19
Personal Relationships, 19 (2012), 128–145. Printed in the United States of America. Copyright © 2011 IARR; DOI: 10.1111/j.1475-6811.2010.01344.x Personality similarity, perceptual accuracy, and relationship satisfaction in dating and married couples MIEKE DECUYPER, MARLEEN DE BOLLE, AND FILIP DE FRUYT Ghent University, Belgium Abstract The association between actual and perceptual personality similarity and perceptual accuracy on relationship satisfaction is examined in 191 couples. Self- and partner ratings of personality were assessed using the Revised NEO Personality Inventory (P. T. Costa & R. R. McCrae, 1992) and relationship satisfaction using the Relationship Assessment Scale (S. S. Hendrick, A. Dicke, & C. Hendrick, 1998). Actual and perceptual similarity and perceptual accuracy were quantified using the index of profile agreement (R. R. McCrae, 1993) and L. J. Cronbach and G. C. Gleser’s (1953) D-indices. These indices showed large variability in personality profiles within couples and considerable perceptual accuracy between raters. Actual similarity was positively associated with female relationship satisfaction, controlling for personality traits of both partners. Moreover, partial support was obtained for the positive associations between perceptual similarity and accuracy and relationship satisfaction. What qualities do people like in their life partner? For decades, researchers have been striving to identify factors that lead to rela- tionship and marital satisfaction, and the link between personality traits and relation- ship satisfaction has received much attention. Research on the association between person- ality and relationship satisfaction has usually taken an individualistic or a couple-centered approach. The individualistic approach inves- tigates self/actor and partner effects of person- ality traits on relationship satisfaction (Kenny, 1996), whereas the couple-centered approach examines characteristics of the couple rather than treating both partners as individuals. A common dyadic variable used to predict relationship satisfaction is couple similarity Mieke Decuyper, Marleen De Bolle, and Filip De Fruyt, Department of Developmental, Personality and Social Psychology, Ghent University, Ghent, Belgium. Correspondence should be addressed to Mieke Decuyper, Department of Developmental, Personality and Social Psychology, Ghent University, H. Dunantlaan 2, B-9000 Ghent, Belgium, e-mail: Mieke.Decuyper@ Ugent.be. (Luo & Klohnen, 2005), and the key question becomes “Am I happy in my relationship because my partner and I are alike and have a lot in common?” Extensive literature has framed this question in terms of similarity versus dissimilarity of partners’ characteris- tics (Klohnen & Mendelsohn, 1998; Watson et al., 2004). Who owns the definition of the partner’s personality is a potentially important though neglected issue in couple-centered studies (Hofstee, 1994). The majority of relation- ship satisfaction studies examined partners’ similarity relying on self-ratings of person- ality (i.e., actual similarity), but one could argue that the perception of the other part- ner’s personality should be considered as the point of reference to examine discrep- ancy with the actor’s self-ratings. A man, for example, may describe himself as an orderly and conscientious individual, but if he is per- ceived and described by his wife as more sloppy and disorganized, then their person- ality profile similarity scores will be largely different depending on the perspective taken. 128

Personality Similarity

Embed Size (px)

Citation preview

Page 1: Personality Similarity

Personal Relationships, 19 (2012), 128–145. Printed in the United States of America.Copyright © 2011 IARR; DOI: 10.1111/j.1475-6811.2010.01344.x

Personality similarity, perceptual accuracy,and relationship satisfaction in datingand married couples

MIEKE DECUYPER, MARLEEN DE BOLLE, AND FILIP DE FRUYT

Ghent University, Belgium

AbstractThe association between actual and perceptual personality similarity and perceptual accuracy on relationshipsatisfaction is examined in 191 couples. Self- and partner ratings of personality were assessed using the RevisedNEO Personality Inventory (P. T. Costa & R. R. McCrae, 1992) and relationship satisfaction using the RelationshipAssessment Scale (S. S. Hendrick, A. Dicke, & C. Hendrick, 1998). Actual and perceptual similarity and perceptualaccuracy were quantified using the index of profile agreement (R. R. McCrae, 1993) and L. J. Cronbach and G. C.Gleser’s (1953) D-indices. These indices showed large variability in personality profiles within couples andconsiderable perceptual accuracy between raters. Actual similarity was positively associated with female relationshipsatisfaction, controlling for personality traits of both partners. Moreover, partial support was obtained for the positiveassociations between perceptual similarity and accuracy and relationship satisfaction.

What qualities do people like in their lifepartner? For decades, researchers have beenstriving to identify factors that lead to rela-tionship and marital satisfaction, and thelink between personality traits and relation-ship satisfaction has received much attention.Research on the association between person-ality and relationship satisfaction has usuallytaken an individualistic or a couple-centeredapproach. The individualistic approach inves-tigates self/actor and partner effects of person-ality traits on relationship satisfaction (Kenny,1996), whereas the couple-centered approachexamines characteristics of the couple ratherthan treating both partners as individuals.A common dyadic variable used to predictrelationship satisfaction is couple similarity

Mieke Decuyper, Marleen De Bolle, and Filip De Fruyt,Department of Developmental, Personality and SocialPsychology, Ghent University, Ghent, Belgium.

Correspondence should be addressed to MiekeDecuyper, Department of Developmental, Personality andSocial Psychology, Ghent University, H. Dunantlaan2, B-9000 Ghent, Belgium, e-mail: [email protected].

(Luo & Klohnen, 2005), and the key questionbecomes “Am I happy in my relationshipbecause my partner and I are alike and havea lot in common?” Extensive literature hasframed this question in terms of similarityversus dissimilarity of partners’ characteris-tics (Klohnen & Mendelsohn, 1998; Watsonet al., 2004).

Who owns the definition of the partner’spersonality is a potentially important thoughneglected issue in couple-centered studies(Hofstee, 1994). The majority of relation-ship satisfaction studies examined partners’similarity relying on self-ratings of person-ality (i.e., actual similarity), but one couldargue that the perception of the other part-ner’s personality should be considered asthe point of reference to examine discrep-ancy with the actor’s self-ratings. A man, forexample, may describe himself as an orderlyand conscientious individual, but if he is per-ceived and described by his wife as moresloppy and disorganized, then their person-ality profile similarity scores will be largelydifferent depending on the perspective taken.

128

Page 2: Personality Similarity

Personality similarity, accuracy, and satisfaction 129

Figure 1. Representation of actual similarity,perceptual similarity, and perceptual accuracy.

Beyond this viewpoint, also the degree ofagreement between an individual’s person-ality self-description and how she or he isdescribed by her or his partner (i.e., percep-tual accuracy) might be an element affectingrelationship satisfaction. Deviations betweena person’s self-description and the descrip-tion by her or his partner may be in itselfa potential source of conflict in the relation-ship and hence influence relationship satis-faction. Moreover, differences between howa person sees his or her own personality andhow he or she perceives the personality ofhis or her partner (i.e., perceptual similar-ity) might also affect relationship satisfaction.These additional perspectives, that is, “per-ceptual similarity,” that is comparing infor-mant ratings by the partner with self-ratings ofthe actor’s personality, and “perceptual accu-racy,” that is, agreement between self andpartner descriptions, are the subject of thisstudy. These perspectives are compared withthe more traditionally researched similarityindices derived from actor and partner self-descriptions. The different personality simi-larity and accuracy indices are presented inFigure 1.

The prediction of marital satisfaction basedon personality characteristics of both part-ners has a long history and previous researchhas demonstrated the robust relationshipsbetween relationship quality, functioning andoutcomes, and broad personality traits. Per-sonality traits are most commonly assessedusing the Five-Factor Model (FFM; Costa &McCrae, 1992), describing general personal-ity traits along the five major dimensions ofNeuroticism (N), Extraversion (E), Opennessto Experience (O), Agreeableness (A), andConscientiousness (C). Although this model isnot designed to assess dysfunction, the FFMtraits appear to be associated with marital dis-tress (Gattis, Berns, Simpson, & Christensen,

2004; Karney & Bradbury, 1995). Of thesefive dimensions, Neuroticism seems to bethe most consistent predictor of marital orrelationship dissatisfaction (Barelds, 2005;Karney & Bradbury, 1995; Kelly & Conley,1987; Robins, Caspi, & Moffitt, 2000, 2002).However, relationship quality is also sig-nificantly related to the other FFM traits(Barelds, 2005; Karney & Bradbury, 1995;McCrae, Stone, Fagan, & Costa, 1998; Robinset al., 2000, 2002; Watson et al., 2004; White,Hendrick, & Hendrick, 2004). This individu-alistic approach has yielded important resultsconcerning the predictive value of personalityand the associations with relationship satisfac-tion but treats the partners as two unrelatedindividuals and hence does not take the “cou-ple” into account (Luo et al., 2008).

A more recent line of research went beyondthis individualistic approach by examiningdyad-level transactions (Letzring & Noftle,2010). Most of this couple-centered researchhas focused on how actual personality simi-larity between partners is related to personal-ity satisfaction, and these studies have yieldedmixed results. Glicksohn and Golan (2001)found no effect of actual personality similar-ity, as measured with the Sensation SeekingScale (Zuckerman, 1979) and the EysenckPersonality Questionnaire (Eysenck, Eysenck,& Barrett, 1985) on relationship satisfactionin 65 married couples. Likewise, Russell andWells (1991) found no effect of actual per-sonality similarity on relationship satisfac-tion and the study of Watson and colleagues(2004) found little support for the idea thatspouses who have more similar personalitiesare more satisfied. Using a 12-year longitudi-nal design, Shiota and Levenson (2007) evenreported that larger overall Big Five similar-ity predicted negative slopes in relationshipsatisfaction in a sample of long-term mar-ried couples. A number of different studies,however, have observed significant effects ofpersonality similarity on relationship satisfac-tion. In a study of married couples, Caspiand Herbener (1990) found that similarity onthe California Q-Sort (Block, 1971) predictedgreater satisfaction among married couples intheir 40s through 60s. In a cross-sectionalstudy of 248 married couples, Gaunt (2006)

Page 3: Personality Similarity

130 M. Decuyper, M. De Bolle, and F. De Fruyt

showed that greater similarity between part-ners was associated with higher levels of mar-ital satisfaction. Robins and colleagues (2000)found that similarity in Neuroticism was asso-ciated with higher relationship satisfaction incollege-age dating couples. Finally, Gonzaga,Campos, and Bradbury (2007) and Luo andKlohnen (2005) even found positive associ-ations between actual personality similarityand satisfaction when actor and partner effectswere controlled.

Another way to investigate the connectionsbetween personality and relationship qualityat the dyadic level is to examine the impor-tance of perceptual similarity. Perceptual sim-ilarity or existence of significant positivecorrelations between a judge’s self-rating ona trait and his or her rating of his or herpartner on that same trait (e.g., between awoman’s self-rated Agreeableness and her rat-ings of her man’s Agreeableness) has alsobeen called assumed similarity (Watson et al.,2004). This is based on evidence that indi-viduals are more likely to rely on their ownpersonality characteristics when they lack suf-ficient trait-related information about the tar-get and is hence also considered as a ratingbias, referred to as self-based heuristic bias.Thus, perceptual similarity reflects the ten-dency for people to rate others as similar tothemselves. When providing personality rat-ings of a friend or partner, individuals maytake into account the degree to which they feela closer psychological bond to that person andthey may also consider how similar they feelthey are to the target (Weller & Watson, 2009).In this way, individuals who rate their part-ner as more similar to themselves may havea better relationship quality. Results of thestudy of Weller and Watson (2009) showedthat people tended to rate others with whomthey felt close as similar to themselves on theBig Five personality traits. People did not uti-lize this rating strategy when they were ratinga person with whom they have experienceda significant conflict. A recent meta-analysisof Montoya, Horton, and Kirchner (2008) onthe impact of actual and perceived similarityon interpersonal attraction indicated that per-ceived similarity was a significant predictor ofinterpersonal attraction. Moreover, perceived

similarity was more strongly associated withinterpersonal attraction than actual personal-ity similarity. However, in this meta-analysis,relationship satisfaction was considered as ameasure of interpersonal attraction and theperceptual and actual personality similaritywere not confined to personality traits.

A third way to examine the associationsbetween personality and relationship qual-ity at the dyadic level is to investigate theeffect of perceptual accuracy or the agree-ment between self- and informant ratings ofpersonality traits. Acquaintance with a tar-get generally increases the agreement betweenself- and informant ratings of personalitytraits (Funder, Kolar, & Blackman, 1995).Intimate partners and spouses are probablythe informants most closely acquainted withtheir mates and studies showed substantialagreement between self- and spouse ratings(Foltz, Morse, Calvo, & Barber, 1997; Funder,1980; McCrae, Stone, et al., 1998). Correla-tions between self- and spouse ratings typ-ically range between .4 and .6 (McCraeet al., 2008), suggesting convergence but alsoleaving room for substantial discrepancy. Dis-crepancies between self- and partner reports,however, do not necessarily mean that one ofthe raters has an inaccurate view of the per-sonality. Different raters may have differentopinions about a target and each perspectivecan provide valuable information on the indi-vidual’s functioning (McCrae, 1994).

It could be expected that concordant viewsof personality within couples contribute torelationship satisfaction. Perceptual accuracymay have several benefits within relation-ships, such as an increase in perceptions ofcontrol and predictions of how one’s part-ner will behave (Swann, Pelham, & Krull,1989), which could contribute to more ful-filling interactions (Letzring & Noftle, 2010).Intimacy and communication lead to bet-ter and more accurate information aboutthoughts, feelings, and values and are knownto be related to marital satisfaction (Karney &Bradbury, 1997; Smith, Ciarrochi, & Heaven,2008). In addition, serious problems in therelationship might motivate distortions in thepartners’ views of each other and vice versa(McCrae, Stone, et al., 1998). Discrepant

Page 4: Personality Similarity

Personality similarity, accuracy, and satisfaction 131

views of each other’s personality might be asource of relationship conflict in itself. Myself-perception as a “warm” and “empathic”person may be opposed to my partner’sdescription and perception of me as being“egocentric,” causing tension within the rela-tionship. However, positive illusions are alsoa possible source of discrepancy between self-and partner ratings (Murray & Holmes, 1997).Positive illusions are assumed to exist whenindividuals rate their partner as more positiveand attractive than the partner rated himselfor herself. Several studies found that indi-viduals often rate their partner overly posi-tively on characteristics such as “kind” and“intelligent” (Barelds & Dijkstra, 2009) andshowed that positive illusions about a part-ner predicted greater satisfaction, love, andtrust, and less conflict in both dating and mar-ital relationships (Murray & Holmes, 1997).From this perspective, discrepancy betweenself-ratings and partner ratings could be asso-ciated with greater relationship satisfaction.

McCrae, Stone, and colleagues (1998),however, found no support for an associa-tion between perceptual accuracy and mar-ital quality in 47 married couples. Neitherresponse style measure nor characteristics ofthe respondents or relationship were consis-tently related to agreement on any of the BigFive factors or the total personality profile.Most of the correlations were very small inmagnitude (median absolute value was .09).Marital adjustment was unrelated to percep-tual accuracy on the total personality profile(r = −.01), and the only noteworthy corre-lation was the positive relationship betweenagreement on the Agreeableness factor andmarital adjustment (r = .17).

Gill and Swann (2004) also showed thatperceptual accuracy on the Big Five traitswas not significantly associated with threemeasures of relationships quality, but largerperceptual accuracy on specific relevant traits(selected on a couple by couple basis) wasuniquely associated with psychological ben-efits such as increased relationship commit-ment, feelings of being known, and knowingone’s partner. In this study, Big Five traitswere operationalized using 20 items to tapthe Big Five personality factors (John, 1990)

instead of a comprehensive FFM instru-ment. Although McCrae, Stone, and col-leagues (1998) and Gill and Swann (2004) didnot find a significant link between perceptualaccuracy on the Big Five traits and relation-ship satisfaction, it is possible that due to thesmall sample size and the operationalizationof the FFM in the study of Gill and Swann,more subtle effects of accuracy on relationshipsatisfaction were not detected. Two recentstudies (Letzring & Noftle, 2010; Neff &Karney, 2005) reported a positive associationbetween perceptual accuracy and relationshipsatisfaction. In a sample of newlyweds, Neffand Karney (2005) showed that for wives,more accurate personality perceptions werepositively associated with their supportivebehaviors, feelings of marital control, andwhether the marriage ended in divorce dur-ing the first 2 years of marriage. These asso-ciations were not found for husbands. Morerecently, Letzring and Noftle (2010) showedthat perceptual accuracy on broad personal-ity traits was a significant predictor of rela-tionship quality, and the results of this studysuggested that perceptual accuracy is moreimportant for relationship quality in marriedcouples than in dating or cohabiting couples.

The literature review on actual person-ality similarity, perceptual personality sim-ilarity, and perceptual personality accuracyhas shown that the available evidence onthe associations with relationship satisfactionin couples is somewhat inconsistent and attimes difficult to interpret. Several factors maycontribute to this rather inconsistent picture.Some studies investigating the effects of per-sonality similarity on relationship satisfactionused relatively small samples (e.g., Glicksohn& Golan, 2001). Moreover, many researchersused various small or more broadband person-ality measures, making it difficult to directlycompare results. A more systematic compar-ison of personality similarity and perceptualaccuracy in couples could be obtained usinga common and comprehensive personalitydescriptive model such as the FFM (Digman,1990). Many studies do not include all fivepersonality dimensions and often rely exclu-sively on self-reports. The actor’s perceptionof the personality of her or his partner is

Page 5: Personality Similarity

132 M. Decuyper, M. De Bolle, and F. De Fruyt

usually neglected and may be considered asan important alternative viewpoint to explainthe actor’s relationship satisfaction. Finally,different methods are used to assess personal-ity similarity and accuracy, including absolutedifference scores and Pearson correlations.These two methods have a number of draw-backs that could add to the rather incon-sistent picture. As Luo and Klohnen (2005)pointed out, absolute value difference scorescan only range from zero to some positivenumber, indicating that spouses have equal ordifferent levels on a particular trait. Method-ological problems with this absolute differ-ence score include unequal variances in theindividual scores that go into the differencescore and difficulty to interpret results dueto its confound with the individual scores.The use of Pearson correlations to computepersonality similarity or agreement also hasa number of disadvantages. Correlating vari-ables across couples only provides informa-tion about the similarity of two arrays ofscores, although it tells nothing about the sim-ilarity of the members of any particular cou-ple. In addition, Pearson correlations betweenpersonality domains provide no informationon how similar any two partners may be interms of a global assessment of personal-ity. As a consequence, it does not allow theinvestigation of factors related to differencesamong couples in similarity (or dissimilar-ity; Klohnen & Mendelsohn, 1998). Betterindices than the Pearson correlation takeinto account the actual distance between thescores (McCrae, Stone, et al., 1998), usuallyusing the D-indices (measure of dissimilarity),the sum of the squared differences betweencorresponding profile elements (Cronbach &Gleser, 1953). The index of profile agree-ment (Ipa) takes into account both the distancebetween profile elements and the extreme-ness of their mean (McCrae, 1993). These twomethods for computing similarity indices cal-culate similarity indices for each couple forthe overall personality profile and the couplebecomes the basic unit of analysis.

To the best of our knowledge, this studyis the first to examine actual personalitysimilarity, perceptual personality similarity,and perceptual personality accuracy in one

study. Whether these three perspectives areassociated with greater relationship satisfac-tion in a large sample of 191 heterosexualcouples will be investigated. As this study isthe first to examine actual personality similar-ity, perceptual personality similarity, and per-ceptual personality accuracy simultaneously,how these indices intercorrelate will also beexplored. Similarity and accuracy will beoperationalized using the D-indices of Cron-bach and Gleser (1953) and the Ipa (McCrae,1993), two methods that are considered supe-rior to absolute difference scores betweenprofile elements and simple Pearson coeffi-cients across profile elements (McCrae, Stone,et al., 1998). Another strength of this study isthe examination of both partner’s personalitytraits. Each partner provided both a self-reportand personality ratings of the partner on theRevised NEO Personality Inventory (NEO-PI–R; Costa & McCrae, 1992). We controlledfor the potential bias of age, as older peopletend to score somewhat higher on A and Cand lower on N, E, and O than young adults(McCrae et al., 1999). Because couples arepositively matched on age, this variable couldmask assortative effects on personality traitsand should be controlled for in analyses wherea wide range of ages is sampled. In line withmost studies on actual similarity of person-ality traits in couples, we expect that therewill be a considerable range in dissimilar-ity/similarity. We expected that the actual andperceptual similarity in the overall personalityprofile will be positively associated with rela-tionship satisfaction. It is further hypothesizedthat perceptual accuracy will correlate posi-tively with relationship satisfaction, becausedivergent personality views of each other are apotential source of conflict and dissatisfaction.

Method

Sample and procedure

One hundred and ninety-one heterosexualcouples participated. Inclusion criteria werea minimum age of 18 years and a relation-ship duration of at least 6 months. This lastselection criterion was installed to ensure thatthe relationship was serious in nature and thatpartners knew each other well. Couples were

Page 6: Personality Similarity

Personality similarity, accuracy, and satisfaction 133

contacted and recruited by fourth-year under-graduate psychology students and through aparticipants’ recruitment advertisement pub-lished in a newspaper and a psychologymagazine. The students provided the question-naires, an information letter, and the informedconsent forms to the participants, and theyasked participants to return them within 3weeks. Couples recruited by an advertise-ment received and returned the question-naires and informed consent form by mail.The couples were informed about the gen-eral objectives and procedures of the researchproject and detailed oral and written instruc-tions were provided on how to complete thequestionnaires. Written informed consent wasobtained from all participants at the momentof assessment.

Men’s average age was 34.14 years (SD =11.05), ranging from 18.16 to 66.84 years,and women’s average age was 32.27 (SD =11.24), ranging from 18.26 to 64.88 years. Interms of marital status, 42.90% of the cou-ples were married, 32.50% lived together, and24.60% did not cohabit. The mean relation-ship duration at the moment of assessmentwas 127.05 months (SD = 126.38), rangingfrom 6 to 534 months. Approximately 21.10%of the sample had a master’s degree, 29.20%had a bachelor’s degree, and 44.40% finishedhigh school education.

Measures

Demographic questionnaire

Participants completed a demographic ques-tionnaire including items regarding gender,age, education, occupation, length of theirrelationship or their marriage, and the numberof children.

Personality

Personality traits were assessed using theDutch authorized version of the NEO-PI–R(Costa & McCrae, 1992; Hoekstra, Ormel,& De Fruyt, 1996). The NEO-PI–R is a240-item measure of the Big Five person-ality dimensions and allows a differentiatedassessment of each dimension in terms ofsix specific facets per dimension. Participants

provided self-ratings and also partner ratingswere obtained. Both the self-report and infor-mant rating versions have been extensivelyvalidated (Costa & McCrae, 1992). Internalconsistency reliability coefficients in the cur-rent sample, as indicated by Cronbach’s α,were .83, .74, .77, .71, and .84 for self-ratingsof Neuroticism, Extraversion, Openness toExperience, Agreeableness, and Conscien-tiousness, respectively, and .82, .75, .73, .80,and .86 for the partner-rated domains.

Relationship satisfaction

Relationship satisfaction of both partnerswas assessed using a Dutch translation ofthe Relationship Assessment Scale (RAS;Hendrick et al., 1998). Translation was doneusing an independent back-translation proce-dure and differences between the original andback-translated versions were discussed andresolved by joint agreement between trans-lators. The RAS consists of seven items tobe rated on a 5-point Likert scale. Thisinstrument has good psychometric properties(Hendrick et al., 1998; White et al., 2004)and is effective in discriminating couples whostay together from couples who will break up(Hendrick, 1988). Cronbach’s α in this studywas .87.

Computing profile similarity

Similarity between the partners’ personalityprofiles and convergence between self- andpartner ratings were measured using Cronbachand Gleser’s (1953) D-indices and the indexof profile agreement (Ipa; McCrae, 1993).

The first approach to evaluate personal-ity profile similarity is based on Cronbachand Gleser’s (1953) indices D2, D′2, and D′′2.Cronbach and Gleser showed that the indi-vidual profiles can vary in three major ways:elevation (the average level of scores), scatter(the variability of scores), and shape (pattern-ing of scores). D2 is sensitive to differences inelevation, scatter, and shape and quantifies thesquared differences between FFM traits in acouple; D′2 is sensitive to differences in scat-ter and shape and quantifies the squared differ-ences between profiles after each profiles hasbeen centered around its mean; and finally,

Page 7: Personality Similarity

134 M. Decuyper, M. De Bolle, and F. De Fruyt

D′′2 only reflects differences in shape andquantifies the squared differences after eachprofile has been standardized (Cronbach &Gleser, 1953).

A recent comparison of different profileagreement indices (McCrae, 2008) showedthat Ipa is superior to the Pearson r andsuggested that its use for estimating agree-ment at the factor level is justified. Ipa takesinto account both the differences betweenthe ratings (d) and the extremeness of theirmean (M): Ipa = k + 2�M2 − �d2/(10k)1/2,where k is the number of profile elements. Ipa

captures pure profile similarity; the extreme-ness of the mean should not be confused withelevation as defined by Cronbach and Gleser(1953). Advantages of Ipa over an intraclasscorrelation coefficient (double entry; ICCDE)

is that it is possible to calculate Ipa for indi-vidual profile elements and identify particularinstances of disagreement for further inves-tigation or clinical interpretation (McCrae,Stone, et al., 1998). The intercorrelations be-tween the Ipa index and the D-indices arepresented in Table 1. As the D-indices aremeasures of dissimilarity (squared differencesbetween corresponding profile elements) andIpa is a measure of similarity, the negativePearson correlations reported in Table 1 arein line with expectations. The correlationsranged between −.74 and −.51 and are con-sidered large.1

Results

FFM traits and relationship satisfaction

Descriptive statistics for NEO-PI–R domains(self- and partner ratings) and RAS totalscores for both men and women are reportedin Table 2. Gender differences on personal-ity traits and RAS total scores were exam-ined using analysis of variance (ANOVA).If the Levene statistic showed that the vari-ances of both groups were heterogeneous, theWelch correction was applied. For the self-rated personality domains, women scored sig-nificantly higher on Neuroticism (F a = 50.13,

1. Pearson r ≤ |.29| is considered small, r = |.30| to|.49| is considered medium, and r ≥ |.50| is consid-ered large (Cohen, 1988).

p < .001), Agreeableness (F = 24.28, p <

.001), and Openness to Experience (F = 5.79,p < .05). No significant gender differenceswere found for the partner-rated personalitytraits and the RAS total score. Table 3 showsthe Pearson correlations between relationshipcharacteristics, demographic variables, andpersonality traits on the one hand, and theRAS total scores for men and women on theother. In women, relationship satisfaction wasnegatively correlated with age, duration ofthe relationship or marriage, and the numberof children, whereas none of these correla-tions was significant in men. Women’s rela-tionship satisfaction was positively associatedwith the relationship satisfaction reported bymen (r = .51, p < .01).

Regarding women’s relationship satisfac-tion and self-rated personality traits, a nega-tive correlation was found with Neuroticism(r = −.18, p < .05) and positive correlationswith Extraversion (r = .26, p < .05), Agree-ableness (r = .18, p < .05), and Conscien-tiousness (r = .20, p < .01). When women’spersonality traits were rated by their partner,relationship satisfaction in women was nega-tively correlated with Neuroticism (r = −.22,p < .01) and positively with Extraversion (r =.24, p < .01) and Conscientiousness (r = .18,p < .05). Relationship satisfaction in womenwas higher when their partners’ self-ratingswere higher on Extraversion (r = .25, p <

.01) and when women rated their partner aslower on Neuroticism (r = −.17, p < .05)and higher on Extraversion (r = .33, p <

.01), Openness (r = .19, p < .05), Agree-ableness (r = .22, p < .01), and Conscien-tiousness (r = .22, p < .01). This pattern ofpersonality–relationship satisfaction associa-tions was remarkably similar to relationshipsatisfaction in men. Relationship satisfactionin men was positively associated with self-rated Extraversion (r = .29, p < .01), neg-atively with partner-rated Neuroticism (r =−.21, p < .01), and positively with partner-rated Extraversion (r = .24, p < .01), Open-ness (r = .16, p < .05), and Conscientiousness(r = .43, p < .01). Relationship satisfactionin men was also significantly associated withboth self- and partner-rated personality traitsin women. Men were more satisfied when

Page 8: Personality Similarity

Personality similarity, accuracy, and satisfaction 135

Table 1. Convergent validityof Ipa and D-indices

Ipa indices

Actualsimilarity

Perceptualsimilarity:Perspective

women

Perceptualsimilarity:Perspective

men

Perceptualaccuracy:Women

Perceptualaccuracy:

Men

D-indicesActual similarity

D2 −.73∗∗

D′2 −.68∗∗

D′′2 −.63∗∗Perceptual similarity:

Perspective womenD2 −.71∗∗

D′2 −.54∗∗

D′′2 -.51∗∗Perceptual similarity:

Perspective menD2 −.66∗∗

D′2 −.57∗∗

D′′2 −.55∗∗Perceptual accuracy:

WomenD2 −.66∗∗

D′2 −.62∗∗

D′′2 −.65∗∗Perceptual accuracy: Men

D2 −.74∗∗

D′2 −.70∗∗

D′′2 −.61∗∗

Note. Ipa = index of profile agreement (McCrae, 1993); D2, D′2 , D′′2 = D-indices (Cronbach & Gleser, 1953) measuresof profile dissimilarity.∗∗p ≤ .01.

their partner rated themselves as less neurotic(r = −.17, p < .05) and more extraverted(r = .17, p < .05), agreeable (r = .21, p <

.01), and conscientious (r = .22, p < .01).Relationship satisfaction in men was alsohigher when they rated their partner asless neurotic (r = −.18, p < .05) and moreextraverted (r = .16, p < .05).

Finally, ANOVA analyses showed thatwomen without children (n = 107, M = 4.47)reported higher relationship satisfaction com-pared to the other women (n = 83, M =4.19, F = 11.58, p < .01), but there was

no difference for men (F = 1.76, p = .19).Relationship satisfaction was also lower inmarried women (F = 5.38, p < .05) com-pared to unmarried women, with again nodifference in RAS scores (F = .23, p = .63)between married (n = 82) and unmarried men(n = 108).

Actual personality similarity

Profile similarity was first investigated usingthe Cronbach and Gleser (1953) profile sim-ilarity indices D2, D′2 , and D′′2 . Parallel to

Page 9: Personality Similarity

136 M. Decuyper, M. De Bolle, and F. De Fruyt

Table 2. Descriptive statistics for NEO-PI–R domains and RAS scale

Women Men

M (SD) Range M (SD) Range F

NEO-PI–RSelf-rated

Neuroticism 3.01 (.45) 1.83–4.15 2.70 (.39) 1.88–3.56 50.13a∗∗∗Extraversion 3.27 (.40) 2.19–4.04 3.25 (.39) 1.96–4.31 0.17Openness 3.37 (.38) 2.38–4.35 3.27 (.43) 2.19–4.31 5.79∗Agreeableness 3.59 (.31) 2.52–4.50 3.43 (.32) 2.50–4.27 24.28∗∗∗Conscientiousness 3.44 (.42) 2.13–4.48 3.48 (.41) 2.10–4.42 1.11

PersonalityPartner rated

Neuroticism 2.82 (.45) 1.48–4.15 2.79 (.45) 1.77–4.04 0.41Extraversion 3.26 (.41) 1.85–4.10 3.26 (.40) 2.19–4.23 0.04Openness 3.21 (.40) 2.06–4.25 3.18 (.40) 1.90–4.33 0.50Agreeableness 3.57 (.38) 1.90–4.90 3.54 (.43) 1.98–4.56 0.37Conscientiousness 3.60 (.41) 2.23–4.58 3.52 (.47) 2.21–4.42 3.22

RAS 4.35 (.57) 2.43–5.00 4.45 (.50) 2.14–5.00 3.69

Note. NEO-PI–R = Revised NEO Personality Inventory; RAS=Relationship Assessment Scale (Hendrick et al., 1998).aWelch F (asymptotically F distributed); NEO-PI–R scales and RAS total score were based on mean scores.∗p ≤ .05. ∗∗∗p ≤ .001.

the study of Robins, Fraley, Roberts, andTrzesniewski (2001) and De Fruyt and col-leagues (2006), probabilities were estimatedby simulating trait scores on a sample of100,000 individuals with identical levels ofelevation, scatter, and shape in the profileof both partners and examining correspond-ing distributions. Simulated trait scores werebased on means, variances, and covariancesestimated from the real data and test–retestreliability coefficients reported in the studyof McCrae, Yik, and colleagues (1998). Thissimulation produced distributions for D2, D′2 ,and D′′2 , with the 95th percentiles at .41,.35, and .66, respectively. Partners with valuesbeyond these 95th percentile values were con-sidered to have significantly different person-ality profiles. Respectively, 89.00%, 81.20%,and 62.80% of the partners had values beyondthe simulated cutoffs for D2, D′2 , and D′′2 ,suggesting that the large majority of partnershad a different personality profile in terms ofelevation, scatter, and shape. Significant neg-ative correlations between actual personalitydissimilarity as represented by the D-indicesand relationship satisfaction in women wereobserved, with r = −.33 (p ≤ .01), r = −.34

(p ≤ .01), and r = −.29 (p ≤ .01) for D2,D′2 , and D′′2 , respectively. In men, only D′′2

was significantly associated with relationshipsatisfaction, with r = −.18 (p ≤ .05).

Actual similarity was additionally exam-ined using the Ipa. The Ipa was normally dis-tributed (Kolmogorov–Smirnov z = .06, p =.08) and ranged from −4.51 to 3.75, with62.20% having a positive sign. The mean Ipa

across couples was .81 (SD = 1.15). Pear-son correlations between relationship satisfac-tion and personality similarity (measured withIpa) showed that for women, similarity wassignificantly and positively associated withrelationship satisfaction (r = .26, p ≤ .01).In men, however, no significant associationwas observed between relationship satisfac-tion and actual personality similarity as mea-sured with Ipa (r = .11, p = .13).

A hierarchical regression analysis wasconducted to examine how much variance inwomen’s relationship satisfaction was ex-plained by actual personality similarity (Table4). To control for the demographic variablesand the personality traits, age, duration ofthe relationship, being married and havingchildren or not were entered in a first step

Page 10: Personality Similarity

Personality similarity, accuracy, and satisfaction 137

Table 3. Pearson correlations between demo-graphic variables, NEO-PI–R domains, simi-larity indices, and RAS total score

RASwomen

RASmen

Demographic variablesAge −.29∗∗ −.07Duration relationship −.32∗∗ −.12Duration marriage −.28∗ −.16Number of children −.19∗∗ −.04

Personality womenSelf-ratingsNeuroticism −.18∗ −.17∗Extraversion .26∗ .17∗Openness .12 .12Agreeableness .18∗ .21∗∗Conscientiousness .20∗∗ .22∗∗

Partner ratingsNeuroticism −.22∗∗ −.18∗Extraversion .24∗∗ .16∗Openness .09 .08Agreeableness .11 .10Conscientiousness .18∗ .08

Personality menSelf-ratings

Neuroticism −.09 −.12Extraversion .25∗∗ .29∗∗Openness .14 .04Agreeableness .07 −.04Conscientiousness .11 .13

Partner ratingsNeuroticism −.17∗ −.21∗∗Extraversion .33∗∗ .24∗∗Openness .19∗ .16∗Agreeableness .22∗∗ .14Conscientiousness .22∗∗ .43∗∗

Similarity indicesActual personalitysimilarity

Ipa .26∗∗ .11D2 −.33∗∗ −.11D′2 −.34∗∗ −.12D′′2 −.29∗∗ −.18∗

Perceptual personalitysimilarity:Perspective women

Ipa .09 .01D2 −.26∗∗ −.11

Table 3. Continued

RASwomen

RASmen

D′2 −.22∗∗ −.06D′′2 −.16∗ −.12

Perceptual personalitysimilarity:Perspective men

Ipa −.05 .10D2 −.10 −.08D′2 −.06 −.03D′′2 −.09 −.13

Perceptual accuracy:Personality women

Ipa .10 .01D2 −.25∗∗ −.04D′2 −.22∗∗ – .05D′′2 −.20∗∗ −.08

Perceptual accuracy:Personality men

Ipa .05 .08D2 −.26∗∗ −.18∗

D′2 −.22∗∗ −.22∗∗

D′′2 −.20∗∗ −.20∗∗

Note. NEO-PI–R = Revised NEO Personality Inventory;RAS = Relationship Assessment Scale (Hendrick et al.,1998); Ipa = index of profile agreement (McCrae, 1993);

D2, D′2 , D′′2 = D-indices (Cronbach & Gleser, 1953)measures of profile dissimilarity.∗p ≤ .05. ∗∗p ≤ .01.

and self-rated personality traits of the womenand their partners were entered in the sec-ond and third steps, respectively. Enteringthe individual personality traits of womenand men in the second and third stepsremoved their confounding effect and allowedto focus solely on the capacity of actualsimilarity to predict relationship satisfac-tion. The demographic variables explained12% of the variance in relationship vari-ance in women (F = 5.75, p < .001). A sig-nificant effect of women’s self-rated per-sonality traits was observed in the secondstep, explaining an additional 10% of thevariance (F = 4.24, p < .01), and self-ratedpersonality traits of their partners explainedan additional 7% of the variance in the third

Page 11: Personality Similarity

138 M. Decuyper, M. De Bolle, and F. De Fruyt

Table 4. Regression analysis relationship satisfaction, demographic variables, personalitytraits, and Ipa actual personality similarity

Criteria andpredictors R2 �R2 Fchange p

Predictor (squared partcorrelation)

RAS womenStep 1: Age, duration

relationship, beingmarried, and children

.12 .12 5.75 <.001 age (.00), duration relationship (.01),married (.00), children (.01)

Step 2: NEO-PI–Rdomains: Women

.22 .10 4.24 <.01 age (.00), duration relationship (.01),married (.00), children (.00), Nw

(.02), Ew (.00), Ow (.00), Aw (.03),Cw (.01)

Step 3: NEO-PI–Rdomains: Men

.28 .07 2.90 <.05 age (.00), duration relationship (.01),married (.00), children (.01), Nw

(.01), Ew (.00), Ow (.00), Aw (.03),Cw (.01), Nm (.00), Em (.00), Om

(.01), Am (.03), Cm (.01)Step 4: Ipa .32 .04 8.80 <.01 age (.00), duration relationship (.02),

married (.00), children (.01), Nw

(.01), Ew (.01), Ow (.00), Aw (.02),Cw (.01), Nm (.00), Em (.00), Om

(.00), Am (.02), Cm (.02), Ipa (.04)

Note. Ipa = index of profile actual personality similarity of men and women (being alike); RAS = RelationshipAssessment Scale (Hendrick et al., 1998); NEO-PI–R = NEO Personality Inventory Revised (Hoekstra et al., 1996),N = Neuroticism; E = Extraversion; O = Openness; A = Agreeableness; C = Conscientiousness; traitw = personalitytrait women; traitm = personality trait men.

step (F = 2.90, p < .05). Actual profile sim-ilarity Ipa computed between self-rated per-sonality traits of the partners explained anadditional 4% of the variance (F = 8.80, p <

.01). The squared part correlations indicatedthat women’s self-rated Neuroticism, Agree-ableness, and Conscientiousness; men’s self-rated Agreeableness and Conscientiousness;and actual personality similarity Ipa explainedmost variance in women’s relationship sat-isfaction. This hierarchical regression analy-sis is an extremely conservative test for theunique contribution of actual profile similar-ity to satisfaction and shows that actual sim-ilarity makes an independent contribution torelationship satisfaction beyond demographicvariables and both men’s and women’s FFMself-ratings. As actual personality similaritywas unrelated to relationship satisfaction inmen, this regression analysis was not con-ducted for men.

Perceptual personality similarity

Perceptual personality similarity was firstinvestigated using the Cronbach and Gleser(1953) profile similarity indices D2, D′2 , andD′′2 . For women, the simulation procedure(see simulation procedure under actual per-sonality similarity) produced distributions ofD2, D′2 , and D′′2 , with the 95th percentilesdefined at .46, .39, and .86, respectively. Themen’s personality profiles rated by womenwith values beyond these 95th percentile val-ues were considered significantly differentthan the self-ratings of the women’s personal-ity. Respectively, 71.20%, 67%, and 38.20%of the men’s personality profiles rated by theirpartner had values beyond these simulatedcutoffs for D2, D′2 , and D′′2 . Approximately62% of the women rated the personalityprofile of their partner similar to their ownpersonality profile in terms of shape and28.90% of the men’s personality profiles rated

Page 12: Personality Similarity

Personality similarity, accuracy, and satisfaction 139

by women were identical to the women’sself-rated personality profiles in terms of ele-vation, scatter, and shape. For men, D2, D′2 ,and D′′2 had 95th percentiles at .41, .35,and .66, respectively. Women’s personalityprofiles rated by their partner with valuesbeyond these 95th percentile values were con-sidered significantly different than the men’sself-rated personality profiles. Respectively70.50%, 64.70%, and 41.60% of the women’spersonality profiles rated by men had val-ues beyond these simulated cutoffs for D2,D′2 , and D′′2 . Approximately 58% of the menrated their partner’s personality profile simi-lar to their own personality profile in termsof shape and approximately 30% of the menrated their partner’s personality profile iden-tical to their own personality profile in termsof elevation, scatter, and shape. In line withour hypotheses, the D-indices representingperceptual dissimilarity from the perspectiveof the women were significantly associatedwith women’s relationship satisfaction, withr = −.26 (p ≤ .01), r = −.22 (p ≤ .01), andr = −.16 (p ≤ .05) for D2, D′2 , and D′′2 ,respectively. However, D-indices representingperceptual personality dissimilarity from themen’s perspective were unrelated relationshipsatisfaction in men.

The Ipa representing perceptual personalitysimilarity from the women’s perspective wasnormally distributed (Kolmogorov–Smirnovz = .07, p = .06) after exclusion of sevenoutliers. We controlled for outliers using case-wise diagnostics in SPSS and examined influ-ential cases (using Cook’s distance). The Ipa

representing perceptual personality similar-ity computed from the women’s perspective,comparing female self-rated personality withtheir ratings of the personality of their part-ner, ranged from −1.40 to 3.06, with 82.90%having a positive sign and a mean Ipa acrosscouples of .89 (SD = 93). From the perspec-tive of men, Ipa coefficients representing per-ceptual personality similarity had a normaldistribution (Kolmogorov–Smirnov z = .05,p = .20) and ranged from −4.51 to 3.75, with84.00% having a positive sign. The mean Ipa

across men was .81 (SD = 1.15). Contrary toour hypotheses, perceptual personality simi-larity did not correlate significantly with the

actor’s relationship satisfaction, either for men(r = .10) or for women (r = .09).

Perceptual accuracy

Profile similarity between self and partner rat-ings was first investigated using the D-indices.For women, the simulation produced distri-butions of D2, D′2 , and D′′2 , with the 95thpercentiles defined at .46, .39, and .86, respec-tively. Women’s personality profiles rated bytheir men with values beyond these 95thpercentile values were considered signifi-cantly different from the personality self-descriptions. Respectively 61.10%, 57.40%,and 38.90% of the partner ratings had val-ues beyond the simulated cutoffs for D2, D′2 ,and D′′2 . Approximately 61% of the part-ner ratings had the same personality profilein terms of shape as the self-rated person-ality, and 38.80% agreed on the profiles interms of elevation, scatter, and shape. In men,D2, D′2 , and D′′2 had 95th percentiles at .41,.35, and .66, respectively. Men’s personalityprofiles rated by their partners with valuesbeyond these 95th percentile values were con-sidered significantly different from the self-described personality profiles. Respectively,71.70%, 61.80%, and 44% of the partner-ratedpersonality traits had values beyond these cut-offs for D2, D′2 , and D′′2 . Fifty-six percent ofthe men had the same self-rated and partner-rated personality profile in terms of shape, and28.30% had identical profiles in terms of ele-vation, scatter, and shape. In line with ourhypotheses, D-indices representing perceptualaccuracy on women’s personality were sig-nificantly correlated with relationship satisfac-tion in women, with r = −.26, r = −.22, andr = −20 for D2, D′2 , and D′′2 , respectively(p ≤ .01), but were unrelated to relationshipsatisfaction in men. Perceptual accuracy onmen’s personality was significantly associatedwith relationship satisfaction in men, withr = −.18 (p ≤ .05), r = −.22 (p ≤ .01), andr = −.20 (p ≤ .01) for D2, D′2 , and D′′2 ,respectively, and also significantly correlatedwith relationship satisfaction in women, withr = −.26, r = −.22, and r = −.20 for D2,D′2 , and D′′2 , respectively (p ≤ .01).

Page 13: Personality Similarity

140 M. Decuyper, M. De Bolle, and F. De Fruyt

The Ipa (McCrae, 1993) representing per-ceptual accuracy on the women’s personalityratings was normally distributed (Kolmo-gorov–Smirnov z = .05, p = .20) after exclu-sion of seven outliers. We controlled foroutliers using casewise diagnostics in SPSSand examined influential cases (using Cook’sdistance). The Ipa coefficients ranged from−1.51 to 3.08, with 83.50% having a pos-itive sign and the mean Ipa was .93 (SD =.88). Fourteen couples were identified as out-liers concerning perceptual accuracy on themen’s personality traits and were removedfrom the analyses. The Ipa coefficients rep-resenting the perceptual accuracy on themen’s personality were normally distributed(Kolmogorov–Smirnov z = .05, p = .20) andranged from −1.82 to 3.04, with 88.40% hav-ing a positive sign. The mean Ipa was .95(SD = .83). The Ipa index representing per-ceptual accuracy on women’s personality wasunrelated to relationship satisfaction reportedby women (r = .10) and men (with r = .01).Moreover, the Ipa representing perceptualaccuracy on men’s personality traits was alsounrelated to relationship satisfaction in men(with r = .08) and in women (with r = .05).

Actual similarity, perceptual similarity,and perceptual accuracy

Pearson correlations between indices of actualsimilarity, perceptual similarity, and percep-tual accuracy were computed to explore howthese three perspectives intercorrelate withanother. Using the Ipa indices, results showedthat the actual similarity was positively asso-ciated with perceptual similarity from thewomen’s and men’s perspective with r = .38(p < .001) and r = .32 (p < .001), respec-tively. In addition, actual personality simi-larity was significantly associated with per-ceptual accuracy on women’s (r = .43, p <

.001) and men’s (r = .30, p < .001) per-sonality traits. Similar results were obtainedusing the D-indices (with r ranging between.36 and .63, all significant at p < .001).These results indicated that couples withhigher actual personality similarity also havehigher perceptual similarity and perceptualaccuracy.

Discussion

This study examined the associations betweenactual and perceptual personality similarity,perceptual accuracy, and relationship satisfac-tion in a sample of 191 heterosexual couplesusing a comprehensive trait measure and dif-ferent methods to represent personality profilesimilarity and agreement.

Individual’s traits and satisfaction

In line with previous studies, both self- andpartner-rated personality traits were signifi-cantly associated with relationship satisfactionof men and women. Thus, whether individualsare more or less satisfied in their relation-ship is, among other factors, a function ofhow they describe themselves and how theyperceive each other. This observation is inline with previous research findings showingthat each partner’s personality traits are deter-minants of relationship happiness, providingsupport for the theoretical perspective empha-sizing consistent and enduring influences ofpersonality traits on relationship behaviorsand experiences (Kelly & Conley, 1987; Luoet al., 2008).

Actual and perceptual personality similarityand satisfaction

Research on assortative mating has focusedprimarily on the questions of mate selec-tion and relationship success. Key questionsin this debate were whether individuals tendto select similar or opposite others to betheir partners and whether spouse similarityis important for relationship success and sat-isfaction. Also in this study, inspection ofthe D-indices showed that most partners hada different personality profile in terms ofelevation, scatter, and shape, analyzing self-descriptions of personality. The Ipa showedthat there was a considerable couple-to-couplevariation in self-described personality sim-ilarity. Actual personality similarity, repre-sented using the D-indices and the Ipa, wassignificantly associated with relationship sat-isfaction, but only in women, confirmingour hypothesis for this subgroup. In women,actual personality similarity was associated

Page 14: Personality Similarity

Personality similarity, accuracy, and satisfaction 141

with relationship satisfaction, even after con-trolling for women’s and men’s personalitycharacteristics. The result of this conservativetest of the effect of actual personality similar-ity, controlling for self-described personalitydimensions of both partners, counters researchfindings suggesting that couple personalitysimilarity makes only a minor contributionto the explanation of relationship satisfac-tion. This study shows that the overall actualpersonality similarity is substantially relatedto women’s relationship satisfaction and thiscannot be explained by age and the durationof the relationship.

The evidence for the association betweenactual personality similarity and relationshipsatisfaction is consistent with previous empir-ical works (Caspi & Herbener, 1990; Gaunt,2006; Gonzaga et al., 2007; Luo & Klohnen,2005; Robins et al., 2000). This associationwas restricted in this study to women, under-scoring women’s position as the “barom-eter” of relationships (Floyd & Markman,1983); that is, women are more attunedto and focus more on the internal dynam-ics in the relationship (Acitelli, 1992). Thisfinding further corresponds with the gen-eral idea that interactions for women tend tohave stronger implications for the relation-ship quality and satisfaction than men (Luo& Klohnen, 2005).

Contrary to our hypotheses, perceptual per-sonality similarity in men—represented byD-indices and Ipa index—was not signifi-cantly associated with relationship satisfac-tion, despite considerable agreement betweenpartner ratings and self-descriptions. Thismeans that in men, the tendency to ratetheir partner as similar to themselves wasunrelated to their relationship satisfaction.Perceptual personality similarity from thewomen’s perspective (i.e., the woman’s per-ception of the partner’s personality relativeto the woman’s self-description) was signif-icantly associated with relationship satisfac-tion in women, but only using the D-indices.This finding was in line with our hypoth-esis and could indicate that women, whorate their partner as more similar to them-selves, may feel closer to their partner andmay have a better relationship quality. Just

as perceptual similarity leads to satisfaction,it is also likely that satisfaction could leadto perceptual similarity (Montoya et al., 2008;Morry, 2005, 2007). Individuals could changetheir beliefs of similarity toward a lovingother because of cognitive biases (Montoyaet al., 2008). Satisfied women could assimi-late their relationship partner into their ownself-concept, leading to higher perceived sim-ilarity. However, this finding should be inter-preted with caution because this associationbetween relationship satisfaction and percep-tual personality similarity was not replicatedusing the Ipa.

These different findings resulting from theIpa and D-indices underscore the importanceof the method used to compute profile sim-ilarity. Indices solely based on D2 make theassumption that the distance between two pro-file elements is equally important at all levelsof the trait. According to McCrae (1993), dis-agreements are less serious when they occurin the extreme ranges of the traits than whenthey occur near the middle. For instance,T scores of 60 and 80 both indicate high lev-els of a trait, even though they differ 2 SD.By contrast, T scores of 40 and 60 gener-ate the same D2, but they have qualitativedifferent interpretations. T score of 40 corre-sponds with a low standing on a trait, whereasT score of 60 indicates a high position ona trait (McCrae, Stone, et al., 1998). More-over, close agreement between two scores caneasily happen by chance if both are near themiddle where most scores are distributed, butit would be very unusual for two very high orlow scores to agree by chance. In contrast withthe D-indices, the Ipa also takes into accountthe extremeness of the mean of the profile ele-ments. The different results obtained in thisstudy using the Ipa or the D-indices indicatethat the significant association between per-ceptual similarity and relationship satisfactionin women found using D-indices appears onlyat the extreme ranges of the traits included inthe personality profile.

Perceptual accuracy and satisfaction

Relying on the D-indices, this study showedthat for both men and women, 56% (men) and

Page 15: Personality Similarity

142 M. Decuyper, M. De Bolle, and F. De Fruyt

61% (women) of the descriptions provided bythe partner had the same personality profileshape as the self-descriptions and 28% (men)to 38% (women) agreed on elevation, scat-ter, and shape, suggesting a somewhat higheragreement on the personality for women. Inaddition, perceptual accuracy as measuredwith the Ipa showed large variability acrosscouples. In line with previous research con-ducted by Neff and Karney (2005) and Let-zring and Noftle (2010), perceptual accuracywas significantly associated with relationshipsatisfaction in both men and women using theD-indices. Concordant views on the women’spersonality profile were associated with rela-tionship satisfaction in women and agreementbetween self- and partner ratings on the men’spersonality profile was linked with relation-ship satisfaction in men and women. Thesefindings indicate that perceptual personalitysimilarity in couples contribute to relationshipsatisfaction and the current results suggestthat it may be that having accurate knowl-edge of a romantic partner is important fora loving and committed relationship. Whenperceptual accuracy was represented using theIpa, the results, however, suggest that percep-tual accuracy on broad personality traits haslittle or no association with relationship sat-isfaction in men and women. These resultsare in line with the study of McCrae, Stone,and colleagues (1998) in which profile sim-ilarity was as well operationalized using theIpa. The different results obtained using the Ipa

or the D-indices indicate that the significantassociation between perceptual accuracy andrelationship satisfaction is only found whendifferences between two profile elements areconsidered equally important at all levels ofthe trait.

Actual similarity, perceptual similarity,and perceptual accuracy

The results of this study indicated that cou-ples with higher actual personality similarityalso have higher perceptual similarity and per-ceptual accuracy for both men and women.These significant associations were obtainedusing both Ipa and D-indices to compute pro-file similarity.

Limitations

This study also has a number of limitations.First, we focused on the associations betweenrelationship satisfaction and personality sim-ilarity in couples and did not consider simi-larity in other domains, such as attachment,values, interests, or intelligence, constrainingthe potential of this study to explain satis-faction variance. Second, individuals in thisstudy were fairly well educated, restricting thegeneralizability of the findings to the broaderpopulation. Moreover, part of the current sam-ple was recruited by fourth-year undergradu-ate psychology students, and this could haveresulted in a somewhat biased sample. Par-ticipating couples further reported on aver-age higher relationship satisfaction, inducingpotential range restriction impacting upon thesize of the observed associations. Fifth, thisstudy has a cross-sectional design with asimultaneous assessment of personality traitsand relationship satisfaction, preventing ananalysis of the causal relationships betweenpersonality and relationship outcomes. Giventhe cross-sectional design of the study, it can-not be excluded that satisfied couples becomeincreasingly similar with time or that highersatisfaction predicts greater perceived similar-ity for traits (Morry, 2007). Similar percep-tions could be made in a strategic fashion andprovide a variety of benefits such as feelingunderstood and validated and hence serve theromantic relationship (Morry, 2005). How-ever, there is research (Glicksohn & Golan,2001; Watson et al., 2004) showing that initialassortment and not convergence is primarilyresponsible for couple similarity.

Conclusions and recommendations

In sum, the hypotheses that actual personalitysimilarity, perceptual similarity, and percep-tual personality accuracy would be positivelyassociated with relationship quality were par-tially supported in this study. The findingsfor actual and perceptual personality similar-ity, however, were restricted to women. Theobservation that actual personality similar-ity and perceptual similarity on the women’spersonality were only significantly associ-ated with women’s satisfaction has not been

Page 16: Personality Similarity

Personality similarity, accuracy, and satisfaction 143

reported before. In addition, the hypothesesfor perceptual personality similarity and per-ceptual accuracy were only supported whenD-indices were used to operationalize pro-file similarity or agreement. The differentresults obtained with the Ipa or the D-indicesindicated that the associations between per-ceptual similarity, perceptual accuracy, andrelationship satisfaction only appeared whendifferences between two profile elements areconsidered equally important at all levels ofthe trait. In women, the most robust associ-ation was found between actual personalitysimilarity and relationship satisfaction, indi-cating that self-descriptions seem importantto evaluate couple similarity. In men, the find-ings suggested that perceptual accuracy on themen’s personality—rather than actual or per-ceptual personality similarity—is an impor-tant correlate of relationship satisfaction.

Future studies should replicate these find-ings before one can generalize and deriveassessment and therapeutic recommendations.This study has demonstrated that it is neces-sary to combine the different couple-centeredapproaches to fully address the connectionsbetween personality and relationship qual-ity at the dyadic level. Longitudinal designswill further help to disentangle the direc-tion of effects. Measurement of personalitytraits early in the relationship, but also atlater stages, complemented with an assess-ment of satisfaction at multiple occasionsseveral years later, will ultimately enable toidentify relationship satisfaction trajectories.

References

Acitelli, L. K. (1992). Gender differences in relation-ship awareness and marital satisfaction among youngmarried-couples. Personality and Social PsychologyBulletin, 18, 102–110.

Barelds, D. P. H. (2005). Self and partner personality inintimate relationships. European Journal of Personal-ity, 19, 501–518.

Barelds, D. P. H., & Dijkstra, P. (2009). Positive illusionsabout a partner’s physical attractiveness and relation-ship quality. Personal Relationships, 16, 263–283.

Block, J. (1971). Lives through time. Berkeley, CA:Brancroft.

Caspi, A., & Herbener, E. S. (1990). Continuity andchange: Assortative marriage and the consistency ofpersonality in adulthood. Journal of Personality andSocial Psychology, 58, 250–258.

Cohen, J. (1988). Statistical power analysis for the behav-ioral sciences. Hillsdale, NJ: Erlbaum.

Costa, P. T., & McCrae, R. R. (1992). Professional man-ual: Revised NEO Personality Inventory (NEO-PI–R)and NEO Five-Factor-Inventory (NEO-FFI). Odessa,FL: Psychological Assessment Resources.

Cronbach, L. J., & Gleser, G. C. (1953). Assessing sim-ilarity between profiles. Psychological Bulletin, 50,456–473.

De Fruyt, F., Bartels, M., Van Leeuwen, K. G., DeClercq, B., Decuyper, M., & Mervielde, I. (2006).Five types of personality continuity in childhood andadolescence. Journal of Personality and Social Psy-chology, 91, 538–552.

Digman, J. M. (1990). Personality structure: Emergenceof the 5-factor model. Annual Review of Psychology,41, 417–440.

Eysenck, S. B. G., Eysenck, H. J., & Barrett, P. (1985).A revised version of the psychoticism scale. Person-ality and Individual Differences, 6, 21–29.

Floyd, F. J., & Markman, H. J. (1983). Observationalbiases in spouse observation: Toward a cognitivebehavioral-model of marriage. Journal of Consultingand Clinical Psychology, 51, 450–457.

Foltz, C., Morse, J. Q., Calvo, N., & Barber, J. P. (1997).Self and observer ratings on the NEO-FFI in couples:Initial evidence of the psychometric properties of anobserver form. Assessment, 4, 287–295.

Funder, D. C. (1980). On seeing ourselves as others seeus: Self-other agreement and discrepancy in person-ality ratings. Journal of Personality, 48, 473–493.

Funder, D. C., Kolar, D. C., & Blackman, M. C. (1995).Agreement among judges of personality: Interper-sonal-relations, similarity, and acquaintanceship.Journal of Personality and Social Psychology, 69,656–672.

Gattis, K. S., Berns, S., Simpson, L. E., & Christensen,A. (2004). Birds of a feather or strange birds? Tiesamong personality dimensions, similarity, and maritalquality. Journal of Family Psychology, 18, 564–574.

Gaunt, R. (2006). Couple similarity and marital satisfac-tion: Are similar spouses happier? Journal of Person-ality, 74, 1401–1420.

Gill, M. J., & Swann, W. B. (2004). On what it meansto know someone: A matter of pragmatics. Journal ofPersonality and Social Psychology, 86, 405–418.

Glicksohn, J., & Golan, H. (2001). Personality, cognitivestyle and assortative mating. Personality and Individ-ual Differences, 30, 1199–1209.

Gonzaga, G. C., Campos, B., & Bradbury, T. (2007).Similarity, convergence, and relationship satisfactionin dating and married couples. Journal of Personalityand Social Psychology, 93, 34–48.

Hendrick, S. S. (1988). A generic measure of relationshipsatisfaction. Journal of Marriage and the Family, 50,93–98.

Hendrick, S. S., Dicke, A., & Hendrick, C. (1998). Therelationship assessment scale. Journal of Social andPersonal Relationships, 15, 137–142.

Hoekstra, H. A., Ormel, J., & De Fruyt, F. (1996).NEO persoonlijkheidsvragenlijsten NEO-PI–R en

Page 17: Personality Similarity

144 M. Decuyper, M. De Bolle, and F. De Fruyt

NEO-FFI. Handleiding. Lisse, The Netherlands: Swets& Zeitlinger.

Hofstee, W. K. B. (1994). Who should own the definitionof personality. European Journal of Personality, 8,149–162.

John, O. P. (1990). The “Big Five” factor taxonomy:Dimensions of personality in the natural language andquestionnaires. In L. A. Pervin (Ed.), Handbook ofpersonality theory and research (pp. 66–100). NewYork: Guilford.

Karney, B. R., & Bradbury, T. N. (1995). The longitudi-nal course of marital quality and stability: A review oftheory, method, and research. Psychological Bulletin,118, 3–34.

Karney, B. R., & Bradbury, T. N. (1997). Neuroticism,marital interaction, and the trajectory of marital satis-faction. Journal of Personality and Social Psychology,72, 1075–1092.

Kelly, E. L., & Conley, J. J. (1987). Personality andcompatibility: A prospective analysis of marital sta-bility and marital satisfaction. Journal of Personalityand Social Psychology, 52, 27–40.

Kenny, D. A. (1996). Models of non-independence indyadic research. Journal of Social and Personal Rela-tionships, 13, 279–294.

Klohnen, E. C., & Mendelsohn, G. A. (1998). Partnerselection for personality characteristics: A couple-centered approach. Personality and Social PsychologyBulletin, 24, 268–278.

Letzring, T. D., & Noftle, E. E. (2010). Predicting rela-tionship quality from self-verification of broad per-sonality traits among romantic couples. Journal ofResearch in Personality, 44, 353–362.

Luo, S. H., Chen, H., Yue, G. A., Zhang, G. J., Zhaoyang,R. X., & Xu, D. (2008). Predicting marital satisfac-tion from self, partner, and couple characteristics:Is it me, you, or us? Journal of Personality, 76,1231–1265.

Luo, S. H., & Klohnen, E. C. (2005). Assortative matingand marital quality in newlyweds: A couple-centeredapproach. Journal of Personality and Social Psychol-ogy, 88, 304–326.

McCrae, R. R. (1993). Agreement of personality profilesacross observers. Multivariate Behavioral Research,28, 25–40.

McCrae, R. R. (1994). The counterpoint of personalityassessment: Self-reports and observer ratings. Assess-ment, 1, 159–172.

McCrae, R. R. (2008). A note on some measures ofprofile agreement. Journal of Personality Assessment,90, 105–109.

McCrae, R. R., Costa, P. T., de Lima, M. P., Simoes, A.,Ostendorf, F., Angleitner, A., Piedmont, R. L. (1999).Age differences in personality across the adult lifespan: Parallels in five cultures. Developmental Psy-chology, 35, 466–477.

McCrae, R. R., Martin, T. A., Hrebickova, M., Urbanek,T., Boomsma, D. I., Willemsen, G., Costa, P. T.(2008). Personality trait similarity between spouses infour cultures. Journal of Personality, 76, 1137–1163.

McCrae, R. R., Stone, S. V., Fagan, P. J., & Costa, P. T.(1998). Identifying causes of disagreement betweenself-reports and spouse ratings of personality. Journalof Personality, 66, 285–313.

McCrae, R. R., Yik, M. S. M., Trapnell, P. D., Bond,M. H., & Paulhus, D. L. (1998). Interpreting person-ality profiles across cultures: Bilingual, acculturation,and peer rating studies of Chinese undergraduates.Journal of Personality and Social Psychology, 74,1041–1055.

Montoya, R. M., Horton, R. S., & Kirchner, J. (2008). Isactual similarity necessary for attraction? A meta-analysis of actual and perceived similarity. Journalof Social and Personal Relationships, 25, 889–922.

Morry, M. M. (2005). Relationship satisfaction as a pre-dictor of similarity ratings: A test of the attraction-similarity hypothesis. Journal of Social and PersonalRelationships, 22, 561–584.

Morry, M. M. (2007). The attraction-similarity hypothesisamong cross-sex friends: Relationship satisfaction,perceived similarities, and self-serving perceptions.Journal of Social and Personal Relationships, 24,117–138.

Murray, S. L., & Holmes, J. G. (1997). A leap of faith?Positive illusions in romantic relationships. Personal-ity and Social Psychology Bulletin, 23, 586–604.

Neff, L. A., & Karney, B. R. (2005). To know you is tolove you: The implications of global adoration andspecific accuracy for marital relationships. Journal ofPersonality and Social Psychology, 88, 480–497.

Robins, R. W., Caspi, A., & Moffitt, T. E. (2000). Twopersonalities, one relationship: Both partners’ person-ality traits shape the quality of their relationship.Journal of Personality and Social Psychology, 79,251–259.

Robins, R. W., Caspi, A., & Moffitt, T. E. (2002). It’snot just who you’re with, it’s who you are: Personalityand relationship experiences across multiple relation-ships. Journal of Personality, 70, 925–964.

Robins, R. W., Fraley, R. C., Roberts, B. W., &Trzesniewski, K. H. (2001). A longitudinal study ofpersonality change in young adulthood. Journal ofPersonality, 69, 617–640.

Russell, R. J. H., & Wells, P. A. (1991). Personality sim-ilarity and quality of marriage. Personality and Indi-vidual Differences, 12, 407–412.

Shiota, M. N., & Levenson, R. W. (2007). Birds of afeather don’t always fly farthest: Similarity in BigFive personality predicts more negative marital satis-faction trajectories in long-term marriages. Psychol-ogy and Aging, 22, 666–675.

Smith, L., Ciarrochi, J., & Heaven, P. C. L. (2008). Thestability and change of trait emotional intelligence,conflict communication patterns, and relationship sat-isfaction: A one-year longitudinal study. Personalityand Individual Differences, 45, 738–743.

Swann, W. B., Pelham, B. W., & Krull, D. S. (1989).Agreeable fancy or disagreeable truth: Reconcil-ing self-enhancement and self-verification. Journal ofPersonality and Social Psychology, 57, 782–791.

Page 18: Personality Similarity

Personality similarity, accuracy, and satisfaction 145

Watson, D., Klohnen, E. C., Casillas, A., Simms, E. N.,Haig, J., & Berry, D. S. (2004). Match makers anddeal breakers: Analyses of assortative mating in new-lywed couples. Journal of Personality,72,1029–1068.

Weller, J., & Watson, D. (2009). Friend or foe? Differ-ential use of the self-based heuristic as a function ofrelationship satisfaction. Journal of Personality, 77,731–760.

White, J. K., Hendrick, S. S., & Hendrick, C. (2004).Big Five personality variables and relationship con-structs. Personality and Individual Differences, 37,1519–1530.

Zuckerman, M. (1979). Sensation seeking: Beyond theoptimal level of arousal. Hillsdale, NJ: Erlbaum.

Page 19: Personality Similarity

Copyright of Personal Relationships is the property of Wiley-Blackwell and its content may not be copied or

emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.

However, users may print, download, or email articles for individual use.