Abstract
This study examined the actor, partner, and (dis)similarity effects of personality on marital satisfaction, considering the interaction with marital duration. Participants were 749 Japanese married couples. The results of an actor–partner interdependence model, which includes interaction with marital duration, showed that there was little interaction between actor and partner effects and marital duration, indicating that both one’s own and the partner’s personality have consistent effects on satisfaction throughout marital duration. In contrast, similarity effects did interact with marital duration. In shorter marital duration, similar couples were less satisfied, whereas similarity had no effect on satisfaction in longer marital duration. We discuss the possible influence of the transition from a romantic relationship to a marital relationship.
Keywords
This study examined the actor, partner, and (dis)similarity effects of personality on marital satisfaction, considering the interaction of marital duration. Actor effects indicate the relationships between each person’s personality trait and his or her own marital satisfaction. Partner effects represent the linkage between a personality of each individual and his or her partner’s marital satisfaction. (Dis)similarity effects estimate whether a partner having a personality that is similar or not to one’s own is associated with higher marital satisfaction.
Personality and Marital Satisfaction
How do personality traits relate to marital satisfaction? In accord with the Vulnerability–Stress–Adaptation model (Karney & Bradbury, 1995), personal dispositions have relationships with both everyday interactions and dyadic adaptation to external stressors. These factors play an important role in romantic relationship quality and stability. Accordingly, personality traits are thought to serve as “enduring strengths and vulnerabilities” for stressful marital experiences (Gonzaga, Campos, & Bradbury, 2007, p. 35).
A lot of previous studies have reported that people’s personality traits relate to satisfaction of their own romantic relationships (e.g., Robins, Caspi, & Moffitt, 2000; Watson, Hubbard, & Wiese, 2000). Neuroticism, Agreeableness, and Conscientiousness are the personality traits with the robust actor effects on relationship satisfaction (Heller, Watson, & Hies, 2004). Neuroticism, which is linked to susceptibility to negative mood inductions (Gross, Sutton, & Ketelaar, 1998), and passive coping (Watson & Hubbard, 1996), tends to be associated with marital dissatisfaction. On the contrary, Agreeableness and Conscientiousness, which are linked to experiencing positive emotions in interpersonal interactions and self-control, tend to be associated with marital satisfaction.
Previous studies showed that not only one’s own personality but also one’s partner’s personality (partner effects) affect satisfaction. A previous study reported significant partner effects on relationship satisfaction using a meta-analysis (Malouff, Thorsteinsson, Schutte, Bhullar, & Rooke, 2010). This meta-analysis revealed that participants with higher Neuroticism scores showed lower satisfaction scores, and participants with higher Agreeableness and Conscientiousness scores showed higher satisfaction scores. Although this meta-analysis stressed only on zero-order correlations, Dyrenforth, Kashy, Donnellan, and Lucas (2010) demonstrated that these effects were maintained in marital couples even when actor effects were controlled.
(Dis)Similarity and Marital Satisfaction
Both members of a married couple influence each other, and they mutually influence on the other partner’s positive psychological factors, for example, happiness (Hoppmann, Gerstorf, Willis, & Schaie, 2011) and marital satisfaction (Gonzaga et al., 2007). The reciprocal influence possibly brought many ways in which married couples can be similar (e.g., in attractiveness, core values, and demographics). In particular, similarity of personality traits has effects on intimate relationships by coordinating partners’ thoughts and behaviors, increasing understanding of each other’s intentions and motivations, and reinforcing their appraisals (Anderson, Keltner, & John, 2003). However, spousal personality similarity is quite low in general (Watson et al., 2004). The commonly used similarity measures are discrepancy scores and profile correlations (Dyrenforth et al., 2010). The discrepancy scores represent the difference between personalities of two partners. To provide an overall index of personality discrepancy, they are calculated by taking the absolute score of the difference between both partners’ scores of personality traits (Dyrenforth et al., 2010). The profile correlation, such as an intraclass correlation (ICC), has also been used recently by some previous studies (Dyrenforth et al., 2010; Furler, Gomez, & Grob, 2013; Luo et al., 2008). The profile correlation represents the degree to which the overall personality profiles of the individuals in each couple have a resemblance to each other.
Is having a partner with similar personality traits related to high marital satisfaction? Or are we satisfied with a partner whose personality is dissimilar to ours? Some studies found that there are significant associations between indices of personality similarity and marital satisfaction (e.g., Robins et al., 2000; Russell & Wells, 1991). However, other studies have not found such kinds of relationships (e.g., Dyrenforth et al., 2010; Furler et al., 2013; Humbad, Donnellan, Iacono, McGue, & Burt, 2013). Gattis, Berns, Simpson, and Christensen (2004) found that discrepancy scores on any of the Big Five traits were not associated with marital satisfaction. Dyrenforth et al. (2010) also reported that there were almost no relationships between discrepancy score of each Big Five trait and relationship satisfaction. In addition, the analyses on overall similarity, which was measured either by an aggregate measure of discrepancy or by calculating a profile correlation coefficient, consistently provided little evidence of the importance of the personality similarity between partners (Dyrenforth et al., 2010).
Moreover, there are only a few studies regarding the importance of dissimilarity. In a value domain, Luo et al. (2008) found that the differences of values between couple were related positively with romantic relationship satisfaction. In an interpersonal domain, Markey and Markey (2007) found that romantic couples who have good quality of relationships were more similar with regard to warmth; however, they were more dissimilar about dominance than couples with low relationship quality.
Actor, Partner, (Dis)Similarity Effect and Marital Duration
Many studies have shown that a longer duration of marriage decreased marital satisfaction (e.g., Bradbury & Karney, 2004; Glenn, 1998; Kurdek, 1999; Vaillant & Vaillant, 1993; VanLaningham, Johnson, & Amato, 2001). However, the interactive relationships among actor, partner, the (dis)similarity effects, and marital duration have been unclear. Watson et al. (2000) reported that partners’ Extraversion and Agreeableness predicted the satisfaction of relationship in married couples, but that only the effect of Agreeableness was replicated in dating couples. In the same way, there is possibly an interactive effect between marital duration and other variables: actor, partner, and similarity effects. The present study is the first attempt to examine the interactive effects of marital duration.
The aim of the present study is to examine the effects of actor, partner, and (dis)similarity on marital satisfaction with consideration for the interaction between them and marital duration. In line with previous research, we expect the effects of personalities of both actor and partner on the marital satisfaction. In particular, we hypothesize that there are positive effects of actors’ and partners’ Agreeableness and Conscientiousness, and negative effects of actors’ and partners’ Neuroticism on marital satisfaction (e.g., Dyrenforth et al., 2010). For the analyses of similarity effects, we examine the effects of personality similarity on marital satisfaction for both each trait level and overall basis. According to the results of Dyrenforth et al. (2013), we expect that there would not be any relationship between the personality similarity and the marital satisfaction. A previous study, however, pointed out that role similarity may impede the goals of each partner or both partners in the later stage of marital choice (Murstein, 1972). If this impediment continues into early marriage, it is plausible that personality similarity would decrease marital satisfaction.
The participants of this study were Japanese married couples. Recent cross-cultural personality research has identified several universal personality dimensions. For example, the common factor structure of personality has been found across almost 50 cultures (McCrae, Terracciano, & 78 Members of the Personality Profiles of Cultures Project, 2005). It is possible that the association between personality and marital satisfaction also shares some common features, although there are few studies on this issue in Eastern culture. East Asian culture, including Japan, is more interdependent than Western culture (Markus & Kitayama, 1991). In the interdependent culture of Japan, marital satisfaction may depend on the partner’s personality rather than one’s own personality. Our study provides the commonalities and differences between Eastern and Western culture.
Method
Participants
A total of 749 Japanese married couples completed a web-based questionnaire. The mean age of husbands was 43.8 years (SD = 13.4), and for wives the mean was 41.9 years (SD = 12.8). Mean marriage duration was 15.17 years (SD = 12.90), ranging from 0 to 50 years.
Materials
Personality was measured using the Japanese version of the Ten-Item Personality Inventory (TIPI-J; Oshio, Abe, & Cutrone, 2012). Two items assess each of the Big Five dimensions. For each dimension, one item is positively keyed (e.g., “Extraverted, enthusiastic”) and the other item is negatively keyed (e.g., “Reserved, quiet”). Items are rated on a 7-point Likert-type scale from 1 (disagree strongly) to 7 (agree strongly). Previous research has indicated that the TIPI-J demonstrates sufficient 2-week test–retest reliability (r = .64 - .84; Oshio et al., 2012), and this scale provides an adequate representation of the Big Five dimensions of personality and correlates sufficiently well with Other Big Five scales (Oshio, Abe, Cutrone, & Gosling, 2013, 2014).
Marital satisfaction was measured using Moroi’s (1996) scale, which contains six items (e.g., “We have a good marriage”) scored on a 4-point scale from 1 (disagree strongly) to 4 (agree strongly). This scale was adapted from the Quality Marriage Index (QMI) originally developed by Norton (1983). In the present study, Cronbach’s alpha coefficients were .94 and .95 for husband and wife, respectively.
Analytic Plan
We used a general measure of (dis)similarity, which was computed by taking the average of the five univariate discrepancy scores into a single value (Dyrenforth et al., 2010). The ICC is a profile correlation that is computed by associating one person’s Big Five trait scores with scores of his or her spouse’s traits. Participants’ Big Five trait scores were standardized by using the grand mean and SDs, and the ICC was calculated using the standard scores (Dyrenforth et al., 2010).
We used the omnibus structural equation modeling (SEM) test for distinguishability (Kenny, Kashy, & Cook, 2006) and examined whether couples were distinguishable by gender. We also used the omnibus test to evaluate gender differences in means, variances, and covariances.
Hierarchical linear modeling was used to estimate the actor–partner interdependence model (APIM; Kenny et al., 2006). First, we focused on each of the Big Five traits one at a time; thus, the analyses involved estimating five separate models. Second, we focused on overall (dis)similarity using an aggregate measure of discrepancy and a profile correlation.
In line with Dyrenforth et al. (2010), actor effects indicate the relationships between each person’s personality trait and his or her own marital satisfaction. Partner effects represent the linkage between a personality of each individual and his or her partner’s marital satisfaction. Similarity effects represent the independent contribution of personality similarity to controlling the main effects of each partner’s personality (Dyrenforth et al., 2010). In this approach, the individual is treated as the lower level unit, whereas marital couple is treated as the upper level unit (Dyrenforth et al., 2010). Similarity indices are upper level predictors while actor and partner predictors are lower level predictors. The estimated general form of the lower level model was
In this formula, Yij represents the marital satisfaction for person i in couple j, Gen is used to represent gender of the person, Act is the person’s own score on one of the personality dimensions, and Part is the person’s partner’s score on the personality dimension (Dyrenforth et al., 2010). We allow for there to be differences in the size of actor or partner effects for husbands and wives by containing the interactions between gender and the actor and partner effects (Dyrenforth et al., 2010). Marital duration and the similarity index are predictors in the upper level equations:
The similarity variable, denoted as Sim in the equations, is a predictor of the intercept, b0j, and the gender effect, b1j, which means a1 assesses if similarity predicts satisfaction on average, and c1 assesses if the effect of similarity on marital satisfaction varies between husbands and wives (Dyrenforth et al., 2010). Marital duration, denoted as Dur in the equations, is another predictor of the intercept. In addition, we included marital duration as a moderator of the actor, partner, and similarity effects. In the simple slope analysis of the interaction with marital duration, we conceptualize 1 SD above the mean as longer duration, and 1 SD below the mean as shorter duration. Finally, the nonindependence of satisfaction scores between the two partners was modeled in the random component of the intercepts, dj (Dyrenforth et al., 2010).
Results
Descriptive Statistics
Table 1 shows descriptive statistics for each of the Big Five personality traits, trait-by-trait discrepancies, and aggregated similarity indices.
Means and Standard Deviations for the Big Five Predictors, Individual Discrepancy Scores, and the Indices of Personality (Dis)Similarity.
Predicting Marital Satisfaction Using Actor Personality, Partner Personality, and Personality Discrepancies for Individual Traits
We first examined the relationships between marital satisfaction and actor, partner, and similarity effects on a trait-by-trait basis (Table 2). Thus, this set of analyses involved estimating five separate models. As the results of the omnibus SEM test for distinguishability, couples were in fact distinguishable by gender, omnibus test: χ2(6) = 29.18 to 115.43, p < .001. A positive effect for discrepancy on satisfaction indicates that less similar couples show more satisfaction. Analyses also included the main effect of gender and marital duration (Step 1), and these analyses allowed the actor, partner, and discrepancy effects to differ by gender (Dyrenforth et al., 2010) and marital duration (Step 2). The main effect for gender was statistically significant (p < .01), indicating that husbands had higher marital satisfaction scores than did wives (M = 2.99, SD = 0.65 for husbands, and M = 2.90, SD = 0.67 for wives), t(748) = 4.97, d = 0.14. The main effect for marital duration was also statistically significant (p < .05), indicating the longer participants were married, the less satisfied they were with their marriage.
Regression Coefficients Predicting Marital Satisfaction Using Separate Trait-by-Trait Analysis.
p < .10. *p < .05. **p < .01.
Considering first the actor effects, people who showed higher Agreeableness, Extraversion, and Conscientiousness, and lower Neuroticism reported higher marital satisfaction. Second, the partner effects indicate that a person’s marital satisfaction was linked with having a partner with high Agreeableness, Extraversion, Openness, and low Neuroticism. Third, dissimilarity effects emerged for three traits—Agreeableness, Neuroticism, and Openness, which means that couples with less similarity of these traits were more satisfied. A gender interaction with actor’s Agreeableness emerged such that agreeable wives were especially satisfied (b = .129, p < .001), whereas the level of the husband’s Agreeableness was not as strongly associated with his own satisfaction (b = .054, p <.05). A gender interaction with partner’s Agreeableness emerged such that husbands with more agreeable wives were especially satisfied (b = .161, p < .001), whereas the level of the husband’s Agreeableness was less predictive of the wife’s marital satisfaction (b = .059, p <.05).
Moreover, marital duration interactions with (dis)similarity effects emerged for three personality dimensions: Extroversion, Conscientiousness, and Neuroticism. These interactions had the same direction. In shorter marital duration, similar couples were less satisfied (i.e., 1 SD below the mean; b = .061, p <.05; b = .077, p <.05; b = .123, p <.001, respectively), whereas similarity had no effect on marital satisfaction in longer marital duration (i.e., 1 SD above the mean; b = −.008, ns; b = −.013, ns; b = −.018, ns, respectively).
Predicting Marital Satisfaction Using Overall Discrepancy
Next, we examined whether overall discrepancy of personality predicted marital satisfaction above and beyond the actor and partner effects (Table 3). Again, couples were in fact distinguishable by gender, omnibus test: χ2(42) = 175.00, p < .001. We included the actor effects and partner effects for all five personality traits to control for the main effects of the partners’ personalities (Dyrenforth et al., 2010). The main effects of gender and marital duration (Step 1) and the interactions of gender and marital duration with overall discrepancy were also included in the model (Step 2). We also conducted the analysis including gender and marital duration interactions with actor and partner effects, but it did not change any of the resulting conclusions.
Regression Coefficients Predicting Marital Satisfaction Using Overall Discrepancy.
Note. Actor effects and partner effects for all five personality traits were included to control for the main effects of the partners’ personalities.
p < .05. **p < .01.
After controlling for the main effects of actors’ and partners’ personality, the overall discrepancy was positively related to marital satisfaction. In addition, marital duration interactions with overall discrepancy emerged (Figure 1). Consistent with the trait-by-trait analysis, in shorter marital duration, similar couples were less satisfied (b = .194, p < .001), whereas similarity had no effect on satisfaction in those with a longer marital duration (b = −.047, ns).

Moderation of the relationship between personality discrepancy and marital satisfaction by marital duration.
We also executed the same analysis using the ICC based on the standardized scores as a profile correlation, instead of overall discrepancy. However, the ICC similarity measure was not statistically significant for the main effect and the interaction with gender and marital duration.
Discussion
The goal of the present study was to examine the actor, partner, and (dis)similarity effects of personality traits on marital satisfaction, by considering the interaction of marital duration. We found medium-sized actor and partner effects. With regard to main effects for personality traits, Neuroticism, Agreeableness, and Conscientiousness were the traits which have the most consistent actor and partner effects for predicting marital satisfaction. Neuroticism was negatively and Agreeableness and Conscientiousness were positively associated with marital satisfaction. These results are consistent with the existing literature (e.g., Dyrenforth et al., 2010) and provide support for the idea that Neuroticism, Agreeableness, and Conscientiousness are personality traits that statistically predict marital satisfaction. In addition, Extroversion was significant for actor and partner effects, and Openness was significant for partner effects.
The results in Japan showed that the patterns of actor and partner effects are consistent across Western and Eastern cultures. Most of the previous research was conducted in a Western culture. This study replicated the findings in an Eastern culture and with the use of a highly abbreviated measure of the Big Five.
On the contrary, unlike previous studies (e.g., Dyrenforth et al., 2013), partner effects were as strong as actor effects in the present study. As we noted in introduction, interdependency in East Asian culture is one of the possible reason of this results. In this culture, people may be more sensitive to their partners’ kindness and conscientiousness.
Agreeableness interacted with gender, such that wives’ agreeableness influenced husband’s marital satisfaction more strongly than the other way around. A previous study in Japan (Ito, Ikeda, & Kawaura, 1999) found that husbands were more dependent on their wives than vice versa. As Ito et al. (1999) addressed, for husbands, sources of support are rare and the wife may be their only source of support. As such, husbands need their wives as a source of support more so than wives need their husbands for support.
There was the only one interaction between actor and partner effects and marital duration, indicating that both one’s own and one’s partner’s personalities had consistent effects on satisfaction throughout the duration of the marriage.
Similar couples were less satisfied only in the instance of short marital duration. The reason for this incongruent result when compared with previous research may be due to the average length of marriage in this study, which was shorter than that found in previous research. For example, the average marital duration in this study was over 5 years shorter than in Dyrenforth et al. (2013). As we noted in introduction, role similarity may impede the goals of one or both partners in early marriage, which in turn, decrease marital satisfaction.
The present study showed the similarity effect using the discrepancy score was significant, whereas the effect using ICC was not significant. In part, this may be because a profile consists of three elements (Furler et al., 2013; Furr, 2010). First, every profile has a shape that represents the pattern of scores in a profile. Second, elevation represents the overall mean across all traits within a profile. Finally, scatter refers to the variability or variance among the scores of a profile. ICC captures all three characteristics of a profile at once (Dyrenforth et al., 2010; Furler et al., 2013). Of these elements, discrepancy scores represent only elevation. Thus, dissimilarity, as reflected in elevation, may only show an effect on marital satisfaction in the early state of marriage.
The present study has some limitations. We examined broad personality constructs assessed with a very short scale. A previous study pointed out that shortened inventories are associated with often substantially reduced criterion validity (Credé, Harms, Niehorster, & Gaye-Valentine, 2012). Thus, future research should conduct the study of personality traits with a scale that has the narrower facets of personality within the Big Five domain. Moreover, future research should also keep on examining the specific mechanisms that link each facet with marital satisfaction.
In sum, the present study replicated the results of previous studies on the effects of personality on marital satisfaction, even in Eastern culture. These findings support the consistency and robustness of the evidence on personality and marital satisfaction. In addition, the moderating effect of marital duration was found. Further research will consider the differences of importance of (dis)similarity by marital duration.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by JSPS KAKENHI Grant 24730610.
