Sunteți pe pagina 1din 9

PERSONALITY PROCESSES AND INDIVIDUAL DIFFERENCES

Two Personalities, One Relationship: Both Partners' Personality Traits


Shape the Quality of Their Relationship

Richard W. Robins Avshalom Caspi and Terrie E. Moffitt


University of California at Davis Institute of Psychiatry, London,
and University of Wisconsin—Madison

This research tested 6 models of the independent and interactive effects of stable personality traits on
each partner's reports of relationship satisfaction and quality. Both members of 360 couples (N = 720)
completed the Multidimensional Personality Questionnaire and were interviewed about their relationship.
Findings show that a woman's relationship happiness is predicted by her partner's low Negative
Emotionality, high Positive Emotionality, and high Constraint, whereas a man's relationship happiness
is predicted only by his partner's low Negative Emotionality. Findings also show evidence of additive but
not interactive effects: Each partner's personality contributed independently to relationship outcomes but
not in a synergistic way. These results are discussed in relation to models that seek to integrate research
on individual differences in personality traits with research on interpersonal processes in intimate
relationships.

Researchers have been interested in understanding the influence relied exclusively on self-report data for both the personality and
of personality on relationships for several decades. A number of relationship outcome measures. Moreover, even when both couple
early studies compared happily married and unhappily married members were studied, the procedures for analyzing couple data
couples and found relatively strong personality correlates (e.g., had not yet been developed. Finally, many of the personality scales
Barry, 1970; Bentler & Newcomb, 1978; Burgess & Wallin, 1953; used in these studies had poor construct validity, and there was
Dean, 1966; Eysenck & Wakefield, 1981; Kelly & Conley, 1987; little attempt to integrate the findings within a broader structural
Terman & Buttenweiser, 1935). About two decades ago, relation- model of personality. This led to a confusing set of findings and a
ship researchers began to abandon the approach adopted in these corresponding belief that research on personality and relationships
early personality studies. One cause of the demise of interest was was not fruitful.
the relatively weak methodologies used in early studies. For the Perhaps more important than these methodological concerns
most part, researchers studied only one member of the couple and was a theoretical shift away from stable individual differences.
Researchers from a number of areas rejected the view that rela-
tionships could be understood by studying the character of indi-
Richard W. Robins, Department of Psychology, University of California vidual partners rather than the relationship itself. Over the past
at Davis; Avshalom Caspi and Terrie E. Moffitt, Social, Genetic, and couple of decades, relationship research has generally gone in
Developmental Psychiatry Research Centre, Institute of Psychiatry, Lon- three directions (Berscheid & Reis, 1998): clinical research on
don, and Department of Psychology, University of Wisconsin—Madison.
microinteractional processes, social psychological research on in-
This study was supported by National Institute of Mental Health Grants
MH-45070, MH-49414, and MH-56344, by grants from the Graduate
terpersonal dynamics in "close" relationships, and sociological
School of the University of Wisconsin and the U. K. Medical Research research on marital happiness and life course dynamics. All three
Council, and by a Faculty Research Grant from the University of California of these research traditions moved the field away from the study of
at Davis. The Dunedin Multidisciplinary Health and Development Re- individual differences, and personality traits do not play a prom-
search Unit is supported by the New Zealand Health Research Council. inent role in contemporary research and theory on relationships.
This article benefited greatly from comments by R. Chris Fraley and The most notable exception is research on attachment style, which
Samuel D. Gosling. We are grateful to the Dunedin Unit investigators and has surged in popularity over the past decade (Cassidy & Shaver,
staff and to the study members and their partners.
1999).
Correspondence concerning this article should be addressed to Richard
The movement away from the study of personality traits oc-
W. Robins, Department of Psychology, University of California, Davis,
California 95616-8686, or to Terrie E. Moffitt, Social, Genetic and De- curred at a time when the person-situation debate was raging.
velopmental Psychiatry Research Centre, Institute of Psychiatry, 111 Den- Although personality research has seen a resurgence over the past
mark Hill, London SE5 8AF, England. Electronic mail may be sent to decade (Pervin & John, 1999), the relationship field has beeixslow
rwrobins @ ucdavis .edu. to return to the study of stable traits. Yet, there are several reasons
Journal of Personality and Social Psychology, 2000, Vol. 79, No. 2, 251-259
Copyright 2000 by the American Psychological Association, Inc. O022-3514/00/$5.00 DOI: 10.1037//O022-3514.79.2.251

251
252 ROBINS, CASPI, AND MOFFTTT

to believe that personality variables can play a prominent role in the woman's personality. These gender differences can take three
relationship research. First, despite its flaws, the early research forms: differences in the actor effect (Paths a and d differ from
literature did provide some evidence that stable personality traits each other), differences in the partner effect (Paths b and c differ
play a crucial role in determining the nature of a relationship from each other), and differences in both the actor and the partner
(Buss, 1991; Eysenck & Wakefield, 1981). Second, we have a effects (Paths a and b differ from Paths c and d).
much better understanding of personality traits now than we did 20 The fourth model {additive effects) predicts that each couple
years ago. We know more about their heritability, developmental member contributes uniquely to the quality of the relationship.
origins, neurobiological substrates, and how they relate to a host of This model can be tested by examining the effects of the man's
other social and psychological variables across the lifespan. We personality on relationship outcomes while controlling for the
also now have better personality measures, with stronger psycho- effects of the woman's personality, and vice versa. Paths a and c
metric properties and more elaborated nomological networks. would indicate additive effects on the man's outcomes, and Paths
Third, recent behavioral genetic research provides another reason b and d would indicate additive effects on the woman's outcomes.
to believe that stable personality traits matter in relationships. Thus, additive effects require the presence of both an actor and a
Evidence from twin studies shows that heritable personality traits partner effect, although the presence of an actor and a partner
can precipitate relationship problems and contribute to marital effect do not necessarily indicate additive effects.
dissolution (Jockin, McGue, & Lykken, 1996). Finally, previous The fifth model (synergy) predicts a synergistic interaction
attempts to understand the role of personality in relationships were between the man's and the woman's personality. In this case, not
not fruitful, in part because they did not take into account how the only do the personality traits of both partners explain unique
personalities of both couple members jointly influence the variance in relationship outcomes, as predicted by the additive
relationship. model, but they do so in a multiplicative (or nonadditive) way. A
In this study, we used the personality traits of both partners to statistical interaction between Paths a and c would suggest syner-
predict each partner's perception of the relationship. Both couple gistic effects on the man's outcomes, and a statistical interaction
members (720 individuals) completed a comprehensive personal- between Paths b and d would suggest synergistic effects on the
ity questionnaire and were interviewed about the positive and woman's outcomes.
negative aspects of their relationship. Using these data, we tested The sixth model (similarity) predicts that greater similarity
six models of the independent and interactive effects of personality between partners will be associated with higher levels of relation-
on reports of relationship satisfaction and quality. These six mod- ship satisfaction and quality (e.g., Eysenck & Wakefield, 1981). In
els are variants of the general model of the effects of personality on this case, it is not the particular traits characterizing each partner
relationship outcomes shown in Figure 1 (Caspi, Herbener, & that influence relationship outcomes, but rather it is how much the
Ozer, 1992; Gonzalez & Griffin, 1997,1999; Kenny, 1995, 1998). couple members resemble each other on those traits that matters
The six models are not mutually exclusive, although in some cases most.
support for one requires or precludes support for another.
The first model {actor effects) predicts that the actor's person-
ality will influence his or her own perception of the relationship
Method
(Paths a and d in Figure 1). The second model {partner effects) Research Participants
predicts that the actor's own personality will influence how his or
her partner perceives the relationship (Paths b and c in Figure 1). We studied 360 couples recruited through the Dunedin Study, a longi-
The presence of both actor and partner effects would suggest that tudinal investigation of health and behavior (for a detailed description of
personality has an effect on the actual quality of the relationship the study, see Silva & Stanton, 1996). The 1,020 living original study
and not just on the individual's perception of the relationship members were asked prior to their follow-up interview at age 21 (held
(which may be biased by idealization, self-deception, mood- during 1993-1994) if they had a partner to whom they were married or
congruent cognition, and other psychological processes). with whom they were engaged, dating for 6 months or more, or cohabiting.
A total of 474 study members indicated that they were involved with a
The third model (gender specificity) predicts that the man's
partner who met one of those criteria, and 76% (360) of the partners agreed
personality will have different effects on the relationship than will
to participate in the interviews. Of the 360 couples interviewed, 52% were
in dating relationships, 41% were in cohabiting relationships, and 7% were
married. The average relationship length was 26 months, with 73% of the
relationships having lasted for more than a year and 45% having lasted for
more than 2 years. All original study members were 21 years old; the men
from the original sample tended to have slightly younger women as
partners (Mdn = 20; range = 16 to 29 years), whereas the women from the
original sample tended to have slightly older men as partners (Mdn = 22;
range = 17 to 39 years). Thus, the female members of our couple sample
Woman's were significantly younger (M = 20.7, SD = 1.3) than the male couple
Relationship members (M = 22.2, SD = 2.8), t(359) = 9.91, p < .01.
Outcome Study members are predominantly of European ancestry. Three of the
Dunedin couples were same-gender couples, which is consistent with
Figure I. A general model of the effects of personality on relationship surveys of sexual-orientation prevalence in the general population (e.g.,
outcomes. Paths a and d represent actor effects; Paths b and c represent Laumann, Gagnon, Micahel, & Michaels, 1994). These couples were
partner effects; Path e represents personality similarity between the part- included in all analyses except those that required both members of the
ners. couple (e.g., tests of the 6 models).
PERSONALITY CORRELATES OF RELATIONSHIP SATISFACTION 253
The 360 study members whose partners participated did not differ theory (e.g., Tellegen & Waller, in press). The two relationship
significantly from the 114 study members whose partners did not partici- outcomes were strongly correlated for men (r = .71) and for
pate in mean educational attainment, r(446) = .32, p = .75, in the average women (r — .68), suggesting that relationships characterized by
ages of their partners, f(468) = 1.32, p = .19, or on any of the measures positive interpersonal interactions (e.g., intimacy and trust) also
of personality and relationship outcomes examined in the present study (all
tend to be characterized by a high level of satisfaction with specific
/>s > .10). However, study members with participating partners were in
domains of the relationship (e.g., finances and sex).
relationships of longer duration, f(450) = 3.04, p < .01, than were study
members whose partners did not participate. There were significant gender differences for Constraint and
Positive Emotionality. Men were higher in Positive Emotionality,
matched pairs r(338) = 4.8, Cohen's d = .36, p < .01. Women
Measurement of Personality
were higher in Constraint, matched pairs f(338) = 8.6, d = .62,
We used the Multidimensional Personality Questionnaire (MPQ; Telle- p < .01. In terms of relationship outcomes, men were more
gen, 1982) to assess the personality traits of each relationship partner. The satisfied in their relationships than women, matched pairs
MPQ assesses a broad range of individual differences in affective and f(349) = 3.9, d = .24, p < .01, but this difference did not hold for
behavioral style and includes 10 primary scales. (The Absorption scale was relationship quality {d = .00).
not included in this version of the MPQ.) Alpha reliabilities in the present To examine similarity between couple members, we correlated
sample ranged from .63 to .80 (Mdn = .75). The 10 primary scales define 3
the man's personality (or relationship outcome) score with the
higher order superfactors: Negative Emotionality, Positive Emotionality,
woman's score on the same variable. The weak correlations for
and Constraint (Tellegen & Waller, in press). The Negative Emotionality
factor consists of the Stress Reaction, Alienation, and Aggression scales; personality (ranging from .05 to .10) indicate that there was
individuals scoring high on Negative Emotionality have a low threshold for relatively little assortative mating in the present sample, which is
the experience of negative emotions such as fear, anxiety, and anger. consistent with previous research (for studies of this topic using
Positive Emotionality consists of the Well-Being, Social Potency, Social the MPQ, see Lykken & Tellegen, 1993; Waller & Shaver, 1994;
Closeness, and Achievement scales; individuals scoring high on Positive see also Krueger, Moffitt, Caspi, Bleske, & Silva, 1998). In con-
Emotionality have a low threshold for the experience of positive emotions trast, men and women did report similar levels of relationship
and tend to view life as an essentially pleasurable experience. The Con- quality (r = .50) and satisfaction (r = .33).2
straint factor consists of the Control, Harm Avoidance, and Traditionalism
scales; individuals scoring high on Constraint tend to act in a cautious and
restrained manner, avoid thrills, and endorse and conform to social norms. Which Personality Traits Contribute to a
The MPQ was completed by the original members of the Dunedin study, Happy Relationship?
when they were 18 years old, and by their relationship partners 3 years
later.1 We used multiple regression analyses to test whether the
personality traits of both partners independently shape their
perceptions of the relationship. In each analysis we simulta-
Measurement of Relationship Quality and Satisfaction neously entered two independent variables—the man's and the
The two couple members were interviewed privately and separately, woman's MPQ score; a separate analysis was conducted for
with confidentiality guaranteed. Couple members were interviewed simul- each of the 3 superfactors and for each of the 10 primary scales.
taneously by different interviewers who were blind to the responses pro- The dependent variables were the male and female reports of
vided by the other member of the couple. Couple members did not know relationship quality and satisfaction; thus, the analyses were
in advance the content of the interview schedule, which eliminated the conducted across 360 couples. The results of these analyses
possibility that couples would confer about their responses prior to the provide tests of actor (Model 1), partner (Model 2), gender-
interview. specific (Model 3), and additive (Model 4) effects.3 We present
Relationship quality was assessed using 40 interview questions that
asked about shared activities and interests, the balance of power, respect
and fairness, emotional intimacy and trust, and open communication. 1
We conducted a series of moderated multiple regression analyses to
Answers were coded as 0 (almost never), 1 (sometimes), or 2 (almost determine whether time of testing moderated any of the findings. Specif-
always) and were summed across items to generate a score for each couple ically, we examined the interaction between personality and a dummy
member (a = .90). Relationship satisfaction was assessed using 14 inter- variable coded for age 18 versus 21 assessment of the MPQ. We did not
view questions that asked how satisfied the respondent was with different find any significant interactions for any of the findings to be reported. The
domains in the relationship (e.g., finances, sex, division of labor, future high stability of personality from age 18 to 21 probably accounts for the
plans, shared time). Answers were coded as 0 (not happy), 1 (somewhat fact that the concurrent effects are not significantly stronger than the
happy), or 2 (very happy) and summed across items to generate a score for longitudinal effects.
each couple member (a = .80). 2
Another way to calculate the similarity between partners is using a
partial pairwise intraclass correlation (Gonzalez & Griffin, 1999, Equation
Results 2). The partial intraclass correlations were similar in magnitude to the
Pearson correlations: .11 (Negative Emotionality), .05 (Positive Emotion-
Intercorrelations, Gender Differences, and Similarity ality), .11 (Constraint), .45 (relationship quality), .32 (relationship satis-
Between Couple Members faction).
3
The gender-specificity model cannot be completely tested using mul-
The three MPQ superfactor scales were weakly intercorrelated tiple regression. Although the multiple regression analyses show differ-
for both men and women (rs ranging from —.26 to .10). This ences in the magnitude of the male and female effects, the statistical
pattern suggests that Negative Emotionality, Positive Emotional- significance of these differences must be tested using structural equation
ity, and Constraint represent three relatively independent dimen- modeling. As recommended by Kenny (1996, 1998) and Gonzales and
sions of personality, which is consistent with previous research and Griffin (1997), we constrained the male and female paths to be equal and
254 ROBINS, CASPI, AND MOFFITT

Man's Report of
Quality/Satisfaction

Woman's Woman's Report of


Negative Quality/Satisfaction
Emotionality

B Man's Man's Report of


16**/.16**
Positive Quality/Satisfaction
Emotionality

Woman's Woman's Report of


Positive Quality/Satisfaction
Emotionality

Man's Man's Report of


.28**7.16**
Constraint Quality/Satisfaction

Woman's Woman's Report of


Constraint Quality/Satisfaction

Figure 2. Panel A: The effects of Negative Emotionality on relationship quality and satisfaction. Panel B: The
effects of Positive Emotionality on relationship quality and satisfaction. Panel C: The effects of Constraint on
relationship quality and satisfaction. Values are standardized beta weights (except for the assortative mating
correlation) from four separate regression analyses, predicting the man's reports of quality and satisfaction and
the woman's reports of quality and satisfaction. * p < .05. ** p < .01.

the tests for synergistic (Model 5) and similarity (Model 6) 1 summarizes the results of the multiple regression analyses for
effects in subsequent sections. The results of the multiple the 10 MPQ primary scales.
regression analyses are shown in Figure 2 for Negative Emo- Negative Emotionality. Negative Emotionality had strong and
tionality, Positive Emotionality, and Constraint. To help inter- consistent effects on the two relationship outcomes. In terms of
pret the effects observed at the level of the superfactors, Table actor effects, both men and women high in Negative Emotionality
reported lower levels of satisfaction and quality. These effects held
for all three primary scales related to Negative Emotionality:
then evaluated the significance of the equality constraints by chi-square Aggressive, alienated, and stress-reactive men and women tended
tests of the difference in the fit of die constrained and unconstrained to be unhappy in their relationships. In terms of partner effects,
models. If the unconstrained model fit significantly better, then there was both men and women were unhappy when their partners were high
a significant gender difference. To test the gender-specificity model, we in Negative Emotionality (even after controlling for their own
conducted three analyses. First, we constrained Paths a and d (see Figure
level of Negative Emotionality). Analyses of the primary scales
1) to be equal, which tests for gender differences in the actor's effect on the
serf. Second, we constrained Paths b and c to be equal, which tests for
showed that the crucial dimension for women was whether their
gender differences in the actor's effect on the partner. Third, to test for partner was aggressive, whereas the crucial dimension for men was
gender dominance, we constrained Paths a and b to equal paths c and d. whether their partner was alienated and reactive to stress. Overall,
These analyses were conducted using the EQS program. the results point to additive effects—that is, the man's and the
PERSONALITY CORRELATES OF RELATIONSHIP SATISFACTION 255
Table 1
Multiple Regression Analyses Predicting Relationship Outcomes From the Multidimensional
Personality Questionnaire (MPQ) Primary Scales

Male outcome Female outcome

MPQ scale Quality Satisfaction Quality Satisfaction

Negative Emotionality
Aggression
Male -.28* -.20* -.21* -.18*
Female -.06 -.04 -.23* -.25*
Alienation
Male -.19* -.26* -.07 -.06
Female -.14* -.12* -.27* -.26*
Stress reaction
Male -.14* -.26* .00 -.07
Female -.22* -.20* -.29* -.34*
Positive Emotionality
Social closeness
Male .21* .17* .14* .16*
Female .04 .01 .25* .18*
Social potency
Male .02 -.03 .07 .07
Female -.07 -.07 -.02 -.01
Well-being
Male .09 .18* .01 .11*
Female .04 .07 .16* .22*
Achievement
Male .12* .08 .08 .00
Female -.03 -.07 .01 .04
Constraint
Control
Male .22* .08 .18* .16*
Female .11* .02 .14* .13*
Harm avoidance
Male .17* .07 -.03 -.06
Female -.02 .00 .02 -.01
Traditionalism
Male .24* .22* .14* .11*
Female -.05 -.04 -.05 .01

Note. Ns ranged from 345 to 348 for men and from 347 to 351 for women. Values in the table are standardized
beta weights from multiple regression analyses in which the man's and the woman's MPQ scores were used to
predict male and female reported relationship outcomes.
* p < .05.

woman's Negative Emotionality independently influenced each We found partner effects for the man's Positive Emotionality
partner's perception of the relationship. but not the woman's. Specifically, men high in Positive Emotion-
In terms of gender differences, we found only one significant ality tended to have satisfied partners.4 This effect was due to
difference: the effect of the woman's personality on her own report social closeness; men with the capacity for intimacy and the ability
of relationship quality was stronger than the effect of the man's to provide comfort and express warmth tended to have more
personality on his own report of relationship quality, A ^ ( l ) = 5.1,
p < .05. This difference was not significant for satisfaction,
Ax^l) = 2.0, p > .10. Overall, the findings support the actor, 4
We tested whether the effect of the man's Positive Emotionality was
partner, and additivity models, and provide marginal support for independent of his Negative Emotionality. Entering the man's Negative
the gender-specificity model. Emotionality into the regression equations produced little change in the
Positive Emotionality. The effects of Positive Emotionality beta weights for Positive Emotionality, and the effect of Positive Emotion-
were weaker than those of Negative Emotionality but showed a ality remained significant for all four dependent variables (man's and
woman's satisfaction and quality ratings). Moreover, entering the man's
consistent pattern across reports of quality and satisfaction (see
Constraint, in addition to his Negative Emotionality, also did not change
Figure 2, Panel B). In terms of actor effects, both men and women the beta weights substantially, and all effects of Positive Emotionality
high in Positive Emotionality were happier in their relation- remained significant except for woman's report of relationship quality
ships. The primary scale analyses showed that these effects were (adding Constraint and Negative Emotionality reduced the beta from. 12 to
due to the social closeness and well-being aspects of Positive .10, p = .07). Thus, the effects of the man's Positive Emotionality are
Emotionality. largely independent of his Negative Emotionality and Constraint.
256 ROBINS, CASPI, AND MOFFTTT

satisfied partners. In contrast, men were not necessarily happier in man's Constraint. An aggressive, stress-reactive woman and an
relationships in which their partners tended to experience and impulsive man seem like a recipe for an unhappy relationship. To
express positive emotions. The gender difference in the partner explore such possibilities, we examined two between-partner in-
effects was statistically significant for satisfaction, A ^ l ) = 4.6, teraction terms (man with woman, woman with man) for each of
p < .05, and marginally significant for quality, A^Cl) = 3.6, the three possible trait interactions (Negative Emotionality with
p = .07. Constraint, Negative Emotionality with Positive Emotionality, and
Overall, then, the findings show that both the man's and the Positive Emotionality with Constraint), producing six interaction
woman's Positive Emotionality contributed independently (i.e., terms. We repeated this set of analyses for male and female
had additive effects) to the happiness of the woman's relationship; outcomes. Out of 24 tests for an interaction, all of the effects were
however, the man's relationship happiness was predicted only by weak in magnitude and none were significant (all ps > .05).5
his own level of Positive Emotionality, in particular his level of Overall, the findings do not support the synergy model.
social closeness. Thus, the findings provide at least partial support
for the actor, partner, gender-specificity, and additivity models. Are Similar Couples Happy Couples?
Constraint. The most noteworthy finding for Constraint was
the effect of the man's Constraint on the relationship (see Figure 2, Is the degree of personality similarity between partners associ-
Panel C). Men high in Constraint were happier in their relation- ated with relationship satisfaction and quality? We calculated a d2
ships, whereas women high in Constraint were not happier, index of personality similarity (which is the sum of the squared
Ay 2 (l) = 7.9, p < .01. The primary scale analyses showed that, in deviations between the two profiles) across all 10 primary scales,
general, men high in self-control, harm avoidance, and tradition- across the 3 superfactors, and across the primary scales defining
alism reported high levels of relationship quality and satisfaction. each superfactor, and then correlated these 5 profile similarity
In terms of partner effects, there was a weak tendency for the indexes with male and female ratings of relationship satisfaction
partners of high Constraint men to have a more positive view of and quality.
the relationship; this effect was significant for quality (/3 = .12, The findings show that men tend to be happier in relationships
p < .05) but did not attain significance for satisfaction (/3 = .08, when they are similar to their partners (see Table 2). This effect is
p = .13). Analyses at the primary scale level showed that two due primarily to similarity on the Negative Emotionality scales,
dimensions—self-control and traditionalism—accounted for this although the correlations for the other two superfactors are in the
partner effect: Women were happier when their partners were same direction. For women, the correlations were weaker and less
traditional and had appropriate impulse control. The overall pattern consistent, but the one significant effect also pointed toward a
of findings is consistent with a male dominance model, in which positive relation between similarity and relationship satisfaction.
the man's Constraint drives the relationship experience of both
members of the couple. A test of this form of gender specificity Discussion
showed that the two male effects were significantly stronger than
the two female effects, A^2(2) = 8.8, p < .05. Thus, the findings The present study examined 360 couples in established relation-
for Constraint support the actor, partner, and gender-specificity ships to test how the personality traits of both members of a couple
models; the findings do not support the additivity model. combine to shape relationship outcomes. Each couple member was
assessed on three broad personality traits (Negative Emotionality,
Positive Emotionality, Constraint) and on 10 specific trait dimen-
Searching for Synergy: Interactive Effects of Personality sions. In addition, couple members were separately interviewed
on Relationship Outcomes about their satisfaction with the relationship and its overall quality.
In the previous section, we found that the personality traits of
both partners make unique contributions to relationship outcomes. Which Personality Traits Contribute to a
However, these additive effects do not demonstrate that the couple Happy Relationship?
functions as a synergistic system in which the whole (the relation-
Low Negative Emotionality was the strongest and most consis-
ship) is more than the sum of the parts (the two individuals'
tent predictor of quality and satisfaction. Both men and women are
personalities). Demonstrating synergy (Model 5) requires evidence
happy in a relationship when neither partner tends to experience
of a multiplicative interaction between the personality character-
and express negative emotions. These effects are consistent with a
istics of the relationship partners.
recent review of longitudinal predictors of marital quality and
We first examined interactions involving a single trait. For
stability, which concluded that Negative Emotionality, or Neurot-
example, are couples particularly unhappy when both partners are
icism, shows stronger effects on marital outcomes than other
high in Negative Emotionality? We used moderated multiple re-
personality factors (Karney & Bradbury, 1995; see also Buss,
gression to predict each relationship outcome from the man's MPQ
1991).
score, the woman's MPQ score, and the product of the standard-
The centrality of negative emotions fits with a core finding of
ized MPQ scores (representing the interaction between the man's
contemporary research on relationship processes: The way in
and the woman's personality). We conducted these analyses for all
which negative affect is expressed toward the partner is the crucial
three MPQ superfactors, separately for male and female outcomes
on each relationship variable. None of the 12 resulting analyses
produced a significant interaction effect (all ps > .05). 5
We also tested for within-person interactions among the MPQ super-
We also tested for interactions involving multiple traits, for factors (e.g., man's Negative Emotionality and man's Constraint) but found
example, between the woman's Negative Emotionality and the no significant effects for either men or women.
PERSONALITY CORRELATES OF RELATIONSHIP SATISFACTION 257

Table 2
Effects of Partner Similarity on Male and Female Reported Relationship Outcomes

Male outcome Female outcome


a
Similarity index Quality Satisfaction Quality Satisfaction

10 primary scales .15* .12* .01 .08


3 superfactors .12* .10 .07 .00
Negative emotionality .18* .13* .00 .11*
Positive emotionality .06 .04 .06 .09
Constraint .04 .06 .06 -.05

Note. N = 336.
a
This index represents the sum of squared discrepancies between the man's and the woman's personality profile.
The index was multiplied by — 1 so that it indicates partner similarity.
*p<.05.

determinant of relationship quality. High Negative Emotionality Evaluating Six Models of the Relation Between
individuals are prone to express four behaviors that Gottman Personality and Relationship Outcomes
(1994) identified as detrimental to relationships: criticism, con-
tempt, defensiveness, and stonewalling. High Negative Emotion- We found robust support for the actor, partner, and additive
ality individuals may also be more physiologically reactive and models, some support for the gender-specific and similarity mod-
more likely to escalate negative affect during conflict with their els, and no support for the synergy model. The actor model
partners, thereby lowering both couple members' satisfaction (e.g., predicts that people's personality traits shape their view of the
Gottman, Coan, Carrere, & Swanson, 1998). relationship (Paths a and d in Figure 1), and the findings suggest
Positive Emotionality and Constraint also influenced relation- that, although romantic partners are entwined in a relationship with
ship satisfaction and quality. Both men and women are happy in another person, it is their own personality that drives much of their
relationships in which the man experiences and expresses positive subjective experience of that relationship. However, the presence
emotions and has the capacity to constrain his impulses. Taken of partner effects suggests that the link between self-reports of
together, our data suggest that a woman's relationship happiness is personality and subjective evaluations of the relationship do not
simply reflect a rose-colored or dark-tinted glasses effect. There
predicted by her partner's level on all three traits—Negative Emo-
were a number of ways in which the man's personality influenced
tionality, Positive Emotionality, and Constraint—whereas a man's
his partner's level of happiness (Path b) and, likewise, a few ways
relationship happiness is predicted by his partner's Negative Emo-
in which the woman's personality influenced her partner's happi-
tionality alone.
ness (Path c).
Why do men seem to be influenced only by their partner's
In several cases, we found that both the man's and the woman's
tendency to be critical and anxious, whereas women are influenced
personality characteristics made independent contributions to the
by the total constellation of personality traits displayed by their
nature of the relationship. These additive effects can be seen as a
partners? Again research on relationship processes may be infor-
special case of a person-environment transaction, in which one's
mative. Women are more likely to raise problems and criticisms in
relationship partner constitutes the environment. There are three
relationships; therefore, the ability of the man to soothe his partner
transactional processes by which personality may affect relation-
and "embrace her anger" is crucial (Gottman, 1994). Men high in
ship outcomes (Caspi, 1998). Consider the case of Negative Emo-
Positive Emotionality may be better able to deescalate their part- tionality—why might a person high in Negative Emotionality be
ner's negative affect through the use of humor and other behaviors more likely to have an unhappy relationship? First, they may select
that defuse conflict. Men high in Positive Emotionality also ex- partners who are similar to themselves. However, our data showed
press a stronger need for close relationships and may be motivated little or no assortative mating on personality, and our multiple
to work to preserve harmony when their girlfriend or wife ex- regression analyses controlled for any assortative mating that was
presses dissatisfaction (Graziano, Jensen-Campbell, & Hair, present. Thus, it is not simply that high Negative Emotionality
1996). Similarly, men high in Constraint may possess the self- individuals are unhappy because they pair off with similarly ag-
control necessary to avoid impulsively disengaging from problem- gressive, alienated, or anxious others. Second, people high in
atic interactions. Evolutionary psychology provides another pos- Negative Emotionality may react to and interpret the behaviors of
sible interpretation of our findings. If women use a man's their partners in a negative light, and thus be primed to perceive
personality traits as cues of his ability to be a stable and committed and magnify relationship difficulties (Rusting, 1998). For example,
relationship partner who will provide resources for their children they may overreact to minor criticism from their partner, believe
(Buss, 1994), then it makes sense that women are more satisfied they are no longer loved when their partner does not call, or
with partners who have the capacity for warm and intimate rela- assume infidelity on the basis of mere flirtation. Third, people high
tionships (low Negative Emotionality, high social closeness) and in Negative Emotionality may evoke behaviors from their partner
who are good at planning and are persistent and thus likely to be that contribute to their own dissatisfaction. For example, an ag-
"good providers" (high self-control). gressive or alienated man may get the "silent treatment" from his
258 ROBINS, CASPI, AND MOFFITT

partner, and consequently become less satisfied with his relation- ality trait on a relationship represents the summed effect of the
ship. The presence of additive effects underscores the need for actor and partner effects (e.g., the multiple R predicting female
relationship researchers to study both members of a relationship. satisfaction from the man's and the woman's Negative Emotion-
We found no support for the synergy model. Relationship re- ality was .40). Moreover, relationship outcomes are determined by
searchers often emphasize the need to understand the couple as an a complex set of factors, and we would not expect personality traits
interactive system rather than in terms of each partner's individual alone to exhaustively account for relationship satisfaction. Finally,
characteristics (e.g., Gottman, 1982). However, our analyses failed we believe it is of vital significance to explain even a small
to show any interactive effects between male and female person- percentage of the variance in an almost universally important life
ality traits, despite testing for interactions in multiple ways and goal—attaining a happy relationship (Roberts & Robins, in press).
with adequate statistical power (Aiken & West, 1991). Thus, each
partner's personality contributes independently-to relationship out-
Finding a Place for Personality in Relationship Research
comes but not in a synergistic way—a whole relationship is not
greater than the sum of its two parts, at least not when the parts are Almost two decades ago, Gottman (1982) argued that we should
each partner's personality traits. This conclusion should be qual- stop studying the dispositional basis of relationship outcomes and
ified by three observations. First, interactions between couple focus on the processes characterizing couple interactions. His call
members' personality traits may be even more complicated than to "free ourselves from this emphasis on individual functioning"
those we tested (e.g., the entire configuration of traits of one (p. 945) was profoundly influential, and the past two decades have
partner may interact with the other partner's configuration of seen an enormous amount of research on microinteractional
traits). Second, we have used the term synergy to refer to a processes.
multiplicative interaction between the traits of two individuals. However, the study of relationship processes is not incompatible
The absence of such synergy does not rule out the existence of with the study of individual differences, and we believe there is
some form of transactional synergy emerging out of microinterac- room for better integration of the two approaches. Individuals
tional relationship dynamics (e.g., escalation of negative affect). create microinteractional processes; these processes do not occur
Finally, the absence of synergistic effects does not imply that only in a vacuum or arise de novo. People bring histories to relation-
one member of the couple matters. On the contrary, our findings ships, and these histories are captured in part by stable personality
highlight the importance of additive effects in the couple system. traits. We have proposed several possible links between stable
These additive effects constitute a unique form of person- traits and microinteractional processes. These mediating links need
environment interaction, called additive coaction (Rutter, 1983), in to be examined more closely in future research. A useful research
which both members of a couple jointly and uniquely shape the design would combine the measurement of personality traits with
relationship. the measurement of relationship-specific behaviors and emotions
and test them in a single model. We predict that the personality
effects reported in the present study will be mediated by ongoing
Limitations relationship-specific behaviors and emotions.
The present study has several limitations. First, it is difficult to The study of individual differences is also not incompatible with
determine whether personality causes relationship outcomes or the idea that relationship outcomes are a property of the couple. It
relationship experiences cause personality change. The antecedent may be true that a relationship constitutes a dynamic system and
nature of some of the data may be informative. When the original not simply two distinct individuals, but an important way to
sample members completed the MPQ (3 years before they com- understand the system is to better understand its constituent parts.
pleted the relationship measures ) only 15% were already involved Every relationship is made up of two personalities, and both
in their current relationship. Thus, for the vast majority, it was not partners' personality traits jointly and uniquely shape the quality of
possible that their personality was altered by their happiness in the their relationship.
current relationship. Moreover, a recent longitudinal study failed
to find any evidence that relationships change personality (at least
not over an 18 month period), although personality did predict 6
We tested whether relationship length or status (living vs. not living
subsequent relationship outcomes (Asendorpf & Wilpers, 1998). together) moderated any of the findings reported in Figure 2. None of the
Second, our sample was comprised almost entirely of young interaction effects was significant for relationship length and only one
adults. It is possible that the present findings will not generalize to interaction was significant for relationship status (p < .05). This interac-
older adults or to long-term marital relationships.6 Other research, tion reflected the fact that the man's Positive Emotionality was associated
however, shows that the MPQ traits are highly stable in adulthood with the woman's satisfaction when they were living together (/3 = .25) but
(McGue, Bacon, & Lykken, 1993) and are thus likely to influence not when they were living apart (j3 = .03).
romantic relationships in later years. Nonetheless, the personality
correlates of relationship outcomes may change developmentally; References
for example, older individuals, whose relationships center more on
companionship and childrearing, may derive satisfaction from Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and
interpreting interactions. Newbury Park, CA: Sage.
partners with different personality traits than those from whom
Asendorpf, J. B., & Wilpers, S. (1998). Personality effects on social
they derived satisfaction when they were younger.
relationships. Journal of Personality and Social Psychology, 74,
Third, the findings may appear to some readers to be too weak 1531-1544.
to be clinically significant. One consideration is that the effects Barry, W. A. (1970). Marriage research and conflict: An integrative re-
were generally additive across partners, so the impact of a person- view. Psychological Bulletin, 73, 185-193.
PERSONALITY CORRELATES OF RELATIONSHIP SATISFACTION 259
Bentler, P. M., & Newcomb, M. D. (1978). Longitudinal study of marital prospective analysis of marital stability and marital satisfaction. Journal
success and failure. Journal of Consulting and Clinical Psychology, 46, of Personality and Social Psychology, 52, 27-40.
1053-1070. Kenny, D. A. (1995). The effect of nonindependence on significance
Berscheid, E., & Reis, H. T. (1998). Attraction and close relationships. In testing in dyadic research. Personal Relationships, 2, 67-75.
D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The handbook of social Kenny, D. A. (1996). Models of non-independence in dyadic research.
psychology, Vol. 2 (4th ed., pp. 193-281). Boston, MA: McGraw-Hill. Journal of Social and Personal Relationships, 13, 279-294.
Burgess, E. W., & Wallin, P. (1953). Engagement and marriage, Phila- Kenny, D. A. (1998). Couples, gender, and time: Comments on method. In
delphia: Lippincott. T. N. Bradbury (Ed.), The developmental course of marital dysfunction
Buss, D. M. (1991). Conflict in married couples: Personality predictors of (pp. 410-422). New York: Cambridge University Press.
anger and upset. Journal of Personality, 59, 663-688. Krueger, R. F., Moffitt, T. E., Caspi, A., Bleske, A., & Silva, P. A. (1998).
Buss, D. M. (1994). The evolution of desire: Strategies of human mating. Assortative mating for antisocial behavior: Developmental and method-
New York: Basic Books. ological implications. Behavior Genetics, 28, 173-186.
Caspi, A. (1998). Personality development across the life course. In W. Laumann, E. O., Gagnon, J. H., Micahel, R. T., & Michaels, S. (1994). The
Damon (Series Ed.) & N. Eisenberg (Vol. Ed.), Handbook of child social organization of sexuality: Sexual practices in the United States.
psychology: Vol. 3. Social, emotional, and personality development (pp. Chicago: University of Chicago Press.
311-388). New York: Wiley. Lykken, D. T., & Tellegen, A. (1993). Is human mating adventitious or the
Caspi, A., Herbener, E. S., & Ozer, D. J. (1992). Shared experiences and result of lawful choice? A twin study of mate selection. Journal of
the similarity of personalities: A longitudinal study of married couples. Personality and Social Psychology, 65, 56-68.
Journal of Personality and Social Psychology, 62, 281-291. McGue, M., Bacon, S., & Lykken, D. T. (1993). Personality stability and
Cassidy, J., & Shaver, P. R. (Eds.). (1999). Handbook of attachment: change in early adulthood: A behavioral genetic analysis. Developmental
Theory, research, and clinical applications. New York: Guilford Press. Psychology, 29, 96-109.
Dean, D. G. (1966). Emotional maturity and marital adjustment. Journal of Pervin, L., & John, O. P. (1999). Handbook of personality: Theory and
Marriage and the Family, 28, 454-457. research. New York: Guilford Press.
Eysenck, H. J., & Wakefield, J. A. (1981). Psychological factors as Roberts, B. W., & Robins, R. W. (in press). Broad dispositions, broad
predictors of marital satisfaction. Advances in Behavior Research and aspirations: The intersection of personality traits and major life goals.
Therapy, 3, 151-192. Personality and Social Psychology Bulletin.
Gonzalez, R., & Griffin, D. (1997). On the statistics of interdependence: Rusting, C. L. (1998). Personality, mood, and cognitive processing of
Treating dyadic data with respect. In S. Duck (Ed.), Handbook of emotional information: Three conceptual frameworks. Psychological
personal relationships: Theory, research, and interventions (2nd ed., pp. Bulletin, 124, 165-196.
1
271-302). Chichester, England: Wiley. Rutter, M. (1983). Statistical and personal interactions: Facets and per-
Gonzalez, R., & Griffin, D. (1999). The correlational analysis of dyad-level spective. In D. Magnusson & V. L. Allen (Eds.), Human development:
data in the distinguishable case. Personal Relationships, 6, 449-469. An interactional perspective (pp. 295-319). New York: Academic Press.
Gottman, J. M. (1982). Temporal form: Toward a new language for Silva, P. A., & Stanton, W. (Eds.). (1996). The Dunedin Study: From child
describing relationships. Journal of Marriage and the Family, 44, to adult. London: Oxford University Press.
943-962. Tellegen, A. (1982). Brief manual for the Multidimensional Personality
Gottman, J. M. (1994). What predicts divorce? The relationship between Questionnaire. Minneapolis: University of Minnesota Press.
marital processes and marital outcomes. Hillsdale, NJ: Erlbaum. Tellegen, A., & Waller, N. (in press). Exploring personality through test
Gottman, J. M., Coan, J., Carrere, S., & Swanson, C. (1998). Predicting construction: Development of the Multidimensional Personality Ques-
marital happiness and stability from newlywed interactions. Journal of tionnaire. In S. R. Briggs & J. M. Cheek (Eds.), Personality measures:
Marriage and the Family, 60, 5-22. Development and evaluation. Greenwich, CT: JAI Press.
Graziano, W. G., Jensen-Campbell, L. A., & Hair, E. C. (1996). Perceived Terman, L. M., & Buttenweiser, P. (1935). Personality factors in marital
interpersonal conflict and reacting to it: The case for agreeableness. compatibility. Journal of Psychology, 6, 143-171.
Journal of Personality and Social Psychology, 70, 820-835. Waller, N. G., & Shaver, P. R. (1994). The importance of nongenetic
Jockin, V., McGue, M., & Lykken, D. T. (1996). Personality and divorce: influences on romantic love styles: A twin-family study. Psychological
A genetic analysis. Journal of Personality and Social Psychology, 71, Science, 5, 268-274.
288-299.
Kamey, B., & Bradbury, T. N. (1995). The longitudinal course of marital
quality and stability: A review of theory, methods, and research. Psy- Received February 10, 1999
chological Bulletin, 118, 3-34. Revision received February 15, 2000
Kelly, E. L., & Conley, J. J. (1987). Personality and compatibility: A Accepted February 18, 2000

S-ar putea să vă placă și