BIG FIVE PERSONALITY AND RELATIONSHIP SATISFACTION:
ACTOR, PARTNER AND SIMILARITY EFFECTS
By
Portia S. Dyrenforth

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Psychology
2010

ABSTRACT
BIG FIVE PERSONALITY AND RELATIONSHIP SATISFACTION:
ACTOR, PARTNER AND SIMILARITY EFFECTS
By
Portia S. Dyrenforth
Data from married couples in the United Kingdom (N = 6,554), Australia (N = 5,278),
and the United Kingdom (N = 11,418) were used to examine the intrapersonal and interpersonal
effects of personality on relationship and life satisfaction. The Actor-Partner Interdependence
Model (APIM; Kenny, Kashy, & Cook, 2006) was used to examine three kinds of personality
effects: actor effects, partner effects, and similarity effects. Across these three large, nationally
representative samples there was clear evidence of actor and partner effects of personality for
both marital and life satisfaction. In addition, despite the sufficient statistical power provided by
these large samples and the use of different indexes to capture similarity in both elevation and
profile shape, there was no consistent evidence that the dyadic variable of similarity matters over
and above the actor and partner effects of personality.

To my mom, who continually teaches me what a smart and compassionate woman can
accomplish.

iii

ACKNOWLEDGEMENTS

This project simply could not have been accomplished without my mentors and
collaborators; Rich Lucas, Debby Kashy, and Brent Donnellan. I will be forever indebted
for your assistance, support, and patience throughout this process.
I am also incredibly grateful for the support of my parents, friends, and HWS
colleagues for standing by me throughout this journey and reminding me that life after
graduate school was something worth fighting for.

iv

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. vi
LIST OF FIGURES ......................................................................................................... viii
Big Five Personality and Relationship Satisfaction: Actor, Partner, and Similarity Effects1
Big Five Personality and Relationship Satisfaction .................................................2
Components of a Relationship: Actor, Partner and Similarity Effects ....................3
Personality and Relationship Satisfaction – Actor Effects.......................................4
Partner’s Big Five Personality and Relationship Satisfaction .................................7
Personality Similarity and Relationship Satisfaction ...............................................9
Statistical Considerations for Analyses of Similarity ............................................14
Overview of Research ............................................................................................22
Method ...................................................................................................................24
Analysis Plan .........................................................................................................27
Results ....................................................................................................................31
Conclusion .............................................................................................................45
APPENDIX ........................................................................................................................55
REFERENCES ..................................................................................................................70

v

LIST OF TABLES

Table 1: Descriptive Statistics for Personality and Relationship Satisfaction Variables 56
Table 2: Univariate analyses predicting relationship satisfaction as a function of
personality, partner personality, and personality similarity (absolute difference between
trait factor scores). ............................................................................................................ 57
Table 3: Simultaneous analyses predicting relationship satisfaction as a function of
personality, partner personality, and absolute differences for all 5 trait factors. ............ 58
Table 4: Discrepancy Analyses By Traits: Predicting relationship satisfaction as a
function of personality, partner personality, and the sum of discrepancies across all 5
trait factors........................................................................................................................ 59
Table 5: Discrepancy Analyses by Item: Predicting relationship satisfaction as a function
of personality, partner personality, and the sum of discrepancies across all personality
items while controlling for trait factors. ........................................................................... 60
Table 6: Profile Correlation Analyses by Traits: Predicting relationship satisfaction as a
function of personality, partner personality, and profile similarity (of trait factor scores).
........................................................................................................................................... 61
Table 7: Profile Correlation Analyses by Items: Predicting relationship satisfaction as a
function of personality, partner personality, and profile similarity (of items) while
controlling for trait factor scores. ..................................................................................... 62
Table 8: Univariate analyses predicting life satisfaction as a function of personality,
partner personality, and personality similarity (absolute difference between trait factor
scores). .............................................................................................................................. 63
Table 9: Simultaneous analyses predicting life satisfaction as a function of personality,
partner personality, and absolute differences for all 5 trait factors. ................................ 64
Table 10: Discrepancy Analyses By Traits: Predicting life satisfaction as a function of
personality, partner personality, and the sum of discrepancies across all 5 trait factors. 65

vi

Table 11: Discrepancy Analyses by Item: Predicting life satisfaction as a function of
personality, partner personality, and the sum of discrepancies across all personality items
while controlling for trait factors...................................................................................... 66
Table 12: Profile Correlation Analyses by Traits: Predicting life satisfaction as a
function of personality, partner personality, and profile similarity (of trait factor scores).
........................................................................................................................................... 67
Table 13: Profile Correlation Analyses by Items: Predicting life satisfaction as a function
of personality, partner personality, and profile similarity (of item scores). ..................... 68

vii

LIST OF FIGURES

Figure 1: Hypothetical Data to Illustrate Three Forms of Similarity............................... 53

viii

Big Five Personality and Relationship Satisfaction: Actor, Partner, and Similarity
Effects
Are individual differences in personality associated with relationship quality? Is
it true that ―birds of a feather flock together‖ and if so, are couples with more similar
personalities more satisfied with their relationships? These questions have received
increasing attention in recent years as researchers have worked to merge evidence
regarding the intrapersonal and interpersonal influences of personality on close
relationships (e.g., Barelds, 2005; Gaunt, 2006; Gonazaga, Campos, & Bradbury, 2007;
Watson, Hubbard, & Wiese, 2000). Amidst growing evidence that an individual’s
personality traits are reliably associated with his or her own satisfaction, there are still
several questions remaining about the interpersonal influences of these traits. The
existing literature contains mixed evidence regarding how relationship satisfaction is
related to the personality traits of a partner or spouse. In addition, there is contradictory
evidence regarding whether personality similarity is important for relationship quality,
with some studies showing that similarity is related to relationship satisfaction and other
studies finding no association.
The current project reviews the evidence addressing these questions and
highlights some of the statistical considerations that may help to explain the diverging
results. In addition, a series of new studies were conducted in order to test the effects of
personality and personality similarity on relationship satisfaction in larger and more
representative samples than have previously been available.

1

Big Five Personality and Relationship Satisfaction
Personality traits are defined as ―consistent patterns of thoughts, feelings, or
actions that distinguish people from one another‖ (Johnson, 1997, p.74). There is
emerging consensus that five broad factors capture most of the variability in normal
personality traits. These ―Big Five‖ trait domains consist of Extraversion (traits like
energetic and sociable), Agreeableness (traits like considerate and kind),
Conscientiousness (traits like hard-working and orderly), Emotional Stability (or its
converse, Neuroticism, that includes traits like nervous and tense), and Openness (traits
like artistic and creative).
Theory and research suggest that the personality traits individuals bring to a
relationship are related to marital satisfaction and functioning. Because personality
shapes the ways in which people interpret and respond to their circumstances, the traits of
each partner can be expected to influence the interactions within a relationship (Robins,
Caspi, & Moffitt, 2002). Personality is believed to be an important, but distal predictor
of relationship quality with personality traits shaping interactions within the relationship
which in turn influence relationship satisfaction (Karney & Bradbury, 1995). Personal
dispositions serve as ―enduring vulnerabilities‖ that shape how couple members react to
one another and to external events. Research evidence supports the idea that interactions
between partners partially mediate the associations between personality and relationship
outcomes (Donnellan, Assad, Robins, & Conger, 2007). Building on this model, it is
possible that the interpersonal effects of personality may also extend beyond the
relationship domain. If personality shapes daily interactions and reactions, the
personality traits that an individual brings to a relationship may not only affect

2

satisfaction specific to the relationship, but also how satisfied his or her partner is with
life in general.

Components of a Relationship: Actor, Partner and Similarity Effects
Relationships can be considered interdependent systems comprised of three parts;
the individual characteristics of each partner, and characteristics resulting from unique
combination of the two individuals. These conceptual distinctions are formalized in a
statistical tool for analyzing data from couples, the actor-partner interdependence model
(APIM; Kenny, Kashy, & Cook, 2006). The APIM is an important means of
investigating the links between personality traits and relationship satisfaction. Actor
effects capture the effects of an individual’s personality trait on her or his own level of
satisfaction with the relationship (e.g., the effects of an individual’s own level of
Agreeableness on her or his judgment of relationship quality), whereas partner effects
capture the effects of the individual’s personality trait on her or his partner’s level of
satisfaction with the relationship (e.g., the effects of an individual’s level of
Agreeableness on her or his partner’s judgment of relationship quality). The existence of
partner effects would strengthen the argument that interpersonal variables influence
relationship functioning. This is particularly convincing because partner effects are
normally free from shared method biases (see Gottman, 1998) because one partner is
reporting on her or his personality whereas the other partner is reporting on her or his
own relationship satisfaction. In short, partner effects provide relatively strong evidence
that personality traits have interpersonal effects on relationships (Kenny & Cook, 1999).

3

In addition to actor and partner effects, relationship satisfaction may be influenced
by characteristics that are unique to the relationship itself. APIM analyses make it
possible to estimate the influence of such dyad-level variables while controlling for the
main effects of each spouse’s initial values. For example, the influence of an actor’s
personality, a partner’s personality, and personality similarity can be estimated using the
Actor-Partner Interdependence model.

Personality and Relationship Satisfaction – Actor Effects
Empirically, a variety of studies have documented that an individual’s stable
personality traits are in fact associated with his or her own relationship satisfaction (e.g.,
Robins, et al., 2000; Watson, et al., 2000). Across the literature, it appears that Emotional
Stability, Agreeableness, and Conscientiousness are the personality traits with the most
robust links to relationship satisfaction. Heller, Watson, and Ilies (2004) recently
published a meta-analysis summarizing the links between the Big Five and marital
satisfaction. They found that the strongest association was between Emotional Stability
and marital satisfaction (r = .29, based on 40 studies [k] involving 7,640 participants [N])
and between Agreeableness and marital satisfaction (r = .29, k = 19, N = 3,071). The
next strongest effect was the association between Conscientiousness and marital
satisfaction (r = .25, k = 6, N = 1,201). The associations between marital satisfaction and
Extraversion (r = .17, k = 22, N = 3,372) and between marital satisfaction and Openness
(r = .10, k = 5, N = 1,154) were generally weaker.
Subsequent research using different measures of personality and satisfaction have
found similar patterns. For example, in a sample of 291 married couples in Iowa, self-

4

rated personality significantly predicted a composite measure of marital and sexual
satisfaction for four of the Big Five traits as measured by the BFI (Watson et al., 2004).
Using hierarchical regression Watson and his colleagues reported the R2 when selfreported personality was entered into the prediction equation for the individual’s own
satisfaction. Similar to the meta-analytic results, the largest effects were for the traits of
Emotional Stability (R2 = .192 for wives, R2 = .103 for husbands), Agreeableness (R2 =
.100 for wives, R2 = .166 for husbands) and Conscientiousness (R2 = .042 for wives, R2 =
.047 for husbands). Self-rated Openness had a small but statistically significant effect on
satisfaction for both husbands and wives (R2 = .078 for wives, R2 = .033 for husbands),
but self-reported Extraversion significantly predicted satisfaction only for wives (R2 =
.019) and not for husbands (R2 = .009).
Another sample from the Midwestern United States also replicated the association
between self-rated personality and satisfaction (Donnellan, Conger, & Bryant, 2004).
This sample of 418 couples completed a 60-item measure of the Big Five (NEO-FFI;
Costa & McCrae, 1989) and rated overall marital quality with a two item measure
assessing happiness and satisfaction. There were significant correlations, ranging from r
= .11 to r = .23, between self-rated personality and marital quality for four of the Big Five
traits (Extraversion, Agreeableness, Conscientiousness, and Emotional Stability). In this
sample, self-rated Openness was not associated with self-rated marital quality (r = .02
and .01 for husbands and wives respectively).
Overall it appears that for at least three traits (Emotional Stability, Agreeableness,
and Conscientiousness), cross-sectional analyses demonstrate a reliable association
between self-rated personality and relationship satisfaction. However, because in most of

5

these research designs both the personality and satisfaction measures are being completed
by the same person, it is possible that these estimates may be influenced by shared
method variance. In other words, it is possible that these associations are due to
similarities in the responses to a measure rather than the hypothesized association
between psychological constructs. One way to address this concern is to test the
association using independent raters for the two variables in order to remove any shared
method variance.
A study adopting this type of design found there was still a significant association
between ratings of personality by acquaintances and self-rated marital satisfaction (Kelly
and Conley, 1987). In fact, informants’ reports of personality were not only associated
with marital satisfaction assessed during the same time period, but the effects held even
when the initial personality assessment was used to predict marital satisfaction and
stability more than forty years later. Specifically, low levels of Emotional Stability for
wives and low levels of impulse control and Emotional Stability for husbands were
associated with lower satisfaction and higher rates of divorce. Husbands’ and wives’
scores on those three personality traits actually accounted for more of the predictable
variance in marital outcomes than the attitudinal, social-environment (stability and
closeness of the family of origin, religious practices, etc.), and sexual history variables
combined. These personality characteristics accounted for 15% of the variance and the
other 14 variables together accounted for only 9% more of the variance in marital
outcomes (Kelly & Conley, 1987). In addition to using multiple respondents, the
longitudinal design of this study demonstrated a lasting association between an
individual’s personality traits and his or her relationship satisfaction over time.

6

Partner’s Big Five Personality and Relationship Satisfaction
The meta-analytic correlations described by Heller et al. (2004) provide a
summary of the individual-level associations between personality and relationships.
However, by definition, relationships are interdependent systems made up of more than
just two distinct individuals. In addition to an individual’s personality affecting her or
his own judgments of relationship satisfaction, a romantic partner’s personality
characteristics may also play a role. Therefore, a more complete understanding of the
personality-relationship association requires a consideration of the personality traits for
both relationship partners.
The literature describing partner effects for personality on relationship outcomes
is not as extensive or as consistent as the evidence regarding actor effects. Although
several studies have examined how relationship satisfaction and functioning relate to the
personality traits of a person’s romantic partner the results do not always replicate across
samples and measures (e.g., Barelds, 2005; Botwin, Buss, & Shackelford, 1997;
Donnellan, Assad, & Conger, 2007; Robins, Caspi, & Moffitt, 2000; Watson, 2000). The
trait with the most consistent evidence of partner effects is Emotional Stability. Having a
partner low in Emotional Stability is associated with higher levels of relationship
dissatisfaction (e.g.; Barelds, 2005; Botwin et al., 1997), and even a greater risk of
marital dissolution (Kurdek, 1993). Similarly, people high in Negative Emotionality have
spouses who are less satisfied (Robins, et al., 2000). However, even with this most
consistent trait, the results sometimes diverge across different samples. For example,
Watson et al. (2000) found that self-reports of Negative Emotionality predicted partners’
satisfaction in a sample of married couples (r = -.27 for women, r = -.58 for men), but

7

Negative Emotionality did not significantly predict partners’ satisfaction in a sample of
dating couples (r = -.13 for women, r = -.16 for men). Inconsistencies such as this one
may simply be the result of insufficient statistical power to detect an effect if it is there.
Partner effects are generally smaller in magnitude than actor effects (probably due at least
in part to the fact that there is no shared method variance for partner effects). However,
additional research with larger samples and greater power is needed to clarify the current
inconsistencies in the literature.
Analyses looking at partner effects for the remaining Big Five traits are even less
consistent than those for Emotional Stability. The same traits are found to be significant
predictors of relationship satisfaction in some samples and not in others. There are also
inconsistent gender differences, with a specific trait predicting satisfaction for one gender
but not the other. Unfortunately, many of these effects do not replicate across samples.
For example, one study found that partners’ Extraversion and Agreeableness were the
only Big Five traits that were significant predictors of marital quality (Barelds, 2005).
However, another study found that Agreeableness, Openness, and Emotional Stability
showed small but consistent partner effects (Watson, Klohnen, Casillas, Simms, Haig, &
Berry, 2004). Adding more inconsistent results to the mix, Watson et al. (2000) reported
that partners’ Extraversion and Agreeableness predicted women’s (but not men’s)
relationship satisfaction in a sample of married couples but that only the Agreeableness
effect was replicated in a sample of dating couples.
The frequency with which these effects appear, albeit sporadically and
inconsistently, suggests that inadequate sample sizes may be at least partially responsible
for the ambiguous results. In their meta-analysis, Karney and Bradbury (1995) point out

8

that much of the relationship research they analyzed had relatively small sample sizes
(33% had 100 or fewer participants), and that this limits researchers’ ability to detect
small effects.
The relative paucity of large sample sizes in research examining partner effects
was further illustrated by a recent meta-analysis that sought to clarify the existing
evidence regarding partner effects for the Big Five traits and relationship satisfaction
(Malouff, Thorsteinsson, Schtte, Bhullar, & Rooke, 2010). Their search of the literature
found only ten published studies with a total of 19 effect sizes for partner effects.
However, when these studies were used to calculate overall effect sizes, Malouff and
colleagues found significant associations between partner personality and relationship
satisfaction for four of the Big Five traits. Consistent with the narrative review above,
the strongest effect was for Neuroticism (r = -.22) followed by Agreeableness (r = .15),
Conscientiousness (r = .12), and Extraversion (r = .06). The meta-analysis did not find a
significant association between relationship satisfaction and partners score on Openness.
These results suggest that using large enough samples to provide sufficient statistical
power may help to clarify the inconsistencies regarding partner effects in the literature.

Personality Similarity and Relationship Satisfaction
The saying, ―birds of a feather flock together,‖ describes a widely held belief that
people tend to congregate and select romantic partners based on similarity on a variety of
characteristics. Based on this folk wisdom, psychologists have examined whether people
are more attracted to similar others (e.g., Botwin et al., 1997) and even whether spouses
become more similar over time (Caspi, Herbener, & Ozer, 1992). Research on assortative

9

mating, or homogomy, indicates that couples do tend to be similar on a variety of factors
including demographic characteristics and attitude domains. Similarity correlations
between spouses can be as high as r = .77 for age and r = .45 for education level (Watson
et al., 2004). Couples also tend to be quite similar on a range of attitude domains; Luo
and Klohnen (2005) reported profile similarity estimates for a sample of newlyweds
ranging from r = .48 for political attitudes to r = .72 for religiosity.
However, there is less evidence for assortative mating for personality traits.
Instead, the correlations between partners on personality traits are generally much lower
than those for values or demographics. For example, published reports of the correlations
between couples’ self-rated personality traits on the BFI range from a low of r = .02 for
both Neuroticism and Conscientiousness to r = .17 for Extraversion (Watson et al., 2004).
Lykken and Tellegen (1993) reported a correlation of just r = .08 between spouses for
scores on 10 personality scales on the MPQ. A Dutch sample of 690 heterosexual
couples also showed low correlations between the partners’ personality scores, with
correlations between spouses of only r = .04 for Neuroticism and r = .04 for Dominance
(Barelds, 2005). However, there is some evidence that congruence between partners may
vary by trait. Rammstedt & Schupp (2008) used data from the German Socio-Economic
Panel Study, a nationally representative sample including over six thousand married
couples, to show that spouses’ scores on Agreeableness, Conscientiousness, and
Openness were moderately correlated (r = .26, .31, and .30 respectively). In contrast, the
same sample showed only small associations between spouses’ scores for Extraversion (r
=.11) and Emotional Stability (r =.06).

10

Regardless of whether, on average, people select partners that are very similar, it
is still possible that similarity is good for relationships. The question remains as to
whether couples with more similar personalities have more successful relationships. A
review of the literature shows that the empirical evidence on the issue is quite mixed. On
one hand, personality similarity has been reported to be a significant predictor of marital
satisfaction in several studies (e.g., Gaunt, 2006; Gonzaga et al., 2007; Karney &
Bradbury, 1995). On the other hand, several other studies find no association between
personality similarity and relationship outcomes (e.g.; Donnellan et al., 2007, Gattis,
Berns, Simpson, & Christensen, 2004; Robins et al., 2000; Watson et al., 2004).
Among the research showing that similarity is important for relationship
satisfaction is a study examining 248 Israeli couples. Gaunt (2006) examined whether
masculine (e.g., self-reliant, analytical), feminine (e.g., compassionate, tender), and
neutral (e.g., friendly, sincere) personality traits were associated with marital satisfaction.
Two different similarity measures were tested; an absolute difference measure assessing
the overall discrepancy between partners’ scores and a profile correlation measure that
assessed similarity in terms of the pattern of responses. For both indexes of similarity,
Gaunt reported that similarity was, ―strongly associated with relationship measures‖ (p.
1416). The profile correlation index of similarity was found to be correlated with marital
satisfaction r = .30 for wives and r = .33 for husbands. The absolute difference measure
of similarity was correlated with marital satisfaction less strongly, with r = -.18 for wives
and r = -.15 for husbands, again indicating that less difference between spouses was
associated with higher marital satisfaction.

11

Another recent study using profile similarity correlations also reported significant
associations between couples’ personality similarity and relationship functioning
(Gonzaga et al., 2007). The authors reported that personality similarity as measured by a
profile correlation significantly predicted relationship quality (d = .76 in Study 1, d =
1.01 in Study 2). Along the same lines, a meta-analysis synthesizing research on the
association between similarity and relationship satisfaction reported that personality
homogomy was reliably associated with marital satisfaction (r = .35 for wives, r = .28 for
husbands; Karney & Bradbury, 1995). However, in discussing these results, the authors
called attention to an important statistical issue inherent in the analysis of similarity
effects. Karney and Bradbury cautioned that most of the studies included in their review
did not control for the initial levels of personality before assessing the effects of
similarity. Consequently, the authors warned that despite their meta-analytic estimates, it
was possible that, ―homogomy itself does not affect marital outcome beyond the initial
levels of particular variables‖ (Karney & Bradbury, 1995, p. 21).
To illustrate the importance of controlling for initial personality traits before
interpreting similarity effects, consider a hypothetical couple who are quite dissimilar on
the trait of Emotional Stability and who report low relationship satisfaction. At first
glance, this would seem to support the idea that personality similarity is important for
relationship satisfaction. However, it is necessary to consider the circumstances that are
required in order for a couple to have dissimilar scores on a trait. In order to find a large
absolute difference between a couples’ scores on Emotional Stability, one spouse must
score quite high and the other spouse must score quite low. Measures of similarity are by
definition confounded with their constituent parts. As noted in the discussion of actor

12

effects, there is a well-established empirical association between low scores on Emotional
Stability and negative relationship outcomes. Individuals scoring low on Emotional
Stability are more likely to be dissatisfied with a relationship and are even more likely to
divorce. Therefore, because similarity is necessarily linked with the individuals’ initial
levels, it is unclear whether the hypothetical couple’s problems are the result of
dissimilarity or simply the result of one partner’s low Emotional Stability. In fact,
because similarity scores are always dependent on the partners’ initial levels, no
conclusions about the effects of similarity can be made without first controlling for the
main effects of personality.
Although failing to control for the initial personality scores for each partner
prevents any clear interpretation for the effects of personality similarity (Kenny &
Acitelli, 1994), the analyses in many of the similarity studies described previously
(Gaunt, 2006; Gonzaga et al., 2007; and the 1995 Karney & Bradbury review) did not
take this step to appropriately control for the main effects of both partners’ personalities.
Therefore, these articles cannot provide a clear interpretation of the unique effect of
personality similarity on relationship satisfaction.
In contrast, those investigations that have appropriately controlled for the main
effects of personality have found either no evidence, or only inconsistent evidence, of
personality similarity predicting relationship satisfaction. For example, Watson et al.
(2004) used a sample of 291 newly married couples to test whether similarity on the Big
Five traits was associated with greater relationship satisfaction after controlling for the
main effects of personality. The authors reported that similarity was unrelated to
relationship satisfaction for Extraversion, Agreeableness, and Emotional Stability

13

(Watson et al., 2004). In the same sample, similarity on Conscientiousness and Openness
was associated with husbands’ (but not wives’) relationship satisfaction.
Several additional studies that take the step of controlling for initial personality
scores have also found little evidence that relationship satisfaction is related to partner
similarity. In fact, Gattis et al. (2004) did not find personality similarity on any of the Big
Five traits to be a significant predictor of marital satisfaction in a sample of distressed
and treatment-seeking couples. Donnellan et al. (2007) also found no evidence that
personality similarity assessed by the Multidimensional Personality Questionnaire (MPQBF) predicted relationship outcomes. Finally, Big Five personality similarity did not
predict relationship satisfaction in a community sample of couples in their 40s and 60s
(Shiota & Levenson, 2007).
Interpretations regarding the influence of personality similarity on relationship
satisfaction depend in large part on whether the analyses control for the main effects of
individuals’ personality traits. The divergence resulting from this analytic choice serves
as a clear example of how important it can be to account for the interdependencies
inherent in relationships when conducting research using relationship outcomes.

Statistical Considerations for Analyses of Similarity
Similarity is by its nature a complicated concept. As noted above, one important
statistical issue is whether or not couple similarity affects relationship quality even after
controlling for the main effects of the personality traits involved in the calculation of
couple similarity (Griffin, Murray, Gonzalez, 1999; Kenny et al., 2006). The concern is
that similarity scores are confounded by their constituents such that the simple correlation

14

between the similarity score and relationship quality is an inappropriate basis for judging
the importance of couple similarity (Griffin et al., 1999). Unfortunately, as described
earlier, the important step of controlling for the ―main effects‖ of personality is frequently
overlooked when the impact of similarity is investigated. This fact makes a clear reading
of the existing literature quite difficult (Watson et al., 2004).
Another important statistical consideration concerns the choice of method for
assessing similarity. All methods of computing similarity indexes involve measuring and
then comparing two or more things. In the case of personality similarity, any index of
similarity gives an estimate of the degree to which two individuals are the same or
different on the trait or traits of interest. However, there are a variety of ways to make
these comparisons and the choice of methods has both theoretical and statistical
implications for the resulting conclusions.
There are two general categories of similarity indexes; those that measure
differences between scores and those that measure similarity in the pattern of those
scores. For example, the simplest version of the first category describing differences is
simply to calculate the absolute difference between spouses’ scores. This type of index
expresses the degree to which the couples’ mean scores are similar or different with larger
numbers indicating greater dissimilarity. Difference indexes are good for assessing
whether couples who have more similar mean level scores are more satisfied but they are
not sensitive to different patterns of similarity across the trait dimensions. For example, a
couple may vary on the mean level of their traits but still represent similarity of a
different kind if they both score highest on Extraversion and lowest on Openness. To
pick up on these kinds of patterns, the second category of similarity measures are needed.

15

The most basic example of these indexes is the profile correlation, where the trait scores
for one spouse are correlated with the scores for the other spouse. In this case, higher
scores indicate more similar patterns of scores. Profile correlations can be used to test
whether couples who are more similar in terms of which traits they score highest or
lowest on are more satisfied, but profile correlations do not pick up on differences in the
mean level of those traits. Because there are conceptually different types of similarity, a
complete examination of whether similarity is related to satisfaction requires an
examination of both types of indexes in order to detect similarity effects from either type.
In fact, the issue of measuring similarity is even more complex than these two
categories might suggest. In 1953, Cronbach and Gleser published a paper describing
many of the issues inherent in the study of similarity. First, they noted that there is no
such thing as ―general‖ similarity and that similarity can only apply to specific
dimensions. Two things or two people cannot simply be ―similar‖; they must be similar
on some particular characteristic or set of characteristics. Furthermore, it is likely that
similarity on different qualities will have different and distinct effects. This is apparent in
investigations of ―personality similarity‖ as it is likely that similarity on one trait (e.g.,
Extraversion) will have quite different implications for satisfaction than partner similarity
on another trait (e.g., Conscientiousness).
In addition to these conceptual issues, the statistical techniques used to study
similarity can also have important implications for the interpretations of similarity
effects. All methods of calculating similarity require ―reducing the configuration or the
relationship between two configurations to a single index, [and require that] we discard
much of the information of the score set‖ (Cronbach & Gleser, 1953, p. 457). Although

16

all indexes lose some information, different indexes of similarity discard different kinds
of information. For instance, calculating similarity by adding up or averaging the
absolute differences between partners’ scores provides an estimate of the effect of mean
levels on these traits but eliminates information about whether similarity on specific
characteristics are related to satisfaction. There is no way to tell whether satisfaction is
different for couples who are slightly different on many characteristics (e.g., score .5
point different on each of the Big Five traits) compared with couples who score the same
on most traits but are very different on just one characteristic (e.g., score the same on
Extraversion, Agreeableness, Conscientiousness and Openness but have scores that are
2.5 points different on Emotional Stability). In contrast, the profile correlation index
described earlier calculates similarity by correlating partners’ scores. The profile
correlation index captures similarity in the pattern or configuration of scores across
partners, but removes evidence about the effect of the mean level scores. Using a profile
correlation approach there is no way to tell whether satisfaction is different for couples
who have the same configuration of scores but different mean levels. For example, there
is no way to tell the difference between a couple who shows the same pattern of scoring
highest on Conscientiousness and Extraversion and 1 point lower on Emotional Stability
and Openness from another couple with the same pattern who score 2 points higher on
the Conscientiousness and Extraversion than on Emotional Stability and Openness.
To address this issue, Cronbach and Gleser (1953) described three distinct
characteristics of similarity profiles; elevation, scatter and shape. Elevation refers to the
average of all scores for a person and can also be referred to as the level (Kenny, et al.,
2006). In other words, similarity in elevation is simply similarity in the mean across all

17

items. Two students would be similar in elevation if they had the same test average at the
end of the semester, regardless of the pattern of grades across the semester. Scatter (also
known as spread) refers to the variability of the scores across items. Two students who
are similar in scatter would have the same overall variability across test grades at the end
of the semester, although not necessarily the same variability across each of the
individual tests. Finally, shape describes the differences between scores after accounting
for difference in elevation and scatter. Students showing similarity in the shape of their
grades would show the same pattern of highs and lows across tests but would not
necessarily have the same average score on specific tests or at the end of the term. For
example, if both students scores lowest on the first test and highest on the third test of the
term they would show similarity in shape, regardless of the actual scores each student
obtained. Importantly, although these three factors all assess ―similarity‖ they measure
different forms of similarity and can vary independently. For example one pair of
students may have similar shape but very different elevation, while another pair might
have similar elevation but different amounts of scatter.
Theoretically, these different forms of similarity across personality traits might
have very different implications for relationship satisfaction. Figure 1 presents a visual
illustration of these different forms of similarity, representing personality profiles for
three hypothetical couples. Figure 1A presents data for a couple who is similar in terms
of shape, but has different elevation. In other words, the husband and wife have the same
pattern of responses such that they rate themselves highest on item O and lowest on item
E, but the man rates himself 1 point higher on each item. An examination of the different
indexes of similarity in the accompanying table shows that the profile correlation index

18

of similarity does not pick up on this type of dissimilarity and the correlation is 1.0. The
discrepancy index of correlation does recognize the difference in elevation and is
represented in the discrepancy index value of 1.0.
Figure 1B on the other hand, represents a couple who have similar shape but
different amounts of scatter. Both spouses have the same peak (item C) and low points
(items E and O), but in this case the man uses more extreme responses and has more
variability in his answers. This similar pattern of their responses is reflected in the high
profile correlation and the discrepancy index registers the differences between partners in
elevation or level. Finally, Figure 1C depicts a couple with the same general shape but
different elevation and different scatter. Again both spouses have the same peak and low
points, but the man has a higher average score across items (elevation) and is more
variable in his responses (scatter). The difference in the response patterns is also indexed
in the smaller correlation between the spouses’ scores.
As Figure 1 illustrates, these different forms of similarity represent different
theoretical ideas of what it means to be similar and are reflected by different similarity
indexes. As such, questions regarding the effects of similarity in one form (e.g., shape)
might result in quite different conclusions than questions regarding the effects of another
form of similarity (e.g., scatter). The appropriate choice of an index of similarity
depends on the characteristics of similarity that are most relevant for the question at hand.
Kenny et al. (2006) describe six different dyadic indexes that can be used to
assess these different forms of similarity; three measuring dissimilarity (discrepancy, d2,
and distance) and three measuring similarity (correlation, covariance, and intraclass
correlation). The authors highlight the forms of similarity described earlier (elevation,

19

scatter and shape) and demonstrate which forms can be assessed by each index of
similarity. Two of these indexes are most relevant to the current questions regarding
personality similarity. For research in which the similarity in the shape of two profiles is
most important, the correlation coefficient is the recommended index. However, in
calculating a correlation the information about the mean level (elevation) is removed.
Therefore, when similarity effects due to differences in elevation are of interest (such as
in Figure 1A), dyadic indexes such as the discrepancy are more appropriate.
Unfortunately, the existing literature examining personality similarity and
relationships is very sparse regarding what kinds of similarity are most important. It is
possible for similarity in any of these three characteristics (elevation, scatter, or shape) to
be an important predictor of relationship satisfaction. For example, it is possible that
similarity of shape is an important factor for satisfaction. If this were the case, then
couples who have similar patterns in terms of the traits they score high on and the traits
they score low on would report the highest satisfaction. Alternatively, it is also possible
that the factor driving personality similarity effects could be scatter, such that matching a
partner in terms of the amount of variability in responses is the most important factor for
satisfaction. Therefore, in order to fully test the similarity-satisfaction relation, it is
necessary to use a combination of similarity indexes that capture these unique types of
similarity. To address these concerns, the current study uses a combination of different
indexes of similarity to examine the complete picture of the effects of any or all of these
forms of similarity.
A final, but important statistical consideration for the study of similarity has to do
with statistical power. It should be kept in mind that the similarity score is essentially an

20

interaction effect (e.g., Kenny et al., 2006) and it is well known that very large samples
are required to reliably detect interaction effects (Aiken & West, 1991; McClelland &
Judd, 1993). This is particularly relevant to research examining partner effects and
similarity because these studies require data from both members of a couple. This often
makes recruitment more difficult and many of the existing studies have limited sample
sizes.
Beyond sheer numbers, another concern is that research on relationships may be
particularly likely to create selection effects such that participants who choose to take part
may be more motivated to focus on their relationships than the general population. For
example, couples who are particularly satisfied or particularly distressed might be more
likely to seek out and participate in research studies examining romantic relationships. In
fact, at least one of the published studies examining personality effects on satisfaction
specifically recruited couples through mental health professionals and included primarily
distressed couples in their analyses (e.g.; Gattis et al., 2004). It is difficult to know
whether participants who volunteer to be a part of research on relationships accurately
represent the general population of married couples. One could argue that the difficulty
and inconvenience inherent in organizing participation for both couple members could
lead to samples including unrepresentative couples who are more motivated to focus on
their relationships than couples in the general population.
In light of these sampling and analytic concerns, and to address the inconsistent
results in the current literature, additional research using large samples and appropriate
statistical techniques is needed to clarify the effects of partner personality and personality
similarity on relationship satisfaction.

21

Overview of Research
To test for effects of personality and personality similarity on relationship
satisfaction, the current project made use of a wealth of data available from two large,
nationally representative samples; the British Household Panel Study (BHPS), and the
Household Income and Labour Dynamics in Australia Survey (HILDA). Participants in
both of these ongoing panel studies are interviewed yearly regarding a variety of
individual and household demographics, economic conditions, values and opinions. Both
studies recently added a measure of the Big Five personality traits and thus provided a
unique opportunity to test for personality effects on relationship satisfaction using large,
representative samples of married couples.
These panel studies provided larger samples than any used in previous attempts to
measure personality effects on relationship satisfaction. The large samples, which
include thousands of married couples, are particularly impressive considering that the
forty studies included in the meta-analysis examining the association between Emotional
Stability and marital satisfaction included a combined total of 7,640 participants (Heller
et al., 2004). Another advantage of these nationally representative panel studies was the
ability to test for personality effects in married couples across a wide variety of ages and
relationship lengths.
In addition to providing adequate statistical power to detect similarity effects if
they exist, these large samples also provided an opportunity to examine how analytic
choices might influence conclusions regarding the associations between personality
similarity and relationship satisfaction. By comparing the results obtained using three

22

different methods of assessing similarity, this project investigated whether some of the
inconsistencies in the current literature result from these analytic choices. Finally, it was
possible to cross-validate the results for all of these analyses by comparing the estimates
across the two distinct samples.
A final goal of this project was to test a related conceptual question regarding
personality effects on well-being. Using data from the same two samples as well as a
third panel study, the German Socio-Economic Panel Study (GSOEP); it was possible to
test whether interpersonal effects of personality extend beyond the specific domain of
relationship satisfaction to a more general measure of overall life satisfaction. Evidence
of partner effects for overall life satisfaction would provide convincing support for more
generalized intrapersonal effects of personality extending beyond the relationship
domain. Thus, the final set of analyses used the APIM to estimate the associations
between the Big Five personality traits, personality similarity, and life satisfaction.
To summarize, the current project used data from large, nationally representative
panel studies to address the following research questions:
1.) What are the associations between the Big Five personality traits of each spouse
and relationship satisfaction?
2.) Is personality similarity associated with relationship satisfaction when
appropriately controlling for the personality traits of each spouse?
3.) Are the conclusions drawn regarding personality similarity effects dependent on
the dyadic index of similarity that is chosen? If similarity matters for relationship
satisfaction, what characteristic (i.e., elevation, scatter, shape) of similarity is
important?

23

4.) Do the interpersonal effects of personality extend beyond the specific domain of
relationship satisfaction? Are there “partner” and similarity effects of
personality for overall life satisfaction?

Method
Sample one - British Household Panel Study
The British Household Panel Study (BHPS) is a nationally representative,
household-based panel study designed to study social and economic change in residents
of the United Kingdom. In the 2005 wave of data collection, the BHPS included a
measure Big Five personality traits for the first time. Therefore, to investigate personality
similarity effects, the current analyses selected those participants who were legally
married and provided data in that wave of assessment. This included 3,277 heterosexual
couples for whom personality and relationship satisfaction data were available for both
spouses (n = 6,554). The average age of the husbands was 51.67 (SD = 14.53) and the
average age of the wives was 49.42 (SD = 14.08). Full descriptive statistics for each
variable are provided in Table 1.
Measures
Marital satisfaction was assessed in the BHPS by asking respondents to report
how satisfied or dissatisfied they felt about their husband or wife using a seven point
scale from 1 ―not at all satisfied‖ to 7 ―completely satisfied‖ (M = 6.32, SD = 1.08). The
correlation between wives and husbands relationship satisfaction scores was .29 (p < .01).
Participants reported their life satisfaction using a one-item measure with endpoints from
1 ―not at all satisfied‖ to 7 ―completely satisfied‖ (M = 5.29, SD = 1.20, r between

24

partners = .26). Personality was assessed using a fifteen-item measure derived from the
Big Five Inventory (BFI; John & Srivastava, 1999). For each trait, participants reported
the extent to which each of a series of three statements described how they see
themselves using a 1-7 scale with endpoints ―does not apply to me at all‖ to ―applies to
me perfectly‖. For example, the indicators of Extraversion (α = .53) were ―is talkative‖,
―is outgoing, sociable‖ and ―is reserved‖ (reverse scored). Each of the remaining traits
were also assessed using three items and showed similar levels of internal consistency
(Agreeableness α = .52, Conscientiousness α = .50, Emotional Stability α = .67, and
Openness α = .66). These short scales have shown to be efficient measures of the Big
Five personality traits and have an impressive degree of convergent validity with the
longer BFI scales. For example, Donnellan and Lucas (2008) reported correlations
between these three-item measures and the full BFI scales ranging from .86 to .90, and
correlations between these three-items and the remaining five to seven items in each BFI
scale ranging from .70 to .73. Finally, information regarding the length of marriage
measured in years was available for 1,837 of the participating couples (M = 21.40, SD =
16.04). Data on the length of marriage was not available for the remaining couples in the
sample.
Sample Two - Household Income and Labour Dynamics in Australia Survey
The Household Income and Labour Dynamics in Australia Survey (HILDA) is a
second household based panel study. After selecting those participants who were legally
married and for whom personality and relationship satisfaction data was collected from
both spouses in 2005, this sample included 2,639 couples (n = 5,278). Demographic
characteristics of this sample were comparable to those in the BHPS, the husbands were

25

an average of 50.96 (SD = 14.64) years old and wives had a mean age of 48.49 (SD =
14.28). Table 1 displays descriptive statistics for all variables used in these analyses.
Measures
Marital satisfaction was assessed by asking respondents to report how satisfied or
dissatisfied they felt about their relationship with their partner using an eleven point scale
(0 = ―completely dissatisfied‖ to 10 = ―completely satisfied‖, M = 8.39, SD = 1.84). In
the HILDA dataset, the correlation between husbands and wives relationship satisfaction
was .53 (p < .01). Participants reported on life satisfaction using a one-item measure
with endpoints of 1 ―completely dissatisfied‖ to 10 ―completely satisfied‖ (M = 8.08, SD
= 1.33, r between partners = .38). The HILDA contains a longer measure of personality
based on Saucier’s (1994) adjective-based measure of the Big Five. Personality was
assessed using a 36-item adjective rating measure. Participants rated the degree to which
a series of adjectives described them (1 = ―does not describe me at all‖ to 7 = ―describes
me very well‖). Eight items served as indicators of Extraversion (α = .78). Six items
were used to assess the trait of Openness (α = .72). The other three personality traits,
Agreeableness (α = .77), Conscientiousness (α = .80), Emotional Stability (α = .79) were
each assessed using seven adjective ratings. Finally, marriage length was measured in
years and data on this variable was available for 2,609 of the responding couples (M =
22.78, SD = 15.88).
Sample Three – German Socio-Economic Panel Study
A third household based, nationally representative panel study was used to
investigate personality and personality similarity effects on overall life satisfaction. The
German Socio-Economic Panel Study (GSOEP) has been assessing a variety of social

26

and demographic characteristics of residents of Germany since 1984. In 2005, the
GSOEP included the same 15-item measure of Big Five personality used in the BHPS
(Sample 1). After selecting those participants who were legally married and for whom
personality and life satisfaction data is available for both spouses, this sample includes
5,709 couples (n = 11,418). In this sample, the husbands were an average of 53.7 (SD =
13.8) years old and the wives had a mean age of 51.0 (SD = 13.6). See Table 1 for full
descriptive statistics.
Measures
Marital satisfaction was not assessed in the GSOEP. Respondent reported on life
satisfaction by reporting how satisfied or dissatisfied they felt about their life overall
from 1 ―not at all satisfied‖ to 10 ―completely satisfied‖ (M = 7.02, SD = 1.78). The
correlation between wives and husbands life satisfaction scores was .56 (p < .01).
Personality was again assessed using the same fifteen-item measure used in Sample 1
(BHPS). The reliabilities of these short scales were generally comparable to those in the
UK sample (Extraversion α = .64, Agreeableness α = .51, Conscientiousness α = .60,
Emotional Stability α = .61, and Openness α = .64).
Analysis Plan
The Actor-Partner Interdependence Model (APIM; Kenny et al., 2006) was used
to test whether relationship satisfaction was related to the Big Five personality traits or to
the degree of similarity or difference between partners’ personality. This multi-level
modeling technique provides estimates of the influence of an individuals’ own personality
(actor effects) and the influence of his or her partners’ personality (partner effects) on
relationship satisfaction while accounting for issues of nonindependence between

27

husbands and wives. Importantly, these analyses also address concerns evident in some
of the existing literature by controlling for the main effects of each partner’s personality.
Prior to analysis, each of the personality traits and satisfaction scores were standardized
across the entire sample.
The step of standardizing scores using the sample mean is particularly important
for testing similarity effects because it eliminates stereotype accuracy as another potential
explanation for agreement across partners’ trait scores. Stereotype accuracy describes the
finding that responses to item measures are not random; there are normative or
―stereotypical‖ responses such that some items are generally more endorsed than others.
For instance, a stereotypical response would be to rate sunny days as more enjoyable than
gray days and it is likely that both spouses would agree on these ratings. However,
similar ratings of weather by a husband and wife do not reflect meaningful similarity in
their psychological experience but simply reflect the normative response. Standardizing
scores across the entire sample removes the increased agreement due to these normative
patterns of responding and allows an uncontaminated test of personality similarity
(Kenny et al., 2006).
To provide a comprehensive test of similarity effects on satisfaction, analyses
were conducted using three different methods of calculating similarity. First, similarity
was calculated as the absolute difference between husbands’ and wives’ scores. This
index captures similarity in level or elevation and can be used to examine the effects of
similarity on a single trait at a time. Next, a discrepancy measure of similarity was
computed to capture the average level of similarity across all five traits and test whether
overall similarity in elevation was related to satisfaction. Finally, the profile correlation

28

was used as the third index of similarity. Although the profile correlation is not sensitive
to mean level differences in personality, it picks up on similarity in the pattern of the
characteristics.
In addition to using three measures of similarity to assess the effects of similarity
in both level and shape, two different sets of analyses were used to test for similarity at
different levels of specificity. Both the discrepancy and profile correlation indexes were
calculated two ways, at both the trait and item levels. For example, discrepancy was first
calculated at the trait level by averaging the differences in mean scores across the Big
Five traits. Next, discrepancy was calculated again at the item level by averaging the
mean differences across all of the items (i.e., all 15 item scores in the BHPS and GSOEP
and all 36 items in the HILDA). Likewise, a trait profile correlation was calculated by
correlating partners’ trait scores and an item-level profile correlation was calculated by
correlating the spouses’ item responses.
The same series of analyses were conducted on both the BHPS and HILDA data
and the results showed a great deal of consistency in the results across samples. For
clarity, the detailed results for the BHPS are presented first and then detailed results for
the HILDA sample are given only when the conclusions vary across the two data sets.
Complete results for both samples are also reported in Tables 2-7 of the appendix.

Relationship satisfaction as a function of actor personality, partner
personality, and similarity.
The literature suggests that the choice of a similarity index can have important
consequences for interpreting the influence of similarity on an outcome of interest (e.g.;

29

Cronbach & Gleser, 1953; Kenny et al., 2006). As described earlier, similarity is not a
general construct and no one index can capture all of the characteristics that fall under the
concept of similarity. Therefore, the analyses that follow examine the question using
three different measures of similarity and dissimilarity. Comparing the results across
indexes, and cross-validating across two different large samples can help to isolate the
effects of specific kinds of similarity (e.g., level or shape) and address concerns in the
existing literature.
The current project uses a combination of similarity indexes that capture unique
aspects of similarity. The first index of similarity is simply the absolute difference
between spouses’ scores for a personality trait. This straightforward measure technically
measures dissimilarity as smaller values imply greater similarity (Kenny et al., 2006).
The absolute difference index picks up on mean-level differences and addresses questions
regarding similarity of elevation (Cronbach & Gleser, 1953), also referred to as level.
However, as discussed earlier, the absolute difference measure is not sensitive to
similarity in profile or shape. An additional advantage of using the absolute difference
index of similarity is that it allows tests of the unique contribution of similarity on
specific traits (e.g., Extraversion) to be examined one at a time in univariate tests.
The second index of similarity, the discrepancy, is the sum of the absolute
differences between spouses divided by the number of items (Kenny et al., 2006). Unlike
the first set of univariate analyses that examined each trait separately and in isolation, the
discrepancy measure indexes the differences in a couple’s scores across all five traits
together. The discrepancy measure provides an index of the overall similarity or
differences in elevation across the Big Five traits.

30

The third index of similarity employed in the current analyses is the profile
correlation. Calculated by correlating spouses’ scores, it is sensitive to similarity in terms
of shape or the pattern of scores. Unlike the discrepancy measure, the profile correlation
index removes information about the mean level so it does not capture similarity in
elevation. The profile correlation measure is sensitive only to similarity or differences in
shape. Together, these indexes provide a comprehensive test for effects due to similarity
in both elevation and similarity in shape as originally described by Cronbach and Gleser
(1953).

Results
Univariate analyses with absolute difference as index of similarity
As an initial test of whether personality similarity predicted relationship
satisfaction, the first set of analyses measured similarity as the absolute difference of trait
scores between the spouses. A set of five APIM analyses were conducted to test whether
similarity on a trait predicted relationship satisfaction after controlling for mean levels of
that personality trait. In other words, an individual’s relationship satisfaction was
predicted to be a function of the person’s own personality score (i.e., the actor effect),
their partner’s personality score (i.e., the partner effect), and the absolute difference
between their personality score and their partner’s score. Interactions of the actor and
partner effects with gender were also included in order to determine if the impact of
personality traits on relationship satisfaction varied for wives versus husbands. A
description of the pattern of results is provided below and a full listing of the APIM
estimates is provided in Table 2.

31

BHPS- Univariate analyses with absolute difference as similarity index
In the British sample (BHPS), the univariate analyses showed consistent evidence
of actor effects. Higher scores on Extraversion, Agreeableness, Conscientiousness,
Emotional Stability and Openness were significantly associated with greater reported
relationship satisfaction. Consistent with previous research examining actor effect, the
largest associations were for Agreeableness (standardized b = .205, p < .01),
Conscientiousness (standardized b = .158, p < .01), and Emotional Stability (standardized
b = .103, p < .01). These univariate analyses also showed significant partner effects for
three of the traits (Agreeableness standardized b = .078, p < .01, Conscientiousness
standardized b = .043, p < .01, and Emotional Stability standardized b = .072, p < .01)
such that a spouses’ standing on these traits were also related to individuals’ relationship
satisfaction. However, there was very little evidence that similarity between partners on
these traits was associated with relationship satisfaction. In fact, only Emotional Stability
had a significant effect for the absolute difference index of similarity, and the effect was
very small (standardized b = .034, p < .05). The estimates for the other four traits were
not significant and close to zero (Extraversion standardized b = .000, Agreeableness
standardized b = .005, Conscientiousness standardized b = .007, and Openness
standardized b = -.001).
HILDA- Univariate analyses with absolute difference as similarity index
Conducting the same analyses on the Australian sample (HILDA) provided a
similar pattern of results. Again, there was consistent evidence of actor effects for all five
traits. An individual’s score on the Big Five traits significantly predicted his or her own
relationship satisfaction such that people higher in Extraversion, Agreeableness,

32

Conscientiousness, and Emotional Stability were also more satisfied with their
relationships. The effect for Openness, though significant, was in the opposite direction
in the HILDA data. That is, in the Australian sample individuals who were higher in
Openness were actually significantly less satisfied with their relationships (standardized b
= -.085, p < .01). Evidence for partner effects was more consistent in the Australian data
than the British sample. The partner’s score on each of the Big Five dimensions was
significantly associated with an individual’s relationship satisfaction.
Despite consistent evidence for actor and partner effects across samples, there
were no consistent associations between relationship satisfaction and the absolute
difference measure of similarity across the two samples. As shown in Table 2, all of the
similarity effects were small and most were not significant. In fact, the only two
similarity effects that were significant in the Australian sample (Extraversion and
Openness) were not replicated in the British sample. Also, the one significant similarity
effect in the British sample (for Emotional Stability) was not found in the Australian
sample.
It should be noted that two interactions with gender emerged as significant in the
univariate analyses just described. In the BHPS, the partner effect of Extraversion
interacted with gender (standardized b = .029, p < .05) such that men with more
Extraverted wives were more satisfied (standardized b = .044, p = .01) and the effect of
husbands’ Extraversion on wives was not significant (and slightly negative, standardized
b = -.014, ns). This effect was not replicated in the HILDA (standardized b = .003, ns),
however a different interaction emerged in the Australian sample such that spouses’

33

Agreeableness was more important for women (standardized b = .169, p < .01) than for
men (standardized b = .097, p < .01).
Relationship satisfaction as a function of all five traits simultaneously
In light of evidence that Emotional Stability is a particularly strong predictor of
marital outcomes (e.g., Karney & Bradbury, 1995); it was important to test whether the
personality and personality similarity effects remained when all five traits were used as
simultaneous predictors of relationship satisfaction. Therefore, an additional
simultaneous APIM analysis estimated the actor, partner, and similarity effects for all five
traits at the same time. This step was important because although the Big Five are
conceptually independent dimensions of personality, scales that measure the Big Five
traits are often correlated in practice (e.g. Funder, 2001). This simultaneous analysis
served as a conservative test to help evaluate the unique influence of each trait. It
directly addressed concerns raised by Karney and Bradbury (1995) identifying the
importance of examining whether other personality traits account for significant variance
after controlling for the influence of Emotional Stability.
As shown in Table 3, including all five traits as simultaneous predictors of
relationship satisfaction did not change the interpretation derived from the univariate tests
described above. In fact, the pattern for both the British (BHPS) and Australian (HILDA)
samples again showed that the actor effects are largest in magnitude. Only the actor
effect for Openness (standardized b = -.021, ns) in the BHPS was not significant. The
other four traits in the BHPS and all of the traits in the HILDA showed significant
associations between individual’s scores on personality and their own relationship
satisfaction. In the simultaneous equation, partner effects remained significant for both

34

Agreeableness (standardized b = .068, p < .01 in BHPS, .070, p < .01 in HILDA) and
Emotional Stability (standardized b = .062, p < .01 in BHPS, .087, p < .01 in HILDA) in
both samples. The absolute difference index of similarity was not significant for any of
the Big Five personality traits in the British sample, although both Extraversion
(standardized b = -.041, p < .05) and Openness (standardized b = -.055, p < .01) were
significantly related to relationship satisfaction in the Australian sample. The picture
remains clear that there are no replicable effects of partner similarity across the two
samples, either in the univariate analyses or the more conservative simultaneous test
including all traits at once.
Discrepancy as an index of similarity
In order to investigate how the choice of similarity index can determine the
interpretation of these associations, the next step was to evaluate the effects using another
index of similarity. The discrepancy measure was chosen as an overall index of similarity
because it is sensitive to similarity in elevation or level. Although it is insensitive to the
pattern of responses, the discrepancy measure averages the differences between spouses’
scores on the five trait scores. Additionally, to address the effects of measuring trait
similarity versus item-level similarity, two sets of analyses were conducted. The first
calculated the average discrepancy across trait scores and controlled for trait scores. The
second set of analyses calculated the average discrepancy across items and controlled for
the item scores. Computing discrepancy at both the trait and item level and controlling
for partners’ trait scores provided a comprehensive test of the associations between
personality similarity and relationship satisfaction while addressing methodological
concerns in some of the existing research addressing this question.

35

BHPS – Trait discrepancy
The results displayed in Table 4 show that the discrepancy measure resulted in the
exact same pattern of associations for the BHPS as those found in the simultaneous
analyses of the absolute difference index. Once again, an individual’s own personality
was significantly associated with relationship satisfaction for all trait domains except
Openness (standardized b = -.021, ns). The same two significant partner effects for
Agreeableness (standardized b = .067, p < .01) and Emotional Stability (standardized b =
.061, p < .01) emerged. Consistent with the absolute difference method of calculating
similarity, the trait-level discrepancy index was not a significant predictor of relationship
satisfaction (standardized b = -.004, ns).
BHPS – Item discrepancy
The possibility remained that important item-level effects of similarity might be
missed when calculating discrepancy scores using trait level data. To address this, the
second set of discrepancy analyses tested whether the results changed when discrepancy
scores were calculated across all fifteen personality items in the BHPS and the item
scores were included as control variables. As shown in Table 5, the results for the BHPS
were identical to those for the trait-level discrepancy measure and the absolute difference
measure of similarity. Once again, discrepancy was not a significant predictor of
relationship satisfaction (standardized b = -.004, ns).
HILDA – Trait discrepancy
Table 4 shows the pattern of results for the HILDA sample when the trait-level
discrepancy measure of similarity was used to predict relationship satisfaction. The
actor and partner effects showed the same pattern of results for the HILDA as in the

36

analyses using the absolute difference measure of similarity (all five actor effects and the
partner effects for Agreeableness and Emotional Stability were significant predictors of
relationship satisfaction). In this case, however, the discrepancy index of similarity
calculated with trait factors was a statistically significant predictor of relationship
satisfaction (standardized b = -.058, p < .01). This indicates that individuals who had
personality trait scores more similar to their spouses’ trait scores reported higher
relationship satisfaction.
HILDA – Item discrepancy
When the discrepancy measure was calculated at the item-level (average
discrepancy across all 36 personality items for each couple) instead of at the more general
trait level, the pattern of results changed somewhat for the HILDA data. In this second
set of analyses, when the item-level discrepancy measure was used to predict relationship
satisfaction, the discrepancy effect dropped to standardized b = -.032, p = .308 and was
no longer statistically significant. Overall, the pattern of results shown in Tables 2, 3, 4,
and 5 do not support the idea that similarity in elevation (as measured by the absolute
difference and discrepancy indexes) is systematically related to relationship satisfaction
for either the BHPS or HILDA samples.
Profile correlation as index of similarity
Although the previous analyses tested whether similarity in level was associated
with relationship satisfaction, the possibility remained that similarity in the shape or
profile of personality traits is important for relationship satisfaction. If so, a similarity
index sensitive to shape might yield a different answer to the question of whether
couples’ personality similarity is associated with relationship satisfaction. To address

37

this, profile correlations were used as the third index of similarity. Profile correlations
were entered into the same sets of analyses used with the discrepancy measure; one using
the profile correlation calculated across trait personality scores and one using profile
correlations calculated across items. Both spouses trait scores were also included in each
equation to control for the main effect of personality.
BHPS – Profile correlation as index of similarity
In the British sample, the analyses using profile correlations calculated as the
index of similarity yielded the same pattern of results as the analyses measuring
similarity using absolute difference and discrepancy indexes. As shown in Table 6, again
there was evidence that an individual’s personality traits are associated with their own
relationship satisfaction as the actor effects were significant for Extraversion
(standardized b = .033, p < .05), Agreeableness (standardized b = .164, p < .01),
Conscientiousness (standardized b = .081, p < .01), and Emotional Stability (standardized
b = .063, p < .01). The same significant partner effects emerged for Agreeableness
(standardized b = .067, p < .01) and Emotional Stability (standardized b = .062, p < .01)
as described earlier for the analyses using other indexes of similarity. And importantly,
the trait-level profile correlation index of partner similarity was again not significantly
related to relationship satisfaction with an estimate close to zero.
The second approach to testing profile correlation as a measure of similarity also
provided very consistent results. Here the profile correlations were calculated by
correlating husbands’ and wives’ scores on all of the personality items rather than
correlating their scale scores. The item-level profile correlations again showed that
couples’ personality similarity was unrelated to relationship satisfaction. When

38

controlling for trait factor scores, the item-level profile correlation was unrelated to
relationship satisfaction. The actor and partner effects were also consistent regardless for
the profile correlation analyses (see Tables 6 and 7 for full list of estimates).
HILDA – Profile correlation as index of similarity
In the Australian sample, the results of the actor and partner effect analyses using
trait-level profile correlations were quite similar to those found using the discrepancy
similarity index. The actor effects were significant for each of the trait domains except
for Conscientiousness where the estimate was in the same direction as previous analyses
but did not reach statistical significance (standardized b = .027, ns). The partner effects
were consistent with both the BHPS data set and other analytic approaches in the HILDA
(individuals with more Agreeable and Emotionally Stable partners were more satisfied).
However, in the HILDA sample, the trait-level profile correlation was significantly
associated with relationship satisfaction (standardized b = .070, p < .01). In the
Australian sample, couples who were more similar in the shape of their personality
profiles reported higher relationship satisfaction.
As seen in Tables 6 and 7, the analyses using profile correlations as the index of
similarity consistently indicated that profile similarity was associated with relationship
satisfaction in the HILDA sample. This was true whether profile correlation was
computed across traits or across items.
Overview of relationship satisfaction results
The data presented in Tables 2 - 7 demonstrate consistent evidence of actor effects
for personality such that an individual’s score on the Big Five traits are associated with
their own ratings of relationship satisfaction. There was also clear and consistent

39

evidence across both the BHPS and HILDA samples that individuals with more agreeable
and emotionally stable spouses are more satisfied with their marriages.
The analyses investigating the effects of personality similarity consistently
showed no association between similarity in elevation or level and relationship
satisfaction. However, one type of similarity did show inconsistent results across
samples. Although similarity in shape (as indexed by the profile correlation) was not
related to relationship satisfaction in the BHPS sample, profile similarity was
significantly associated with relationship satisfaction in the HILDA sample. It is not
entirely clear why these differences emerged across samples. One unlikely possibility is
that personality plays a different role in marriage in Australia compared to the UK.
Another possibility is that the differences arise from the different personality measures
used in the two samples, either because the personality measures capture different content
of the personality constructs or because the scales have different levels of precision and
measurement error. Although the exact explanation for these differences is not readily
apparent, an additional set of analyses using the same samples but examining a different
outcome variable can help to shed light on some of these potential explanations and
further elucidate the effects of personality within marriage.
Life Satisfaction: A More General Measure of Well-being
In order to broaden the scope of the question regarding the importance of spouses’
personality similarity on important outcomes, the same series of analyses just described
was also conducted with a measure of general life satisfaction as the outcome. Testing
for actor, partner, and similarity effects of personality on a more general measure of wellbeing provided both the ability to look for broader implications of personality and to

40

provide context to interpret the effects found for relationship satisfaction. An additional
advantage of examining the outcome of overall life satisfaction is the ability to use a
third, still-larger nationally representative sample from Germany to cross-validate the
findings.
All of the same analyses described for relationship satisfaction were conducted
using life satisfaction as the outcome. To clarify the presentation, the results for all three
datasets are reviewed together in order to highlight the consistencies and differences
across samples.
Actor Effects for Life Satisfaction
Consistent with previous research (e.g., Heller et al., 2004), there was clear
evidence of actor effects such that an individuals’ personality traits were systematically
related to their own ratings of life satisfaction. The impressive evidence of intrapersonal
effects of personality can be seen in the univariate analyses shown in Table 8. In fact, all
five traits were consistently significantly associated with life satisfaction across all three
samples and all methods of calculating similarity with only one exception. The only
exception was for the trait of Openness in the BHPS. Although the actor effect for
Openness was significant in the univariate analyses where only one trait domain was
tested at a time, in the more conservative simultaneous test including all five personality
traits Openness was not significantly related to life satisfaction. (Table 9 includes a full
report of the effect estimates for actor, partner, and similarity effects for the simultaneous
analysis.)

41

Partner Effects for Life Satisfaction
These three large data sets also provided additional evidence that ratings of
overall life satisfaction are systematically associated with the personality traits possessed
by a spouse. Finding any evidence of partner effects is compelling support for the
importance of the intrapersonal effects of personality, particularly when examining a
domain outside of the romantic relationship such as life satisfaction. Partner effects
demonstrate that an individual’s satisfaction with life is reliably associated with the
personality characteristics of his or her spouse, a particularly high standard considering
that the reports of personality and satisfaction come from different people. Therefore, it
is not surprising that the partner effects were generally smaller in magnitude and less
consistent across the three samples compared with the actor effects. This is also
consistent with previous research, and the smaller magnitudes of these effects are likely
due at least in part to the fact that the data is not biased by shared method variance.
An examination of univariate results in Table 8 shows that individuals with more
Agreeable, Conscientious, and Emotionally stable partners reported significantly higher
life satisfaction across all three samples. In addition to those three consistent partner
effects, the HILDA sample showed a significant partner effect for Extraversion and the
GSOEP sample showed significant partner effects for all five personality domains in the
univariate analyses.
A less consistent pattern of partner effects emerged across the three nationally
representative samples when all five traits were entered simultaneously to predict life
satisfaction. As shown in Table 9, partner’s Agreeableness was a significant predictor of
life satisfaction in the BHPS but not in the HILDA or GSOEP. In contrast, partners’

42

scores on Extraversion and Conscientiousness were significantly associated with life
satisfaction in the HILDA but not in the other two samples. Finally, the trait of Openness
showed a partner effect in only the GSOEP sample. The pattern of conclusions just
described for the simultaneous analyses using the absolute difference measure were
consistent across each of the additional indexes of similarity. Tables 10, 11, 12, and 13
display the estimates and show the same pattern of partner effects emerged regardless of
the index of similarity or whether it was calculated across the scale or the item scores.
Although the evidence for partner effects for the other traits varied across the
samples, it is important to note that even in the conservative simultaneous test, the
partners’ level of Emotional Stability consistently emerged as a significant predictor of
life satisfaction in all three nationally representative samples (see Table 9). The partner’s
Emotional Stability score was a significant predictor across all samples and all methods
of computing similarity. The fact that these effects replicated across all three samples and
for a non-relationship domain is new and particularly impressive evidence of the
intrapersonal effects of personality.
Similarity Effects for Life Satisfaction
Tables 8-13 also display the results from testing for personality similarity effects
on life satisfaction. There was no consistent evidence that personality similarity between
spouses was associated with life satisfaction after controlling for the main effects of
personality. The analyses examining absolute difference measures of similarity did show
some inconsistent effects that were different across samples. For example, in the
simultaneous analysis, individuals in the British sample who were more similar to their
partners on Agreeableness reported lower overall life satisfaction (standardized b = .035,

43

p < .05) whereas the same set of analyses in the German sample found that individuals
who were more similar to their partners on Agreeableness reported higher life satisfaction
(standardized b = -.025, p < .05). Neither of these contradictory findings were replicated
in the Australian sample which showed no significant effect for similarity in
Agreeableness (standardized b = .010, ns). See Table 9 for full report of the estimates. In
contrast, the HILDA data did show small but significant associations between life
satisfaction and similarity on Extraversion (standardized b = -.034, p < .05),
Conscientiousness (standardized b = .035, p < .05), and Openness (standardized b = .031, p < .05).
The analyses using the discrepancy indexes of similarity calculated with trait
scores showed no evidence of similarity effects on life satisfaction for any of the three
samples (standardized b = .015, ns in BHPS, standardized b = .004, ns in HILDA, and
standardized b = -.002, ns in GSOEP). When discrepancy was calculated at the itemlevel, the results in the HILDA and GSEOP were consistent, showing that similarity was
not related to life satisfaction (standardized b = .020, ns controlling for traits and .025, ns
controlling for items for HILDA and standardized b = .012, ns controlling for traits and
.009, ns controlling for items for GSOEP). The only exception to this was when
discrepancy was calculated at the item level in the BHPS. In the British sample, the itemlevel discrepancy measure of similarity was significantly associated with life satisfaction
(standardized b = .041, p < .01 controlling for traits and .028, p < .05 controlling for
items standardized). (See Table 11 for full information and specific estimates for each
trait.)

44

The final set of analyses used the profile correlation measure of similarity to test
whether individuals whose partners showed similar patterns of personality traits were
more satisfied with life. The results for the profile correlation index were very consistent;
there was no significant association between profile correlations and life satisfaction
across any of the three samples regardless of whether the profile correlation was
calculated across trait or item scores. The profile correlation calculated with trait scores
was not a significant predictor of life satisfaction in the BHPS (standardized b = -.015,
ns), the HILDA (standardized b = -.013, ns), or the GSOEP (standardized b = -.009, ns)
samples. When profile correlations were calculated across items instead of trait scores
there was still no significant association in any sample (see Table 11 for full results).

Conclusion
The current studies provided a unique opportunity to test for actor, partner, and
similarity effects of personality for relationship satisfaction. Using nationally
representative samples, these analyses had sufficient power to help clarify the
associations between Big Five personality traits and marital satisfaction. To put these
samples into context, together the BHPS and HILDA samples included 5,916 couples or
11,832 individuals. This represents more people than the total participants of all 40
studies included in the Heller et al. (2004) meta-analysis that demonstrated reliable actor
effects personality on marital satisfaction. The BHPS sample alone included 3,277
couples, almost twice as many participants as the 3,848 individuals represented by the ten
separate studies included in the single meta-analysis to date examining partner effects of
personality (Malouff et al., 2010).

45

Not only are the BHPS and HILDA samples larger than those used in previous
analyses, but they represent the full spectrum of legally married residents of the United
Kingdom, Australia, and Germany. The large sample sizes and the inclusion of a wide
variety of ages and marital lengths provided a valuable opportunity to address the
inconsistencies and confusion in the existing literature with sufficient statistical power to
detect even small effects of personality. An additional benefit of using representative
panel studies to address these questions is that these representative samples are free from
potential bias associated with many studies that rely on volunteer participants who are
particularly motivated to be part of research examining relationships. Also, by using
multiple samples, the results could be cross-validated in separate samples with different
measures of personality and satisfaction to further clarify the effects of personality.
Finally, by using the Actor-Partner Interdependence model to appropriately control for
the main effects of personality, these analyses provided an unmatched opportunity to
estimate the associations between personality similarity and relationship satisfaction.
Consistent with existing research, the data from these nationally representative
samples replicated the consistent actor effects previously found for the Big Five
personality traits and relationship satisfaction. Self-reports of personality were reliably
associated with reports of relationship and life satisfaction across each of the three
samples. The current studies add to the extensive literature demonstrating intrapersonal
effects of personality.
In addition, this study provided support for the intrapersonal effects of personality
in marriage and provided estimates of the effects of partner’s personality for relationship
satisfaction using larger samples than previously available. The question of how

46

personality traits relate to the satisfaction of a romantic partner is a particularly
challenging one to answer. Specifically, not only must researchers avoid selection biases
often inherent in relationship studies and obtain data from both members of the couple,
the sample must also be large enough to provide sufficient power to detect an effect even
without the benefit of shared method biases. The three nationally representative studies
examined here were particularly well-suited to address these concerns.
These challenges related to measuring interpersonal effects of personality have
limited the available research and resulted in inconsistent evidence across studies,
hampering the development of theory regarding the consequences of personality in
relationships. By addressing the methodological issues of selective recruitment,
statistical power, and appropriately controlling for the main effects of personality, these
studies provide a potential explanation for the inconsistencies in the existing literature.
They also help to develop the body of evidence regarding interpersonal and dyadic effects
of personality. Specifically, they demonstrate the importance of controlling for main
effects of personality and rule out insufficient power as an alternative explanation for the
lack of significant similarity effects.
The general pattern of evidence did not reveal any consistent evidence that
personality similarity is reliably associated with relationship or life satisfaction.
However, a primary goal of these analyses was to test whether the conclusions drawn
regarding personality similarity are dependent on the dyadic index of similarity used and
whether this could help explain inconsistencies in the existing literature. To this aim,
analyses were repeated using several different indexes of similarity that captured different
types of similarity. The evidence was clear for similarity in elevation, the average value

47

of personality scores. Using both the absolute difference and discrepancy measures of
similarity across both the BHPS and HILDA samples, there was no consistent evidence
that similarity was associated with relationship satisfaction. Together with the previous
studies that used appropriate statistical controls for personality, these findings clarify that
similarity in elevation or mean scores is not an important factor for relationship
functioning. In other words, couples who are more discrepant on the Big Five traits are
no more likely to be particularly satisfied or unsatisfied with their relationships.
When similarity in shape, or the pattern of scores was tested by the profile
correlation index of similarity the BHPS again showed no association between profile
similarity and relationship satisfaction. Couples with a more similar pattern of
personality scores were no more or less satisfied than couples who showed different
profiles. However, the results from the HILDA diverged, indicating that couples with
more similar profiles were more satisfied. There are several potential explanations for
this divergence. The failure to find consistent effects could result from the different
personality measures used; it is possible that the larger item set for the HILDA
personality measure may have allowed for a more precise measure of profile shape.
However, the failure to find the same profile similarity effects in the BHPS is still
surprising considering the personality measure it included is brief but reliable and the
large sample size should have provided sufficient power to pick up consistent but small
effects. Another possibility is that culture is a moderator for the effect of profile
similarity. Although personality and the Big Five traits are generally considered
universal and not culturally specific (McCrae & Costa, 1997), cultural differences cannot
be ruled out by the current analyses and may play an important role in the expectations

48

and interactions inherent in marital relationships. In the end, it is not possible to prove
that an association does not exist in a single study, and demonstrating the lack of a
similarity effect is a conservative test. Only future research with additional measures and
sufficiently large samples will be able to clarify the current inconsistencies regarding
profile similarity.
Despite finding some inconsistencies across samples, these studies provide clear
evidence that the personality traits of both an individual and his or her spouse are
important for relationship satisfaction. There is also consistent evidence that similarity or
differences in the mean levels of these traits are not important for relationship success.
The discrepancy between samples for the profile correlation index suggests that the
configuration of traits may possibly be a more important type of similarity for
relationship satisfaction. Importantly, these differences confirm that similarity is not a
unitary construct and that analytic choices are of particular importance for testing
questions of similarity.
These data also provided an opportunity to extend beyond the specific domain of
relationship satisfaction and examine the associations between personality and an overall
measure of well-being. The final set of analyses provided strong support for the literature
showing that intrapersonal effects of personality are reliable and important. Actor effects
across three large and nationally representative samples provided consistent evidence that
personality traits are consistently related to well-being. In addition, the life satisfaction
analyses provide new evidence of the interpersonal effects of personality. This
particularly conservative test showing that a spouse’s personality traits are reliably
associated with life satisfaction demonstrates that a distal predictor of behavior shapes

49

satisfaction, even beyond the relationship domain. Although the current studies do not
directly address the mechanisms of how these interpersonal effects occur, they are
consistent with the vulnerability-stress-adaptation model of marriage and existing
evidence regarding interactional processes as a mediator for the link between personality
and relationship outcomes (Karney & Bradbury, 1995; Donnellan et al., 2007). In
addition, finding that interpersonal effects extend to general life satisfaction suggests that
future research into the processes that create these links is likely to help build a more
complete understanding of how far-reaching the link is between personal dispositions and
relationship outcomes.
Finally, the use of three large samples representing the UK, Australia, and
Germany provided a unique opportunity to cross-validate the results. By identifying the
consistencies across cultures and the areas of divergence, these analyses can help
highlight the effects of personality while also identifying the effects of methodological
choices. These analyses demonstrated compelling consistencies across cultures with
robust actor effects across the Big Five traits, and consistent interpersonal effects,
particularly for Emotional Stability. Perhaps most importantly, the current study shows
that after controlling for the main effects of personality, there is no consistent evidence
that personality similarity is important for relationship or life satisfaction.
Future Directions
The current studies provide a more comprehensive test of actor, partner and
similarity effects of personality for satisfaction in married couples than previously
available. However, several questions still remain. First, the ability to use such large and
nationally representative samples required a trade-off in terms of the scope of personality

50

measure used. Research often requires a trade-off between fidelity and breadth.
Although there is good evidence that the brief personality measures used in these samples
are valid indicators of the Big Five (e.g., Donnellan & Lucas, 2008), the shorter measures
of personality used in these large scale projects inevitably have lower reliability than
more comprehensive personality scales. If longer, more reliable measures of personality
could be applied in the large samples required for this type of analysis, it is possible that
small effects of personality similarity might be uncovered.
Another possibility is that similarity in the broad traits of the Big Five are not related
to satisfaction but an investigation of more specific facets of personality have unique
dyadic effects. In particular, it might be worth investigating the facets underlying the
traits with the most consistent evidence of intrapersonal and interpersonal effects to test
for potential dyadic effects. It is possible that the consistent associations between
relationship satisfaction and Emotional Stability, Agreeableness, and Conscientiousness
might reflect more specific dyadic combinations at the facet level.
Another question that is important to a full understanding of the interpersonal effects
of personality similarity is to explore whether similarity predicts future relationship
outcomes. The current studies examine cross-sectional associations between partners’
personality and satisfaction but it is also possible that personality similarity better
predicts the trajectory of a relationship rather than current satisfaction. In fact, a study
examining 67 couples over 12 years suggested that greater personality similarity was
actually associated with more negative slopes in martial satisfaction trajectories (Shiota
& Levenson, 2007). The panel data used in the current studies will yield a unique
opportunity to test this question in the coming years with the advantages of large and

51

nationally representative samples. All three samples are continuing yearly data
collection, and the personality data collected in 2005 will provide an opportunity to test
personality as a predictor of changes in marital satisfaction. An additional benefit of
using longitudinal data of this nature is that behavioral measures can also be included.
For example, personality similarity can be tested as a predictor of important life outcomes
including marital stability or divorce.
Although the current studies address many of the methodological issues that make it
difficult to design a clear test of the effects of personality similarity on relationship
satisfaction, future research will continue to refine our understanding of how individual
and dyadic variables are related to relationship functioning. For now, the current studies
support previous research and provide unique evidence of intrapersonal and interpersonal
effects of personality for relationship and life satisfaction. Across several large,
nationally representative samples, there is clear evidence of actor and partner effects of
personality for both marital and life satisfaction. These studies also show that despite the
sufficient statistical power provided by these large samples and the use of different
indexes to capture similarity in both elevation and profile shape, there is no consistent
evidence that the dyadic variable of similarity matters over and above the actor and
partner effects of personality.

52

Figure 1:
Hypothetical Data to Illustrate Three Forms of Similarity
Adapted from Kenny, Kashy & Cook, 2006, p.324-325

E
A
C
ES
O
Mean
SD
Discrepancy Index
Correlation Index

Dyad 1
Wife
Husband
1
2
2
3
2
3
3
4
4
5
2.4
3.4
1.02
1.02
1.00
1.00

Dyad 2
Wife
Husband
1
0
2
2
3
5
2
2
1
0
1.8
1.8
0.75
1.83
0.80
0.99

53

Dyad 3
Wife
Husband
1
2
2
3
3
7
2
5
1
3
1.8
4
0.75
1.79
2.20
0.90

54

APPENDIX

55

Table 1:
Descriptive Statistics for Personality and Relationship Satisfaction Variables
Men
Women
r between
d
Mean (SD) Mean (SD)
couples
BHPS
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness
Relationship Satisfaction
Life Satisfaction

4.29 (1.14)
5.29 (1.00)
5.34 (1.04)
4.76 (1.21)
4.47 (1.19)
6.38 (1.04)
5.26 (1.18)

4.53 (1.19)
5.65 (.93)
5.49 (1.05)
4.02 (1.29)
4.36 (1.20)
6.27 (1.12)
5.32 (1.22)

-.20*
-.37*
-.14*
.57*
.10*
.10*
-.04

.03
.06*
.09*
.04*
.16*
.29*
.26*

HILDA
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness
Relationship Satisfaction
Life Satisfaction

4.50
5.29
5.17
5.15
4.19
8.52
8.00

(.92)
(.84)
(.95)
(.98)
(1.01)
(1.74)
(1.32)

4.75
5.75
5.41
5.19
4.08
8.27
8.15

(.99)
(.73)
(.95)
(.98)
(1.04)
(1.93)
(1.35)

-.26*
-.56*
-.25*
-.03
.11*
.13*
-.11

.01
.18*
.07*
.20*
.22*
.53*
.38*

4.68
5.27
5.98
4.23
4.38
7.00

(1.11)
(1.00)
(.91)
(1.18)
(1.19)
(1.81)

4.88
5.62
6.05
3.77
4.52
7.05

(1.11)
(.93)
(.86)
(1.20)
(1.23)
(1.73)

-.18*
-.35*
-.07*
.38*
-.12*
-.03

.10*
.26*
.31*
.16*
.34*
.56*

GSEOP
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness
Life Satisfaction
* p < .05

56

Table 2:
Univariate analyses predicting relationship satisfaction as a function of personality, partner personality, and personality similarity
(absolute difference between trait factor scores).
Extraversion
Agreeableness
Conscientiousness Emotional Stability Openness
BHPS
Gender
.054*
.071*
.056*
.039*
.046*
Actor
.076*
.205*
.158*
.103*
.050*
Partner
.015
.078*
.043*
.072*
.012
Absolute Difference
.000
.005
.007
.034*
-.001
Actor X Gender
-.008
-.001
.020
.012
.017
Partner X Gender
.029*
-.010
-.005
-.005
-.022
HILDA
Gender
Actor
Partner
Absolute Difference
Actor X Gender
Partner X Gender
* p < .05

.071*
.119*
.082*
-.056*
.001
.003

.084*
.197*
.133*
.006
.019
-.036*

.069*
.116*
.097*
-.026
.022
-.016

57

.067*
.176*
.146*
-.012
-.017
-.005

.069*
-.085*
-.026*
-.055*
.012
-.021

Table 3:
Simultaneous analyses predicting relationship satisfaction as a function of personality, partner
personality, and absolute differences for all 5 trait factors.
BHPS
HILDA
Intercept
-.010
.019
Gender
.076*
.093*
Actor effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

.032*
.166*
.081*
.063*
-.021

.044*
.135*
.032*
.073*
-.086*

Partner effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.011
.068*
.001
.062*
-.003

.017
.070*
.023
.087*
-.017

Absolute Differences
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.011
.004
.003
.010
-.010

-.041*
.017
-.016
-.006
-.055*

Actor X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.013
-.007
.019
.005
.010

-.005
.034
.022
-.038*
.000

.037*
-.008
.002
-.011
-.025

.022
-.043*
-.002
.001
-.016

Partner X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness
* p < .05

58

Table 4:
Discrepancy Analyses By Traits: Predicting relationship satisfaction as a function of
personality, partner personality, and the sum of discrepancies across all 5 trait factors.
BHPS
HILDA
Intercept
-.012
.011
Gender
.076*
.093*
Discrepancy (trait factors)
-.004
-.058*
Actor effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

.032*
.165*
.081*
.062*
-.021

.045*
.130*
.032*
.070*
-.086*

Partner effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.011
.067*
.001
.061*
-.003

.018
.065*
.023
.084*
-.017

Actor X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.011
-.007
.019
.009
.009

-.001
.018
.020
-.038*
-.001

.036*
-.007
.003
-.015
-.024

.018
-.028
.000
.001
-.015

Partner X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness
* p < .05

59

Table 5:
Discrepancy Analyses by Item: Predicting relationship satisfaction as a function of personality,
partner personality, and the sum of discrepancies across all personality items while controlling
for trait factors.
BHPS
HILDA
Intercept
-.011
.015
Gender
.076*
.093*
Discrepancy (items)
-.004
-.032
Actor effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

.032*
.166*
.081*
.062*
-.020

.043*
.132*
.032*
.070*
-.088*

Partner effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.011
.067*
.001
.061*
-.002

.016
.066*
.024
.082*
-.019

Actor X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.011
-.007
.019
.008
.009

-.000
.023
.022
-.038*
-.004

.036*
-.008
.002
-.014
-.024

.017
-.033
-.002
.001
-.013

Partner X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness
* p < .05

60

Table 6:
Profile Correlation Analyses by Traits: Predicting relationship satisfaction as a function of
personality, partner personality, and profile similarity (of trait factor scores).
BHPS
HILDA
Intercept
-.013
.012
Gender
.077*
.093*
Profile Correlation (traits)
.005
.070*
Actor effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

.033*
.164*
.081*
.063*
-.019

.052*
.121*
.027
.070*
-.070*

Partner effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.011
.067*
-.000
.062*
-.003

.025
.054*
.018
.083*
-.001

Actor X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.011
-.008
.018
.009
.009

-.000
.018
.022
-.032
-.001

.036*
-.007
.002
-.016
-.024

.017
-.028
.002
-.004
-.015

Partner X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness
* p < .05

61

Table 7:
Profile Correlation Analyses by Items: Predicting relationship satisfaction as a function of
personality, partner personality, and profile similarity (of items) while controlling for trait factor
scores.
BHPS
HILDA
Intercept
-.015
.012
Gender
.076*
.094*
Profile Correlation (items)
.022
.089*
Actor effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

.034*
.161*
.078*
.065*
-.017

.052*
.120*
.026
.063*
-.069*

Partner effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.009
.064*
-.002
.064*
-.001

.025
.052*
.016
.077*
-.001

Actor X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.012
-.009
.017
.012
.009

-.001
.019
.021
-.034*
-.001

.036*
-.004
.003
-.018
-.024

.018
-.029
-.001
-.002
-.015

Partner X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness
* p < .05

62

Table 8:
Univariate analyses predicting life satisfaction as a function of personality, partner personality, and personality similarity (absolute
difference between trait factor scores).
Extraversion
Agreeableness
Conscientiousness Emotional Stability Openness
BHPS
Gender
-.011
.007
-.008
-.090*
-.026*
Actor
.125*
.215*
.230*
.337*
.093*
Partner
.019
.055*
.034*
.100*
.017
Absolute Difference
.002
.037 *
.002
.044*
.014
Actor X Gender
.001
-.001
.039*
.019
.018
Partner X Gender
.016
-.005
-.002
.006
.009
HILDA
Gender
Actor
Partner
Absolute Difference
Actor X Gender
Partner X Gender
GSOEP
Gender
Actor
Partner
Absolute Difference
Actor X Gender
Partner X Gender
* p < .05

-.041*
.187*
-.098*
-.033*
.006
.021

-.015
.231*
.095*
.024
.008
-.011

-.044*
.176*
.103*
.035*
.038*
-.014

-.051*
.259*
.119*
.031*
-.013
.006

-.052*
-.038*
-.022
-.022
.013
.003

-.011
.134*
.076*
-.008
-.003
.018

-.002
.123*
.041*
-.038*
.006
-.007

-.013*
.117*
.049*
-.005
.020
-.026*

-.044*
.263*
.119*
.014
.040*
-.011

-.013*
.141*
.079*
-.021
.018
.001

63

Table 9:
Simultaneous analyses predicting life satisfaction as a function of personality, partner
personality, and absolute differences for all 5 trait factors.
BHPS
HILDA
GSOEP
Intercept
-.003
.008
-.007
Gender
-.048*
-.019
-.026*
Actor effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

.030*
.128*
.131*
.293*
-.000

.092*
.094*
.067*
.162*
-.034*

.046*
.055*
.044*
.230*
.086*

Partner effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.018
.038*
-.002
.093*
.007

.041*
.016
.041*
.075*
-.015

.016
.005
.001
.099*
.055*

Absolute Differences
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.010
.035*
-.008
.019
-.001

-.034*
.010
.035*
.029
-.031*

.014
-.025*
-.004
.009
-.005

Actor X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.003
-.017
.026*
.010
.008

-.001
.009
.037*
-.030
-.001

-.010
-.011
.004
.035*
.024*

.004
-.007
-.009
.001
.012

.032*
-.013
-.013
.002
.005

.017
.002
-.016
-.009
-.010

Partner X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness
* p < .05

64

Table 10:
Discrepancy Analyses By Traits: Predicting life satisfaction as a function of personality, partner
personality, and the sum of discrepancies across all 5 trait factors.
BHPS
HILDA
GSOEP
Intercept
-.007
.006
-.006
Gender
-.048
-.019
-.026*
Discrepancy
.015
.004
-.002
Actor effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

.031*
.123*
.133*
.292*
-.001

.094*
.094*
.062*
.157*
-.034*

.045*
.058*
.044*
.231*
.087*

Partner effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.017
.033*
.000
.092*
.006

.043*
.016
.036*
.070*
-.015

.015
.008
.001
.099*
.056*

Actor X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.001
-.024
.028*
.014
.007

.004
.005
.032*
-.030
-.002

-.012
-.005
.004
.037*
.024*

.002
-.000
-.002
-.003
.012

.026
-.008
-.007
.001
.006

.019
-.004
-.016
-.011
-.010

Partner X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness
* p < .05

65

Table 11:
Discrepancy Analyses by Item: Predicting life satisfaction as a function of personality, partner
personality, and the sum of discrepancies across all personality items while controlling for trait
factors.
BHPS
HILDA
GSOEP
Intercept
-.004
.007
-.005
Gender
-.048*
-.019
-.026*
Discrepancy
.041*
.020
.012
Actor effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

.032*
.124*
.135*
.292*
.002

.095*
.095*
.063*
.158*
-.034*

.045*
.059*
.046*
.232*
.087*

Partner effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.016
.035*
.002
.092*
.009

.044*
.016
.037*
.072*
-.015

.015
.009
.002
.100*
.056*

Actor X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.001
-.002
.029*
.010
.008

.005
.007
.032*
-.030
-.002

-.011
-.004
.005
.036*
.025*

.002
-.002
-.003
.000
.011

.025
-.010
-.007
.001
.006

.018
-.005
-.017
-.010
-.011

Partner X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness
* p < .05

66

Table 12:
Profile Correlation Analyses by Traits: Predicting life satisfaction as a function of personality,
partner personality, and profile similarity (of trait factor scores).
BHPS
HILDA
GSOEP
Intercept
-.008
.006
-.004
Gender
-.047*
-.018
-.027*
Profile Correlation (traits)
-.015
-.013
-.009
Actor effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

.028*
.125*
.136*
.290*
-.003

.093*
.097*
.063*
.157*
-.037*

.045*
.059*
.046*
.230*
.086*

Partner effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.021
.037*
.003
.090*
.003

.041*
.019
.038*
.071*
-.018

.014
.009
.002
.097*
.055*

Actor X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.000
-.024
.026*
.014
.006

.005
.005
.032*
-.031
-.003

-.012
-.004
.004
.035*
.025*

.001
.000
-.002
-.004
.013

.026
-.010
-.008
.002
.007

.019
-.005
-.016
-.009
-.011

Partner X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness
* p < .05

67

Table 13:
Profile Correlation Analyses by Items: Predicting life satisfaction as a function of personality,
partner personality, and profile similarity (of item scores).
BHPS
HILDA
GSOEP
Intercept
-.007
.005
-.003
Gender
-.048*
-.018
-.027*
Profile Correlation (items)
-.015
.005
-.023
Actor effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

.028*
.125*
.136*
.290*
-.003

.095*
.094*
.061*
.156*
-.032*

.044*
.061*
.049*
.229*
.085*

Partner effects
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.020
.036*
.002
.090*
.004

.043*
.016
.035*
.070*
-.013

.013
.011
.005
.096*
.054*

Actor X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness

-.001
-.024
.027*
.013
.006

.005
.004
.032*
-.029
-.002

-.011
-.004
.004
.033*
.026*

.001
.000
-.002
-.004
.013

.026
-.009
-.008
.001
.006

.018
-.006
-.016
-.007
-.012

Partner X Gender Interactions
Extraversion
Agreeableness
Conscientiousness
Emotional Stability
Openness
* p < .05

68

Notes
The data used in Sample 1 of this dissertation were made available through the ESRC Data
Archive. The data were originally collected by the ESRC Research Centre on Micro-social
Change at the University of Essex (now incorporated within the Institute for Social and
Economic Research). Neither the original collectors of the data nor the Archive bear any
responsibility for the analyses or interpretations presented here.

Sample 2 of this paper uses confidentialised unit record file from the Household, Income and
Labour Dynamics in Australia (HILDA) survey. The HILDA Project was initiated and is funded
by the Commonwealth Department of Families, Community Services and Indigenous Affairs
(FaCSIA) and is managed by the Melbourne Institute of Applied Economic and Social Research
(MIAESR). The findings and views reported in this paper, however, are those of the author and
should not be attributed to either FaCSIA or the MIAESR.

The data from Sample 3 were made available by the German Socio-Economic Panel Study
(SOEP) at the German Institute for Economic Research (DIW), Berlin.

69

REFERENCES

70

REFERENCES

Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions.
Thousand Oaks, CA: Sage.
Barelds, D. P. H. (2005). Self and partner personality in intimate relationships. European
Journal of Personality, 19, 501-518.
British Household Panel Survey [computer file] ESRC Research Centre on Micro-social Change.
Colchester: The Data Archive [distributor], 2008 . Data files and associated
documentation.
Botwin, M. D., Buss, D. M., & Shackelford, T. K. (1997). Personality and mate preferences: Five
factors in mate selection and marital satisfaction. Journal of Personality, 65, 107-136.
Caspi, A., Herbener, E. S., & Ozer, D. J. (1992). Shared experiences and the similarity of
personalities: A longitudinal study of married couples. Journal of Personality, 62, 281291.
Cronbach, L. J., & Gleser, G. C. (1953). Assessing similarity between profiles. Psychological
Bulletin, 50, 456-473.
Donnellan, M. B., Conger, R. D., & Bryant, C. M. (2004). The big five and enduring marriages.
Journal of Research in Personality, 38, 481-504.
Donnellan, M. B., Assad, K. K., Robins, R. W., & Conger, R. D. (2007). Do negative interactions
mediate the effects of negative emotionality, communal positive emotionality, and
constraint on relationship satisfaction? Journal of Social and Personal Relationships, 24,
557-573.
Donnellan, M. B., & Lucas, R. E. (2008). Age differences in the big five across the life span:
Evidence from two national samples. Psychology and Aging, 23, 558-566.
Funder, D. C. (2001). Personality. Annual Review of Psychology, 52, 197-221.
Gaunt, R. (2006). Couple similarity and marital satisfaction: Are similar spouses happier?
Journal of Personality, 74, 1401-1420.
Gattis, K. S., Berns, S., Simpson, L. E., Christensen, A. (2004). Birds of a feather or strange
birds? Ties among personality dimensions, similarity, and marital quality. Journal of
Family Psychology, 18, 564-574.

71

German Institute for Economic Research (DIW). German Socio-Economic Panel Study—2005
(Wave 2005) [Data file]. Berlin, Germany: Author.
Gonzaga, G. C., Campos, B., & Bradbury, T. (2007). Similarity, convergence, and relationship
satisfaction in dating and married couples. Journal of Personality and Social Psychology,
93, 34-48.
Gottman, J. M. (1998). Psychology and the study of the marital processes. Annual Review of
Psychology, 49, 169-197.
Griffin, D., Murray, S., & Gonzalez, R. (1999). Difference score correlations in relationship
research: A conceptual primer. Personal Relationships, 6, 505-518.
Heller, D., Watson, D., & Ilies, R. (2004). The role of person versus situation in life satisfaction:
A critical examination. Psychological Bulletin, 130, 574-600.
John, O. P., & Srivastava, S. (1999). The Big Five trait taxonomy: History, measurement, and
theoretical perspectives. In L. A. Pervin & O. P. John (Eds.). Handbook of personality:
Theory and Research (2nd ed.) (pp. 102-138). New York: Guilford Press.
Johnson, J. A. (1997). Units of analysis for the description and explanation of personality. In R.
Hogan, J. Johnson, & S. Briggs (Eds.). Handbook of Personality Psychology (pp. 73 –
93). New York: Academic Press.
Karney, B.R., & Bradbury, T.N. (1995). The longitudinal course of marital quality and stability:
A review of theory, method, and research. Psychological Bulletin, 118, 3-34.
Kelly, E. L., & Conley, J. J. (1987). Personality and compatibility: A prospective analysis of
marital stability and marital satisfaction. Journal of Personality and Social Psychology,
52, 27-40.
Kenny, D. A., & Acitelli, L. K. (1994). Measuring similarity in couples. Journal of Family
Psychology, 8, 417-431.
Kenny, D. A., & Cook, W. (2006). Partner effects in relationship research: Conceptual issues,
analytic difficulties, and illustrations. Personal Relationships, 6, 433-448.
Kenny, D. A., Kashy, D. A., & Cook, W. L. (2006). Dyadic data analysis. New York: Guildford.
Kurdek, L. A. (1993). Predicting marital dissolution: A 5-year prospective longitudinal study of
newlywed couples. Journal of Personality and Social Psychology, 64, 221-242. Lykken,
D. T., & Tellegen, A. (1993). Is human mating adventitious or the result of lawful
choice?: A twin study of mate selection. Journal of Personality and Social
Psychology,65, 56-68.

72

Luo, S., & Klohnen, E. C. (2005). Assortative mating and marital quality in newlyweds: A
couple-centered approach. Journal of Personality and Social Psychology, 88, 304-326.
Malouff, J. M., Thorsteinnson, E. B., Schutte, N. S., Bhullar, N., & Rooke, S. E. (2010). The
five-factor model of personality and relationship satisfaction of intimate partners: A metaanalysis. Journal of Research in Personality, 44, 124-127.
McClelland, G. H., & Judd, C. M. (1993). Statistical difficulties of detecting interactions and
moderator effects. Psychological Bulletin, 114, 376-390.
McCrae, R. R., & Costa, P. T. (1997). Personality trait structure as a human universal. American
Psychologist, 52, 509-516.
Rammstedt, B. & Schupp, J. (2008). Only the congruent survive – personality similarities in
couples. Personality and Individual Differences, 45, 533-535.
Robins, R. W., Caspi, A., & Moffitt, T. E. (2000). Two personalities, one relationship: Both
partners’ personality traits shape the quality of their relationship. Journal of Personality
and Social Psychology, 79, 251-259.
Shiota, M. N., & Levenson, R. W. (2007). Birds of a feather don’t always fly farthest: Similarity
in Big Five personality predicts more negative marital satisfaction trajectories in longterm marriages. Psychology and Aging, 22, 666-675.
Watson, D., Hubbard, B., & Wiese, D. (2000). General traits of personality and affectivity as
predictors of satisfaction in intimate relationships: Evidence from self- and partnerratings. Journal of Personality, 68, 413-449.
Watson, D., Klohnen, E. C., Casillas, A., Nus Simms, E., Haig, J., & Berry, D. S. (2004). Match
makers and deal breakers: Analyses of assortative mating in newlywed couples. Journal
of Personality, 72, 1029-1068.
Watson, N. (ed) (2005) HILDA User Manual – Release 5.0, Melbourne Institute of Applied
Economic and Social Research, University of Melbourne.

73