)	−1.98*	0.15	−0.28 (0.13)	−2.15*	0.16
Results for Negative Affect
Husband psychopathy	0.98 (0.29)	3.43**	0.26	1.13 (0.28)	4.09**	0.30
Wife psychopathy	0.21 (0.26)	0.81	0.06	0.87 (0.35)	2.48*	0.19
Partner Ratings of Psychopathic Traits
Results for Positive Affect
Husband psychopathy	−0.03 (0.09)	−0.37	0.03	−0.06 (0.08)	−0.73	0.06
Wife psychopathy	−0.27 (0.10)	−2.70**	0.20	−0.26 (0.10)	−2.65**	0.20
Results for Negative Affect
Husband psychopathy	0.67 (0.21)	3.13**	0.24	1.31 (0.32)	4.04**	0.30
Wife psychopathy	0.79 (0.24)	3.27**	0.25	0.84 (0.27)	3.11**	0.23

Notes. Husbands’ and wives’ psychopathic traits were grand-mean centered. N = 170 couples. df = 167 for all analyses. Self- and partner-ratings were examined in separate equations. Effect size r = sqrt [t²/(t² + df)].

^*p < .05.

^**p < .01.

To examine whether there were unique effects of self- and partner-ratings on communication, we analyzed a third model in which self- and partner-ratings were included simultaneously at Level 2. For positive affect, model fit (deviance = 2218.45, parameters = 14) for the combined model did not differ from the model using only self-ratings [deviance = 2224.21, parameters = 10, χ² (4) = 5.75, p > .10] or from the model using only partner-ratings [deviance = 2222.15, parameters = 10, χ² (4) = 3.70, p > .10]. For negative affect, model fit (deviance = 2924.09, parameters = 14) for the combined model was significantly better than the model using only self-ratings [deviance = 2944.32, parameters = 10, χ² (4) = 20.23, p < .001] but not compared to the model using only partner-ratings [deviance = 2929.15, parameters = 10, χ² (4) = 5.06, p > .10]. Results are shown in Table 3. There were no significant effects for self-ratings in this model, but partner-ratings remained significant in the manner described above, indicating that partner-ratings explained variance in observed communication above and beyond self-ratings. Specifically, husbands and wives exhibited lower levels of positive affect and higher levels of negative affect if wives were rated by their husbands as being higher in psychopathic traits. Husbands and wives also exhibited higher levels of negative affect if husbands were rated by their wives as being higher on psychopathic traits.

Table 3.

Unique Effects of Husbands’ and Wives’ Self- and Partner-Rated Psychopathic Traits on Observed Communication

	Husbands’ Communication			Wives’ Communication
	Β (SE)	t ratio	Effect size r	Β (SE)	t ratio	Effect size r
Results for Positive Affect
Husband psychopathy (self-rated)	0.15 (0.18)	0.87	0.07	0.17 (0.16)	1.01	0.08
Wife psychopathy (self-rated)	−0.14 (0.13)	−1.07	0.08	−0.17 (0.13)	−1.28	0.10
Husband psychopathy rated by wives	−0.05 (0.09)	−0.59	0.05	−0.08 (0.09)	−0.87	0.07
Wife psychopathy rated by husbands	−0.27 (0.10)	−2.65**	0.20	−0.26 (0.11)	−2.46*	0.19
Results for Negative Affect Affect
Husband psychopathy (self-rated)	0.37 (0.31)	1.19	0.09	0.27 (0.35)	0.77	0.06
Wife psychopathy (self-rated)	−0.21 (0.26)	−0.79	0.06	0.33 (0.38)	0.88	0.07
Husband psychopathy rated by wives	0.59 (0.23)	2.52*	0.19	1.14 (0.39)	2.93**	0.22
Wife psychopathy rated by husbands	0.73 (0.28)	2.64**	0.20	0.63 (0.31)	2.08*	0.16

Notes. Husbands’ and wives’ psychopathic traits were grand-mean centered. Self- and partner-ratings were examined simultaneously. N = 170 couples. df = 165 for all analyses. Effect size r = sqrt [t²/(t² + df)].

^*p < .05.

^**p < .01.

Psychopathic Traits and Couples’ Four-Year Marital Satisfaction Trajectories

We then examined how self- and partner-rated psychopathic traits predicted couples’ marital satisfaction trajectories over the first four years of marriage. Results for wives’ psychopathic traits were consistent when using self- and partner-ratings (Table 4). In both models, higher levels of wives’ psychopathic traits predicted lower initial levels of satisfaction (intercepts) for themselves and for their husbands. Results for husbands’ psychopathic traits varied depending on whether self- or partner-ratings of personality were used. Using self-ratings, husbands’ psychopathic traits were associated with their own intercepts, such that husbands with higher levels of psychopathic traits reported lower initial levels of marital satisfaction. Using partner-ratings, husbands who were rated by their wives as having higher levels of psychopathic traits showed significantly steeper declines in marital satisfaction over time (slopes), and their wives reported lower initial satisfaction levels.

Table 4.

Husbands’ and Wives’ Self- and Partner-Rated Psychopathic Traits and Four-Year Marital Satisfaction Trajectories

	Marital Satisfaction Intercept			Marital Satisfaction Linear Slope
	Β (SE)	t ratio	Effect size r	Β (SE)	t ratio	Effect size r
Self-Ratings of Psychopathic Traits
Results for Husbands’ Satisfaction
Main effect	40.88 (0.34)	—	—	−0.41 (0.08)	−5.38**	0.38
Husband psychopathy	−0.16 (0.07)	−2.37*	0.18	0.00 (0.01)	0.19	0.01
Wife psychopathy	−0.20 (0.06)	−3.46**	0.26	−0.01 (0.01)	−0.72	0.06
Results for Wives’ Satisfaction
Main effect	41.48 (0.33)	—	—	−0.50 (0.08)	−6.34**	0.44
Husband psychopathy	−0.07 (0.06)	−1.18	0.09	−0.01 (0.01)	−1.06	0.08
Wife psychopathy	−0.16 (0.05)	−2.85**	0.22	−0.02 (0.01)	−1.50	0.12
Partner Ratings of Psychopathic Traits
Results for Husbands’ Satisfaction
Main effect	40.88 (0.34)	—	—	−0.42 (0.08)	−5.51**	0.39
Husband psychopathy	−0.06 (0.04)	−1.46	0.11	−0.03 (0.01)	−2.79**	0.21
Wife psychopathy	−0.25 (0.06)	−4.61**	0.34	0.02 (0.01)	1.30	0.10
Results for Wives’ Satisfaction
Main effect	41.47 (0.32)	—	—	−0.50 (0.08)	−6.36**	0.44
Husband psychopathy	−0.17 (0.05)	−3.44**	0.26	−0.02 (0.01)	−1.89	0.14
Wife psychopathy	−0.14 (0.05)	−2.68**	0.20	0.01 (0.01)	0.89	0.07

Notes. Husbands’ and wives’ psychopathic traits were grand-mean centered. Intercepts were significant p < .01 because the lowest possible score was greater than zero, so these statistics are not reported. Self- and partner-ratings were examined in separate equations. N = 170 couples. df = 167 for all analyses. Effect size r = sqrt [t²/(t² + df)].

^*p < .05.

^**p < .01.

We analyzed a third model in which self- and partner-ratings were included simultaneously at Level 2 to examine whether there were unique effects of self- and partner-ratings on marital satisfaction trajectories. Model fit (deviance = 13348.11, parameters = 31) for the combined model was significantly better than the model using only self-ratings [deviance = 13376.15, parameters = 23, χ² (8) = 28.04, p < .001] but not compared to the model using only partner-ratings [deviance = 13361.02, parameters = 23, χ² (8) = 12.91, p > .10]. Results are shown in Table 5. Husbands’ initial satisfaction was associated with their wives’ self- and partner-rated psychopathic traits, such that husbands had lower initial levels of marital satisfaction if they rated their wives higher in psychopathic traits and if their wives rated themselves higher in psychopathic traits. In addition, husbands’ satisfaction declined significantly more if they were rated by their wives as having higher levels of psychopathic traits. Wives’ initial satisfaction was significantly associated with partner-rated psychopathic traits only: Wives had lower initial levels of marital satisfaction if they were rated as having higher levels of psychopathic traits by their husbands and if they rated their husbands as having higher levels of psychopathic traits.

Table 5.

Unique Effects of Husbands’ and Wives’ Self- and Partner-Rated Psychopathic Traits on their Four-Year Marital Satisfaction Trajectories

	Marital Satisfaction Intercept			Marital Satisfaction Linear Slope
	Β (SE)	t ratio	Effect size r	Β (SE)	t ratio	Effect size r
Results for Husbands’ Satisfaction
Main effect	40.88 (0.34)	—	—	−0.41 (0.07)	−5.51**	0.39
Husband psychopathy (self-rated)	−0.08 (0.07)	−1.24	0.10	0.01 (0.02)	0.72	0.06
Wife psychopathy (self-rated)	−0.13 (0.07)	−1.98*	0.15	−0.01 (0.01)	−0.37	0.03
Husband psychopathy rated by wives	−0.00 (0.05)	−0.07	0.01	−0.03 (0.01)	−2.37*	0.18
Wife psychopathy rated by husbands	−0.18 (0.05)	−3.26**	0.25	0.01 (0.02)	0.75	0.06
Results for Wives’ Satisfaction
Main effect	41.47 (0.32)	—	—	−0.51 (0.08)	−6.43**	0.45
Husband psychopathy (self-rated)	0.10 (0.07)	1.44	0.11	−0.02 (0.02)	−1.25	0.10
Wife psychopathy (self-rated)	−0.05 (0.06)	−0.81	0.06	−0.03 (0.01)	−1.75	0.13
Husband psychopathy rated by wives	−0.19 (0.06)	−3.23**	0.24	−0.01 (0.01)	−0.53	0.04
Wife psychopathy rated by husbands	−0.16 (0.06)	−2.55*	0.19	0.03 (0.01)	1.82	0.14

Notes. Husbands’ and wives’ psychopathic traits were grand-mean centered. Intercepts were significant p < .01 because the lowest possible score was greater than zero, so these statistics are not reported. Self- and partner-ratings were examined simultaneously. N = 170 couples. df = 165 for all analyses. Effect size r = sqrt [t²/(t² + df)].

^*p < .05.

^**p < .01.

Psychopathic Traits and 10-Year Divorce Rates

Last, we examined the association between self- and partner-rated psychopathic traits and 10-year divorce rates using binary logistic regressions. Twenty-two percent of couples divorced by 10-year follow-up (n = 38). Husbands’ partner-rated psychopathic traits (i.e., husbands’ traits rated by their wives) were a significant predictor of divorce, b = .06 (SE = .03), Odds ratio = 1.07, p = .02. Husbands’ self-rated psychopathic traits, wives’ self-rated psychopathic traits, and wives’ partner-rated psychopathic traits (i.e., wives’ traits rated by their husbands) did not significantly predict divorce (p > .10). When all four variables were entered simultaneously, husbands’ partner-rated psychopathic traits remained a significant predictor of divorce, b = .07 (SE = .03), Odds ratio = 1.07, p = .04.

Discussion

Psychopathic personality traits include a number of characteristics that have been associated with relationship dysfunction, raising important questions about how these traits affect marital relationships. These data from 172 couples assessed over the first 10 years of marriage provide several new insights into the role of psychopathic personality traits in marriage.

First, at the initial assessment, individuals self-reporting higher levels of psychopathic traits tended to be married to partners who also self-reported higher levels of psychopathic traits, consistent with the hypothesis that individuals with psychopathic personality engage in some degree of assortative mating. The current findings are in line with previous work demonstrating homophily for psychopathic traits (r = .39; n = 45 couples) for undergraduate dating couples (Smith et al., 2014), but also indicate a more modest level of homophily for these traits in our married sample compared to the undergraduate dating sample. Previous research on narcissism in two other samples similarly showed more consistent evidence of homophily among dating couples (Lamkin, Campbell, vanDellen, & Miller, 2015) than among married couples (Lavner, Lamkin, Miller, Campbell, & Karney, 2016). Accordingly, across these studies, there is evidence that homophily effects for maladaptive traits such as psychopathy or narcissism may be stronger among dating couples than among married couples, reflecting the possibility that the types of partners with whom individuals high in these types of maladaptive traits are willing or able to form less serious dating relationships may differ from marriage relationships, or that these pairings are not stable enough to result in marriage. These small effects in homophily for psychopathic traits in our sample are consistent with previous research showing a small degree of assortative mating in personality among newlywed couples (Luo & Klohen, 2005; Watson et al., 2004). Given the small degree of homophily seen here, it will be important to understand the mate selection processes leading to these pairings. For example, it would be interesting to examine whether this type of assortative mating occurs because individuals with higher levels of psychopathic traits view these traits as being likable, as is the case for narcissism (Hart & Adams, 2014), or because they view them as being less dislikable (Lamkin & Miller, 2016). Such research would help determine whether individuals with psychopathic traits might purposely seek out other individuals with these same traits, or whether they end up partnered with these individuals, in part, because they do not see these traits as being detrimental and thus do not exclude someone for consideration as an intimate partner.

Second, psychopathic personality traits were correlated with observed positive and negative affect during discussions about sources of tension in the relationship. In general, wives’ psychopathic traits were associated with lower levels of observed positive affect (i.e., humor, affection and interest) and greater negative affect (i.e., anger and contempt) for themselves and for their husbands, and husbands’ psychopathic traits were associated with higher levels of observed negative affect for themselves and for their wives. The communication difficulties associated with high psychopathic traits are likely multifactorial in nature, and could reflect problems associated with greater interpersonal antagonism (e.g., callousness towards partners’ feelings; egocentrism; deceptiveness), difficulties with impulse control and delay of gratification, as well as difficulties with recognizing partners’ emotional distress (e.g., Blair et al., 2004). That is, individuals with high levels of psychopathic traits may not be concerned if their communication approach causes their partner distress, and even if they are, they may be less capable of detecting these affective states and changing course so as to mitigate these experiences, resulting in the more aversive emotional states (more negativity, less positivity) observed here.

Third, the use of longitudinal data on marital quality allowed for a prospective examination of how psychopathic traits affected how marriages unfolded over time. Overall, results generally indicated that the impairment associated with psychopathic traits was typically present from the beginning of couples’ marriages, such that wives’ and husbands’ psychopathic traits predicted lower initial levels of marital satisfaction for themselves and for their partners. Husbands’ partner-rated psychopathic traits (i.e., husbands’ traits rated by wives) were the only significant predictor of steeper declines in satisfaction over time, and only for husbands’ own satisfaction. These findings are consistent with the enduring dynamics model of marital functioning in which couples’ difficulties can be traced to factors present early in marriage (Huston, Caughlin, Houts, Smith, & George, 2001). Future research is needed to explore the mechanisms that explain lower initial and sustained levels of satisfaction and husbands’ declines in satisfaction over time. For example, to the extent that psychopathy has been linked to lower interest in committed relationships (e.g., Jonason, Luevano, & Adams, 2012), maintaining intimacy (Ali & Chamorro-Premuzic, 2010), and maintaining sexual fidelity (e.g., Brewer et al., 2015), being involved in a committed marital relationship may be less satisfying for individuals with higher levels of psychopathic traits.

Fourth, the use of longitudinal data on divorce allowed for a prospective examination of how psychopathic traits affected the longevity of marriages. Strikingly, wives’ ratings of their husbands’ psychopathic traits were associated with higher rates of divorce within 10 years, indicating that wives who view their husbands as more antagonistic and disinhibited have marriages that are more likely to end, although the effect was modest in size. These findings are consistent with prior longitudinal findings showing that low levels of agreeableness and conscientiousness (assessed separately) lead to marital dissolution (Solomon & Jackson, 2014). Future research is needed to examine why wives’ ratings of their husbands’ psychopathic traits prove to be especially powerful predictors of marital dissolution. For example, to the extent that newlywed wives have previously been shown to value agreeableness and conscientiousness more than newlywed husbands (Botwin, Buss, & Shackelford, 1997), wives who view their husbands as low in these traits (i.e., have higher levels of psychopathy) may be especially dissatisfied over the long-term with their marriages and more likely to initiate divorce. Alternatively, the fact that wives’ ratings of husbands’ psychopathic traits predicted declines in husbands’ satisfaction but not initial levels of satisfaction could suggest that the negative effects of wives’ views on their husbands’ satisfaction take time to unfold, and that these men find being married increasingly dissatisfying as time passes.

Finally, one of the strengths of the current study was the simultaneous use of both self- and informant/partner-ratings in order to more thoroughly capture individuals’ standing on the psychopathy construct. The multi-method assessment is especially important in the case of psychopathic traits since questions persist as to whether psychopathic individuals possess the insight and/or willingness to accurately report on their personality (cf. Miller et al., 2011) and whether informant-ratings of psychopathy explain incremental variance in behavioral outcomes related to psychopathy (e.g., Fowler & Lilienfeld, 2007; Jones & Miller, 2012). Our results indicated a moderate degree of self-other agreement in ratings of psychopathic traits, consistent with previous findings showing a moderate level of agreement between spouses on personality traits (Brock et al., 2016; South et al., 2011). The examination of the unique effects of self- and partner-ratings of psychopathic traits when all four variants of self- and partner-ratings were placed in the model together allowed for a test of each predictor’s unique incremental utility in the marital functioning outcomes. Although self-ratings of psychopathic traits were generally predictive of affective experiences and marital outcomes (8 significant effects), these effects mostly disappeared when examined simultaneously with partner ratings of psychopathic traits (only 1 of 8 effects remained significant), suggesting that partner-ratings but not self-ratings provide substantial incremental validity in the study of outcomes associated with psychopathy studied in a marital context. These findings offer perhaps the strongest support yet seen for the incremental validity of informant-ratings in predicting outcomes related to psychopathic personality, building upon previous findings in which more limited incremental validity was found (Fowler & Lilienfeld, 2007; Jones & Miller, 2012). While incremental validity has been limited in studies to date on psychopathy, these findings are consistent with data found on the substantial incremental validity of informant ratings for other related personality disorders such as antisocial personality disorder (e.g., Miller, Pilkonis, & Clifton, 2005; South et al., 2008; Stroud et al., 2010) and in relationship research more generally (e.g., Brock et al., 2016).

Limitations

Despite these many strengths, it is important to acknowledge some of the study’s limitations. First, although psychopathic traits were assessed using an empirically derived and validated measure of personality traits known to be strongly associated with psychopathy, the study would have been enhanced by including a direct and explicit measure of psychopathy. Nonetheless, there is a significant history of using personality-based proxies of psychopathy in the empirical literature (e.g., Benning, Patrick, Blonigen, Hicks, & Iacono, 2005; Blonigen, Hicks, Krueger, Patrick, & Iacono, 2005; Miller et al., 2001; Miller & Lynam, 2003) to study important issues in samples in which explicit measures of psychopathy are not available. It is worth noting, however, that the current psychopathy proxy did not include content associated with fearless dominance/boldness, which is included in many but not all models and measures of psychopathy (e.g., Patrick et al., 2009) and is the subject of substantial debate regarding its place in the construct (e.g., Miller & Lynam, 2012; Lilienfeld et al., 2012; Miller, Lamkin, Maples-Keller, & Lynam, 2016). As such, the current data cannot speak to how fearless dominance/ boldness alone or in concert with the current content focused on antagonism/meanness and disinhibition would affect marital functioning. Second, our study captures only a portion of marriage. Additional data on marital quality beyond the fourth year of marriage, and on divorce rates beyond the tenth year of marriage, would be valuable. Third, this study was structured in such a way that self- and partner-reported personality ratings were assessed at the same time point as assessments of husbands’ and wives’ initial marital satisfaction. As a result, where traits were found to be related to initial satisfaction, we cannot determine whether the trait perceptions influenced ratings of satisfaction, whether satisfaction influenced perceptions of partners’ personality, or if these relations are bidirectional in nature.

Fourth, interpretation of results from studies that use both self- and informant-report methodologies can be difficult due to issues related to mono-method bias, wherein measuring the predictor and outcome variables using the same methodology (e.g., using self-reported psychopathic traits and self-reported marital satisfaction), may artificially inflate the effect size of tested relationships (Jones & Miller, 2012). In the current study, however, we avoided this problem in two ways: First, by examining partner-rated personality in relation to self-rated satisfaction, and second, by examining self- and partner-rated personality in relation to observed behavioral outcomes (i.e., communication) that were not contingent upon either partner’s ratings. When predicting self-ratings of marital satisfaction, there were 4 significant monomethod-based findings (e.g., self-ratings of psychopathy predicting self-ratings of satisfaction; husbands’ ratings of wives’ psychopathy predicting husbands’ satisfaction) and 3 significant heteromethod-based findings (e.g., partner-rated psychopathy predicting self-rated satisfaction). Similarly, when predicting the coded experiences of affect during the communication task, there were a mixture of both significant self-based (i.e., 5) and partner-rated (i.e., 6) effects. As such, the current findings do not seem to have been unduly influenced by issues of method variance. Finally, we studied these effects among a community sample of couples with relatively low levels of psychopathic traits. Associations between psychopathic traits and relationship quality and stability may be stronger among samples with higher levels of psychopathy.

Conclusion

Taken together, these results indicate that husbands’ and wives’ psychopathic traits are related to more negative and less positive affective experiences during couples’ communication and to lower levels of marital satisfaction, and that husbands’ psychopathic traits as rated by their wives are associated with elevated divorce rates 10 years later. These findings underscore the importance of examining relationship functioning prospectively to understand how psychopathic personality affects marital functioning, and call for further research examining the processes by which psychopathic traits exert their influence on marital dynamics and outcomes.