Embed Size (px)
Citation preview
Journal of Research in Personality 47 (2013) 369–375
Contents lists available at SciVerse ScienceDirect
Journal of Research in Personality
journal homepage: www.elsevier .com/ locate/ j rp
Personality similarity and life satisfaction in couples
Katrin Furler ⇑, Veronica Gomez, Alexander Grob ⇑Department of Psychology, University of Basel, Switzerland
a r t i c l e i n f o a b s t r a c t
Article history:Available online 15 March 2013
Keywords:Personality similarityLife satisfactionCouplesDifferentiated approachActor effectsPartner effects
0092-6566/$ - see front matter � 2013 Elsevier Inc. Ahttp://dx.doi.org/10.1016/j.jrp.2013.03.002
⇑ Corresponding authors. Address: Department of PsMissionsstrasse 62, 4055 Basel, Switzerland.
E-mail addresses: [email protected] (K. Furle(A. Grob).
The present study examined the association between personality similarity and life satisfaction in a large,nationally representative sample of 1608 romantic couples. Similarity effects were computed for the BigFive personality traits as well as for personality profiles with global and differentiated indices of similar-ity. Results showed substantial actor and partner effects, indicating that both partners’ personality traitswere related to both partners’ life satisfaction. Personality similarity, however, was not related to eitherpartner’s life satisfaction. We emphasize the importance of thoroughly controlling for each partner’s per-sonality and for applying appropriate analytical methods for dyadic data when assessing the effect of per-sonality similarity in couples.
� 2013 Elsevier Inc. All rights reserved.
1. Introduction
Personal relationships in general, and romantic relationships inparticular, are essential for people’s well-being. Romantic relation-ships represent a unique social unit in which partners spend a con-siderable amount of time together and share closeness andintimacy with each other (Argyle, 1999). A recent longitudinalstudy on development of life satisfaction in couples revealed thatboth members of a couple influence each other and mutually affectthe other partner’s well-being (Hoppmann, Gerstorf, Willis, &Schaie, 2011). In this regard, a dyadic approach that focuses oncouples as the unit of analysis is of crucial interest for personalitypsychology. Correspondingly, within- and between-person as wellas similarity effects of personality on well-being have receivedconsiderable attention in recent research and both one’s own aswell as the partner’s personality characteristics have been foundto be important factors in intimate relationships (e.g., Barelds,2005; Dyrenforth, Kashy, Donnellan, & Lucas, 2010; Gonzaga, Cam-pos, & Bradbury, 2007; Luo & Klohnen, 2005). Evidence so farspeaks for consistent effects of one’s own personality on well-being(actor effects), such that being extraverted, agreeable, conscien-tious, emotionally stable and open to experience is positively re-lated to a person’s well-being (e.g., Steel, Schmidt, & Shultz,2008). But it is not only one’s own personality that affects well-being, it is also the partner’s personality (partner effects). Beingin an intimate relationship with someone who is extraverted,agreeable, conscientious, and emotionally stable is associated with
ll rights reserved.
ychology, University of Basel,
higher well-being as well (Barelds, 2005; Dyrenforth et al., 2010;Headey, Muffels, & Wagner, 2010; Robins, Caspi, & Moffitt, 2000).
However, as far as personality similarity in romantic couples isconcerned, the everlasting question about who is a person’s perfectmatch has not yet been answered satisfactorily. Is having a partnerwhose personality is similar to one’s own associated with higherlevels of satisfaction? Or do we prefer someone rather dissimilar?The present study tries to shed light on this issue and examineshow personality similarity in couples relates to each partner’s lifesatisfaction.
Research to date leans more towards ‘‘like attracts like’’ ratherthan ‘‘opposites attract’’ and assumes that similarity, as opposedto complementarity (Antill, 1983), is an important factor in roman-tic relationships. Studies on mate selection for instance support thenotion that similarity between partners is essential in forming inti-mate relationships. People tend to choose partners who are similarto themselves on a number of variables, including age, education,values, physical attractiveness, and intelligence (Epstein & Gutt-man, 1985; Vandenberg, 1972). Similarity is theorized to be bene-ficial for intimate relationships by coordinating partners’ thoughtsand behaviors, increasing understanding of each other’s intentionsand motivations, and reinforcing their appraisals, leading to rela-tionship satisfaction and longevity (Anderson, Keltner, & John,2003). Nevertheless, spousal similarity in personality traits isgenerally quite low (Watson et al., 2004) although people tendto prefer partners with similar personalities (Botwin, Buss, &Shackelford, 1997).
The existing body of research on the association between per-sonality similarity and well-being in couples does not provide aclear picture. Some studies show positive effects for relationshipor life satisfaction respectively, such that the more similar couples’personalities, the higher each partner’s satisfaction in the
370 K. Furler et al. / Journal of Research in Personality 47 (2013) 369–375
respective domain (Arrindell & Luteijn, 2000; Gaunt, 2006; Gon-zaga, Carter, & Buckwalter, 2010; Gonzaga et al., 2007; Luo & Kloh-nen, 2005; Robins et al., 2000). However, a recent study usingnationally representative panel data from Great Britain, Australia,and Germany revealed no or only small associations between per-sonality similarity and relationship or life satisfaction and neitherof these small effects was consistent across the three samples(Dyrenforth et al., 2010). Thus, even with big sample sizes havingthe power to detect very small effects, there was little evidencethat personality similarity predicted relationship or life satisfactionin those three large samples. Furthermore, in a representative sam-ple of Dutch couples, personality similarity did not affect maritalquality (Barelds, 2005) and in a sample of distressed and treat-ment-seeking couples, similarity on any of the Big Five traits wasnot a predictor of marital satisfaction (Gattis, Berns, Simpson, &Christensen, 2004).
Given the inconsistent empirical evidence, research is needed inorder to deepen our understanding of how personality similarity inromantic relationships relates to each partner’s well-being. For thispurpose, three important factors that have been neglected in sev-eral previous studies have to be considered. First, when assessingthe effect of personality similarity in couples, it is essential to con-trol for each partner’s individual level of personality to get the un-ique similarity effect beyond each partner’s individual contribution(Griffin, Murray, & Gonzalez, 1999; Kenny & Acitelli, 1994). Notcontrolling for initial levels of both partners’ personality leads toan overestimation of the association between similarity and well-being. Some of the studies reporting significant similarity effectsdid not or did not thoroughly control for individual levels of part-ners’ personality (Gonzaga et al., 2007; Robins et al., 2000). Whenmain effects were included in the analysis, however, similarity wasno longer a unique predictor of well-being (Barelds, 2005; Dyren-forth et al., 2010; Gattis et al., 2004).
Second, to gain generalizable results and a more accurate senseof the association between personality similarity and well-being, itis necessary to examine a large representative sample of couples.Several studies that found similarity effects on well-being analyzedrelatively small (Arrindell & Luteijn, 2000; Gonzaga et al., 2007) orvery specific and thus possibly biased samples such as newlyweds(Luo & Klohnen, 2005), cohort study members (Robins et al., 2000)or couples who met via online dating platforms (Gonzaga et al.,2010). However, the few studies examining large representativesamples failed to find similarity effects on well-being (Barelds,2005; Dyrenforth et al., 2010).
Third, in light of the dyadic nature of couple data it is necessaryto make use of proper analytic techniques and similarity measures.Thus far, many studies have treated dyadic data as if they wereindividual data, for instance by conducting analyses separatelyfor husbands and wives or using simple correlational methods thatfail to capture the interdependent nature of couple data. Thisshortcoming in previous studies has been pointed out as a majorproblem by relationship researchers and can be overcome byapplying appropriate analytical methods that take the interdepen-dence of dyadic data into account and are able to test for the un-ique effect of each independent variable on well-being (seeKenny, Kashy, & Cook, 2006).
Furthermore, studies on personality similarity in couples haveused a wide array of similarity measures. The most common mea-sures reported include difference scores and profile correlations.Difference scores are straightforward and intuitively understand-able in the way we usually think about differences in daily life(Griffin et al., 1999). These scores are typically computed at thetrait level, taking the absolute value of difference between twopartners’ scores on a given trait. They thus indicate how dissimilartwo members of a couple are with respect to a specific trait. Yetresearchers criticize this approach mainly due to the lowered
reliability inherent in difference scores. If the two component vari-ables are positively correlated, as is often the case given that thescores are usually measured with the same instrument, the reli-ability of the difference between those two components becomesless reliable (Edwards, 1994, 2001). Furthermore, difference scorescan be confounded with each partner’s individual score. A simplesolution to avoid this problem is to include both partners’ individ-ual scores in the same analysis (Griffin et al., 1999).
Profile correlations, on the other hand, are more difficult tointerpret. Generally, personality similarity computed at the profilelevel represents the degree to which both couple members’ overallpersonality profiles are similar to each other; that is, how well twopartners match on a set of personality traits (see Cronbach & Gle-ser, 1953). To interpret the effects of profile similarity adequately,it is important to know that a profile consists of three elements(Cronbach & Gleser, 1953; see also Furr, 2010). First, every profilehas a shape that represents the pattern of scores in a profile. It re-flects which traits have relatively high scores and which ones haverelatively low scores within the same profile. Elevation representsthe overall mean across all traits within a profile. And scatter refersto the variability or variance among the scores of a profile. It thusreflects how much the trait scores deviate within the same profile.
Because a profile comprises these three elements, similarity be-tween different profiles can be measured in various ways. A com-monly used profile correlation measure is the IntraclassCorrelation Coefficient (ICC). It reflects a global index because itcaptures all three characteristics of a profile at once (e.g., Dyren-forth et al., 2010). However, assuming ICC as a global index of per-sonality similarity might confound findings because shape,elevation, and scatter are conceptually different from one anotherand should not be mixed within the same analysis. Instead, Furr(2010) suggests a differentiated analysis and argues for the neces-sity to separately examine all elements of a profile. Hence, shapesimilarity is calculated by correlating the scores in one partner’sprofile with the scores of the other partner’s profile using Pearsoncorrelations. Second, elevation similarity is measured using differ-ence scores (i.e., absolute value of difference between the overallmean across traits: mean similarity). Third, scatter similarity is alsocomputed using difference scores (i.e., absolute value of differencebetween the variances across all traits within the profiles: variancesimilarity).
In light of the range of possible measures of similarity, anothergoal of the present study is to rule out that the use of different sim-ilarity indices results in different findings. By adopting a differen-tiated approach we can determine whether results varydepending on how profile similarity is measured. Some researcherswho applied difference scores and profile correlations in the samestudy reported that profile-based similarity was more stronglyassociated with satisfaction (Gaunt, 2006; Luo & Klohnen, 2005),whereas others did not find different effects with different similar-ity measures (Dyrenforth et al., 2010). No study to date has ana-lyzed similarity by means of a differentiated approach; that is,examining each element of a profile separately (as proposed byFurr, 2010).
In brief, the aim of the present study is to clarify the associationbetween personality similarity and life satisfaction in couples.Three reasons speak for the relevance of this research question.First, we think that some previous findings mostly represent anoverestimation of the effect of similarity because many studiesdid not take each partner’s personality into account (e.g., Robinset al., 2000). Thus, we will control for each partner’s individual le-vel of personality to examine the effect of personality similarity onlife satisfaction beyond the effects of one’s own and the partner’spersonality. In line with previous research, we expect actor andpartner effects for the association between personality and lifesatisfaction in couples (positive actor and partner effects for
T itTraitSimilarity Life Satisfactiony
Mane1
Big Five TraitMan Life Satisfaction
Womane2
Big Five TraitWoman
Fig. 1. Model specification for the five trait level models with trait similarity andthe Big Five traits of both partners as predictors of each partner’s life satisfaction.Trait similarity is represented by the absolute value of difference between twopartners’ scores on a given trait (multiplied by �1).
K. Furler et al. / Journal of Research in Personality 47 (2013) 369–375 371
Extraversion, Agreeableness, Conscientiousness, Emotional Stabil-ity as well as positive actor effects for Openness; e.g., Dyrenforthet al., 2010). Second, with the exception of only a few studies(Barelds, 2005; Dyrenforth et al., 2010), previous research mainlyused small or specific samples (e.g., Arrindell & Luteijn, 2000;Luo & Klohnen, 2005). Thus, in order to provide generalizable re-sults we use a nationally representative sample of couples livingin Switzerland. We expect to replicate findings of Dyrenforthet al., 2010 who analyzed comparable data sets of other countries.Third, we apply proper methods for dyadic data and extend priorresearch on similarity and well-being by adopting a differentiatedapproach: associations between personality similarity and life sat-isfaction in couples will be analyzed both at the trait and at theprofile level. The latter includes global profile similarity (ICC), aswell as similarity of shape, elevation (mean), and scatter(variance).
2. Method
2.1. Participants
Participants were drawn from the Swiss Household Panel (SHP),a nationally representative household panel study in Switzerland.The SHP is an annual survey, which started in 1999, of a randomsample of Swiss households. All adult members of selected house-holds were asked to participate using computer-assisted phoneinterviewing. Since its start, the SHP has collected data on closeto 8000 households, including a variety of social and demographicmeasures (for further information see Voorpostel et al., 2010).
The sample for the present study consists of all heterosexualromantic couples for which life satisfaction and personality dataof both partners were available in 2009, leading to a total samplesize of 1608 couples (Nind = 3216). Men were on average51.88 years old (SD = 13.58), women 49.10 years (SD = 13.35). Themean length of the relationship was 24.21 years (SD = 14.11) andthe majority of couples were married (85%).
2.2. Measures
Life satisfaction was assessed with a single item asking respon-dents how satisfied they felt with their life in general ranging from0 (not at all satisfied) to 10 (completely satisfied) (M = 8.14,SD = 1.24). Using single items is nearly custom and widely ac-cepted within large-scale panel studies to minimize subject burden(see also Lucas & Donnellan, 2012). Men and women did not differin their life satisfaction scores (t(1606) = .51, p = .61).
Personality was assessed using a German 10-item instrumentderived from the Big Five Inventory (BFI; see John & Srivastava,1999 for the original version). For each trait, participants reportedthe extent to which each of the two statements per personalitytrait described themselves on an 11-point scale ranging from 0(completely disagree) to 10 (completely agree). Correlations betweenthe two items for Extraversion, Agreeableness, Conscientiousness,Emotional Stability, and Openness were .39, .11, .24, .40, and .21,respectively, and are consistent with previously published researchusing personality data from the SHP (Anusic, Lucas, & Donnellan,2012). Furthermore, the 10-item BFI correlates highly with the ori-ginal BFI-44 scale (r = .83) and predicts around 70% of the varianceof the full scale (Rammstedt & John, 2007).
2.3. Rationale of analysis
We applied structural equation modeling to estimate differentsets of Actor-Partner Interdependence Models (APIM; Kennyet al., 2006). The APIM is a dyadic relationship model that takesinterdependencies between two members of a dyad into account
and models each dyad member’s effect on his/her own outcome(actor effect) as well as on the partner’s outcome (partner effect).Two sets of analyses were conducted.
The first set of analyses focused on personality similarity at thetrait level and comprised five models (trait level models; one foreach Big Five trait). In these models, life satisfaction of both couplemembers was predicted by each partner’s personality (actor andpartner effects) as well as the couple’s personality similarity on agiven trait. As mentioned earlier, personality similarity at the traitlevel is represented as the absolute value of difference betweentwo partners’ scores on a given trait. This difference score was thenmultiplied by �1 in order to be indicative for similarity instead ofdifference. Hence, five models were estimated to analyze actor,partner, and similarity effects in the association between the BigFive traits and life satisfaction of couples (see Fig. 1 for an illustra-tion of the trait level models).
The second set of analyses focused on personality similarity atthe profile level and consisted of two models. In both models lifesatisfaction of both couple members was predicted by each part-ner’s personality (actor and partner effects of all Big Five traits)as well as couples’ personality similarity using different profilesimilarity indices for the two models. (1) The global profile modelincluded a global index of profile similarity represented by theIntraclass Correlation Coefficient (ICC), which captures all elementsof a profile at once. This model estimated the overall or global ef-fect of profile similarity above and beyond each partner’s individ-ual level of personality. (2) The differentiated approach modelincluded three different profile similarity indices to capture eachelement of the profile separately (i.e., shape, mean, and variancesimilarity). Thus, this model provided insight into the unique pre-dictive power of each element as proposed by Furr (2010). More-over, these analyses allowed a thorough examination of whetherthe application of a differentiated approach with distinguishedprofile similarity indices for shape, elevation, and scatter yieldsvaluable additional information that goes beyond a global profilesimilarity index. The similarity indices used in the differentiatedapproach model were represented by (a) Pearson correlation coef-ficient for shape similarity, (b) the absolute value of difference be-tween partners’ overall mean across all Big Five traits (multipliedby �1) for mean similarity, and (c) the absolute value of differencebetween partners’ variance within their Big Five traits (multipliedby �1) for variance similarity. Hence, the differentiated approachmodel estimated the effect of three different profile similarity indi-ces above and beyond each partner’s individual level of personality(see Figs. 2 and 3 for an illustration of the global profile and the dif-ferentiated approach model, respectively).
3. Results
Descriptive statistics and correlations of all variables are pre-sented in Table 1. Intercorrelations between similarity indices
MeanSimilarity
Life SatisfactionWomanVariance
Similarity
ShapeSimilarity Life Satisfaction
Mane1
e2
Big Five TraitMan
Big Five TraitWoman
Fig. 3. Model specification for the differentiated approach model with shape, mean,and variance similarity as predictors of each partner’s life satisfaction, controllingfor all Big Five traits of both partners. Shape similarity is represented by the Pearsoncorrelation coefficient; mean similarity is represented by the absolute value ofdifference between partners’ overall mean across all Big Five traits (multiplied by�1); variance similarity is represented by the absolute value of difference betweenpartners’ variance within their Big Five traits (multiplied by �1). Covariancesbetween all predictor variables were included but are not displayed for reasons ofclarity.
Life SatisfactionWoman
Global Similarity
Life SatisfactionMan
e1
e2Big Five TraitMan
Big Five TraitWoman
Fig. 2. Model specification for the global profile model with the global index ofpersonality similarity as predictor of each partner’s life satisfaction, controlling forall Big Five traits of both partners. Global similarity is represented by the IntraclassCorrelation Coefficient (ICC). Covariances between all predictor variables wereincluded but are not displayed for reasons of clarity.
1 The invariant standardized regression coefficients across men and women wereaveraged using Fisher’s z-transformation.
2 An important issue concerns the fact that people tend to respond to items in anormative or stereotypical way. Because of this response pattern, profile correlationsbetween two random individuals are likely not zero (Furr, 2008; Kenny et al., 2006).To account for this potential source of confounding, the analyses were also conductedusing stereotype adjusted scores instead of raw scores (according to Kenny et al.,2006). However, results remained the same and thus, we only reported the results forthe raw scores.
372 K. Furler et al. / Journal of Research in Personality 47 (2013) 369–375
were generally low or moderate, which indicates independenceand conceptual differences between these indices. As an exception,the global similarity index (measured by ICC) and the shape simi-larity index (measured by Pearson correlation) were strongly cor-related. This is not surprising given the conceptual overlapbetween these two indices. As mentioned earlier, the ICC capturesall three elements of a profile (shape, elevation, and scatter),whereas the Pearson correlation captures the shape of a profile.Values of the different indices ranged from very low to very high(e.g., �.98 to .97 for global similarity) meaning that there are bothcouples that are highly similar as well as couples whose personal-ities are fairly dissimilar.
The first set of analyses at the trait level consisted of estimatingfive separate models for each Big Five personality trait (see Fig. 1).These were saturated models with zero degrees of freedom. Withineach model, estimates of actor, partner, and similarity effects didnot differ between men and women because constraining therespective paths to be invariant did not lead to a significant dete-rioration in model fit in either model (Extraversion: Dv2
(Ddf = 3) = 3.35, p = .34; Agreeableness: Dv2 (Ddf = 3) = 1.00,p = .80; Conscientiousness: Dv2 (Ddf = 3) = 1.37, p = .71; Emotional
Stability: Dv2 (Ddf = 3) = 1.83, p = .61; Openness: Dv2
(Ddf = 3) = 1.57, p = .67). Table 2 presents results of the five trait le-vel models with invariant actor, partner, and similarity effectsacross men and women.1 In line with previous research, one’sown as well as the partner’s personality was related to one’s lifesatisfaction. Actor effects were found for each Big Five trait such thatpeople high in Extraversion, Agreeableness, Conscientiousness,Emotional Stability, and Openness reported higher levels of life satis-faction. In a similar vein, having an extraverted, agreeable, conscien-tious, and emotional stable partner was also positively associatedwith one’s life satisfaction. Effect sizes for actor and partner estimateswere generally small (according to Cohen, 1988). No partner effectswere identified for Openness. Regarding personality similarity in cou-ples on a trait level, partners tended to be more similar than dissimilaron single Big Five personality traits (see means and ranges of traitsimilarity indices in Table 1). However, there was little evidence forsimilarity effects beyond the individual contribution of each partner’spersonality (see Table 2). Coefficients were generally small and onlyone effect reached significance, which has to be interpreted withcaution considering the large number of analyses conducted.
The second set of analyses focused on personality profiles pre-dicting life satisfaction. First, we established a global profile model(Fig. 2) with the ICC as global profile similarity index controlling foractor and partner main effects of the Big Five traits. This modelprovides an overall profile similarity effect on both couple mem-bers’ life satisfaction above and beyond the effects of one’s ownand the partner’s personality. Second, we adopted a differentiatedapproach to profile similarity and estimated a differentiated ap-proach model (Fig. 3) with profile indices of shape, mean, and var-iance similarity. Again, we controlled for actor and partner maineffects of all Big Five traits. This model provides the unique profilesimilarity effect of each element of a profile on both couple mem-bers’ life satisfaction above and beyond the effects of one’s ownand the partner’s personality. Actor, partner, and similarity esti-mates did not differ between men and women because constrain-ing the respective paths to be invariant did not lead to a significantdeterioration in model fit in either model (global profile model:Dv2 (Ddf = 11) = 10.05, p = .53; differentiated approach model:Dv2 (Ddf = 13) = 14.09, p = .37). Even though the similarity indicesreflect conceptually different estimates, results at the profile levelrevealed no similarity effects, neither in the global profile model(b = .00, p = .84; R2 = .13), nor in the differentiated approach model(bshape similarity = �.02, p = .32; bmean similarity = .00, p = 1.00; bvariance
similarity = .00, p = .95; R2 = .13). In short, after controlling for eachpartner’s personality traits, profile similarity was unrelated to bothpartners’ life satisfaction, both in the global model as well as in thedifferentiated approach model.2
Additionally, because our sample comprised a variety of coupleswith a broad range of relationship length, we examined the poten-tial influence of how long couples had been together by includingrelationship length as a moderator in all analyses. We added aninteraction term (relationship length � similarity) to all our traitas well as profile level models. Again, estimates did not differ be-tween men and women because constraining the respective pathsto be invariant did not lead to a significant deterioration in modelfit in either trait model (Extraversion: Dv2 (Ddf = 5) = 2.25, p = .66;Agreeableness: Dv2 (Ddf = 5) = 1.99, p = .85; Conscientiousness:
Table 1Intercorrelations and descriptives for study variables (N = 1608).
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.
Men1. Life satisfaction –2. Extraversion .18 –3. Agreeableness .23 .21 –4. Conscientiousness .22 .24 .27 –5. Emotional
stability.26 .25 .31 .29 –
6. Openness .08 .16 .04 .10 .12 –
Women7. Life satisfaction .29 .09 .10 .12 .13 .01 –8. Extraversion .09 .14 .03 .01 .08 �.06 .19 –9. Agreeableness .12 .07 .10 .07 .08 .03 .20 .22 –10. Conscientiousness .10 .07 .07 .05 .09 .03 .22 .21 .32 –11. Emotional
stability.11 .09 .06 .05 .04 .01 .29 .20 .31 .29 –
12. Openness .03 .00 .04 �.02 .01 .13 .12 .18 .11 .13 .09 –
Trait similarity indices13. Extraversion .01 .14 .02 .00 .03 .01 �.05 .00 �.02 �.07 �.02 .01 –14. Agreeableness �.02 �.03 .11 �.07 �.04 �.03 �.04 �.02 .04 �.01 �.01 .03 .08 –15. Conscientiousness .00 .04 .04 .20 .03 .01 .00 .01 �.03 .05 .02 �.03 .09 .11 –16. Emotional
stability�.01 .01 �.05 �.05 �.09 �.04 .07 .04 .06 .04 .32 �.01 .12 .06 .13 –
17. Openness �.02 �.02 �.09 �.03 �.07 .06 �.02 �.03 �.06 �.06 �.03 .00 .07 .03 .06 .06 –
Profile similarity indices18. Global .01 �.02 .03 .18 �.11 �.15 �.01 �.09 .03 .14 .01 �.19 .35 .17 .31 .33 .30 –19. Shape .02 .00 .06 .17 �.11 �.12 �.02 �.09 .05 .13 .08 �.17 .27 .07 .19 .24 .23 .85 –20. Mean �.03 �.01 �.04 .01 �.03 �.04 .00 .02 �.04 �.01 �.04 �.02 .36 .29 .35 .32 .28 .43 .01 –21. Variance .00 .07 .01 �.11 .06 .08 .03 .06 �.06 �.09 .14 .14 .27 .18 .18 .23 .21 �.11 �.05 .06 –M 8.13 6.44 6.81 7.49 6.81 6.24 8.14 6.75 6.82 7.67 6.28 6.34 �1.97 �1.45 �1.56 �1.90 �1.91 0.03 0.14 �1.02 �2.15SD 1.21 1.86 1.36 1.40 1.62 1.75 1.26 1.92 1.42 1.44 1.75 1.86 1.54 1.17 1.21 1.45 1.44 0.44 0.50 0.79 2.28Min 0.00 0.00 0.00 0.00 0.00 0.00 3.00 0.00 1.00 1.00 0.00 0.00 �9.00 �8.00 �8.00 �8.50 �8.00 �0.98 �0.97 �4.60 �17.78Max 10.00 10.00 10.00 10.00 10.00 10.00 10.00 10.00 10.00 10.00 10.00 10.00 0.00 0.00 0.00 0.00 0.00 0.97 1.00 0.00 0.00
Note: Correlation coefficients printed in bold are significant at the p < .05 or p < .01 level.
Table 2Standardized estimates of the five trait level models predicting life satisfaction ofboth partners.
Trait Standardized estimates R2
Actor Partner Similarity
Extraversion .18*** .07*** �.03 .04Agreeableness .21*** .10*** �.05** .06Conscientiousness .22*** .11*** �.04 .06Emotional stability .27*** .11*** �.01 .09Openness .10*** .01 �.02 .01
Note: Models were estimated separately for each trait. For each of the five models,both partners’ personality on the given trait and trait similarity were included aspredictors of life satisfaction. Estimates and explained variance (R2) refer to modelswith invariant actor, partner, and similarity effects across men and women.** p < .01.*** p < .001.
K. Furler et al. / Journal of Research in Personality 47 (2013) 369–375 373
Dv2 (Ddf = 5) = 0.51, p = .99; Emotional Stability: Dv2
(Ddf = 5) = 5.28, p = .38; Openness: Dv2 (Ddf = 5) = 2.21, p = .82)or profile model (global profile model: Dv2 (Ddf = 13) = 10.51,p = .65; differentiated approach model: Dv2 (Ddf = 17) = 16.70,p = .48). More importantly, no substantial effects point to similarityas a function of relationship length. Standardized estimates for theinteraction effects ranged from �.02 to .01, and were all not signif-icant (all p’s > .35). Only one interaction effect for similarity in Con-scientiousness reached significance (b = �.05, p < .01), but theeffect was too small to merit attention.
4. Discussion
The goal of the present study was to clarify the effects betweenpersonality similarity and life satisfaction in couples by examining
a large representative sample of romantic couples and adopting adifferentiated methodological approach. Overall, the results ofour study reveal little evidence for associations between personal-ity similarity in couples and each partner’s life satisfaction. Thisdoes not mean, however, that two partners’ personalities are unre-lated to their individual life satisfaction—rather the opposite istrue: in line with previous research, one’s own personality as wellas the partner’s personality is substantially related to both part-ners’ life satisfaction (Barelds, 2005; Dyrenforth et al., 2010). Thefact that not only within-person characteristics but also partnercharacteristics affect each partner’s life satisfaction gains remark-able importance, and even more so when bearing in mind the myr-iad of potential (within-person) correlates of subjective well-being.Nevertheless, we tried to plunge even further into dyadic effectsand laid our main emphasis on personality similarity as a predictorof life satisfaction over and above actor and partner effects of theBig Five traits.
Neither at the trait nor at the profile level did our results pro-vide evidence for personality similarity effects on life satisfaction.Only one effect for similarity at the trait level reached significance.However, the effect was too small to merit interpretation. Our re-sults rather affirm previous research with comparable data sets ofnationally representative samples that did not find any evidencefor effects of personality similarity on well-being (Barelds, 2005;Dyrenforth et al., 2010).
Although an issue in everyday life, empirical evidence does notsupport the common saying ‘‘like attracts like’’, that is, personalitysimilarity is not substantially related to life satisfaction in couples.Three main factors may account for this lack of relevance of per-sonality similarity for each couple member’s well-being. First, thepresent study controlled for actor and partner effects and results
374 K. Furler et al. / Journal of Research in Personality 47 (2013) 369–375
confirm previous research showing that personality similarity isnot a predictor of life satisfaction above and beyond each partner’sindividual level of personality (Barelds, 2005; Dyrenforth et al.,2010; Gattis et al., 2004). Thus, given the magnitude of both actorand partner effects, we again emphasize the importance of thor-oughly controlling for each partner’s individual level of personalitywhen assessing the effect of similarity in couples. Several studiesreporting significant similarity effects have neglected this issue(e.g., Gonzaga et al., 2007; Robins et al., 2000).
Second, one of the strengths of the present study is the use ofpanel data. Most previous research had smaller sample sizes(Arrindell & Luteijn, 2000) or laid the focus on specific samplessuch as newlyweds (Luo & Klohnen, 2005). In contrast, results pre-sented here are representative of heterosexual couples living inSwitzerland. Our results support findings of Dyrenforth et al.(2010) as well as Barelds (2005), who also made use of nationallyrepresentative samples of couples. When examining very largedata sets, similarity in personality is not systematically related tosatisfaction.
Third, we estimated different sets of Actor-Partner Interdepen-dence Models to take account of the dyadic nature of the data. Sev-eral previous studies have failed to capture the interdependentdata structure in their analyses, which renders interpretation ofsimilarity effects in couples problematic (e.g., Gonzaga et al.,2007). Thus, our results demonstrate why it is necessary to applyadequate analytic techniques to dyadic data (e.g., as proposed byKenny et al., 2006). Furthermore, a recurring issue in this area ofresearch is the question of which similarity measure to use. Hence,another asset of our study is the adoption of a differentiated ap-proach to similarity. We captured conceptually different estimatesof personality similarity and fragmented our analyses through sep-arate examinations of trait and profile (global, shape, mean, andvariance) similarity indices as predictors of life satisfaction. Eventhough the similarity measures used in our study are conceptuallydifferent from one another, none of them had an effect on eitherpartner’s life satisfaction. Thus, adopting a differentiated approachdid not provide additional information that goes beyond the globalsimilarity index. This supports the conclusion of Dyrenforth et al.(2010) that as long as main effects are controlled, findings suggestno relation between personality similarity and life satisfactionregardless of the similarity index used. However, we do not ruleout the possible advantage of a differentiated approach in general.It is essential to understand the conceptual difference of the ap-plied measures and we therefore believe that future research couldbenefit from this differentiated approach to analyze similarity.
Alternative theoretical explanations for mixed findings in pastresearch can be ascribed to the assumption of general effects ofsimilarity on well-being. Future research should rather investigatedifferential effects between couples: whereas being similar mightbe beneficial for some couples, for others it may not be beneficialas they are more satisfied with a partner who is dissimilar and thuscomplements them. Alternatively, underlying relationship struc-tures or relationship maintenance processes might lead some peo-ple to be more satisfied with a similar partner, whereas othersprefer a dissimilar partner. Moreover, future research should morethoroughly investigate the processes and mechanisms throughwhich personality is linked to important life outcomes. Motiva-tional aspects of personality for instance could explain associationsbetween personality similarity on single traits and life satisfactionin couples and could serve as a potential mediator. Such analyseshave been fairly neglected in couple research so far. Furthermore,although we did not find evidence that similarity on the Big Fivetraits is related to life satisfaction in couples, similarity in other do-mains (within or outside personality) may very well be importantfor well-being. In this vein, previous research provides evidence forthe relevance of similar attachment characteristics for marital sat-
isfaction in a sample of newlywed couples (Luo & Klohnen, 2005)or for the relevance of gender-related attitudes on dyadic adjust-ment (Aube & Koestner, 1995). Finally, one could argue that per-sonality similarity may operate differently as a function ofrelationship stage. For instance, being similar to one’s partnermight be of greater relevance for one’s well-being during acquain-tanceship and mating, whereas complementing each other mightbe more beneficial in a more advanced phase of the relationship.Because our sample comprised a broad range of couples in differ-ent relationship stages, we included relationship length as a poten-tial moderator into our models. However, these additional analysesbarely yielded evidence for alternative associations between per-sonality similarity and satisfaction as a function of relationshipstage.
Besides these strengths, we should also consider the limitationsof this study. First, similar to other studies using panel data, wewere confined to using short scales for the analyses. To ensure abroad assessment of the construct, personality traits were assessedwith very few items per trait with rather low intercorrelations. As aconsequence, internal consistencies of the Big Five traits wererather low in this sample. However, short scales are known to havelower reliabilities and are widely used in large-scale panel studiesdue to time constraints.
Another limitation of this study concerns life satisfaction as theprimary outcome variable within a dyadic research question ofromantic couples. Life satisfaction represents a cognitive evalua-tion of one’s life in general (Diener, Suh, Lucas, & Smith, 1999;Schimmack, 2008) and is thus a more or less highly individual ap-praisal of one’s life. At least in the context of romantic relation-ships, personality similarity may be more relevant for a dyadicwell-being outcome such as relationship quality or relationshipsatisfaction. Unfortunately, however, the SHP does not assess rela-tionship satisfaction.
In sum, personality is highly relevant in explaining life satisfac-tion. Of remarkable interest, not only one’s own personality butalso the partner’s personality matters for each partner’s well-being.However, after controlling for these actor and partner effects, beingsimilar to one’s partner is not related to life satisfaction. Adopting adifferentiated approach revealed similar results for all similarityindices.
Acknowledgements
This study has been realized using the data collected by theSwiss Household Panel (SHP), which is based at the Swiss Centreof Expertise in the Social Sciences FORS. The SHP is financed bythe Swiss National Science Foundation.
This research was supported by the Swiss National ScienceFoundation Sinergia program ‘‘Co-Development in Personality:Longitudinal Approaches to Personality Development in Dyadsacross the Life Span’’ (CRSI11_130432/1).
We thank Laura Wiles for proofreading.
References
Anderson, C., Keltner, D., & John, O. P. (2003). Emotional convergence betweenpeople over time. Journal of Personality and Social Psychology, 84, 1054–1068.http://dx.doi.org/10.1037/0022-3514.84.5.1054.
Antill, J. K. (1983). Sex role complementarity versus similarity in married couples.Journal of Personality and Social Psychology, 45, 145–155. http://dx.doi.org/10.1037/0022-3514.45.1.145.
Anusic, I., Lucas, R. E., & Donnellan, M. B. (2012). Cross-sectional age differences inpersonality: Evidence from nationally representative samples from Switzerlandand the United States. Journal of Research in Personality, 46, 116–120. http://dx.doi.org/10.1016/j.jrp.2011.11.002.
Argyle, M. (1999). Causes and correlates of happiness. In D. Kahneman, E. Diener, &N. Schwarz (Eds.), Well-being: The foundations of hedonic psychology(pp. 353–373). New York: Russell Sage Foundation.
K. Furler et al. / Journal of Research in Personality 47 (2013) 369–375 375
Arrindell, W. A., & Luteijn, F. (2000). Similarity between intimate partners forpersonality traits as related to individual levels of satisfaction with life.Personality and Individual Differences, 28, 629–637. http://dx.doi.org/10.1016/S0191-8869(99)00125-7.
Aube, J., & Koestner, R. (1995). Gender characteristics and relationship adjustment:Another look at similarity–complementarity hypotheses. Journal of Personality,63, 879–904. http://dx.doi.org/10.1111/j.1467-6494.1995.tb00319.x.
Barelds, D. P. H. (2005). Self and partner personality in intimate relationships.European Journal of Personality, 19, 501–518. http://dx.doi.org/10.1002/per.549.
Botwin, M. D., Buss, D. M., & Shackelford, T. K. (1997). Personality and matepreferences: Five factors in mate selection and marital satisfaction. Journalof Personality, 65, 107–136. http://dx.doi.org/10.1111/j.1467-6494.1997.tb00531.x.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.).Hillsdale, NJ: Erlbaum.
Cronbach, L. J., & Gleser, G. C. (1953). Assessing similarity between profiles.Psychological Bulletin, 50, 456–473. http://dx.doi.org/10.1037/h0057173.
Diener, E., Suh, E. M., Lucas, R. E., & Smith, H. L. (1999). Subjective well-being: Threedecades of progress. Psychological Bulletin, 125, 276–302. http://dx.doi.org/10.1037/0033-2909.125.2.276.
Dyrenforth, P. S., Kashy, D. A., Donnellan, M. B., & Lucas, R. E. (2010). Predictingrelationship and life satisfaction from personality in nationally representativesamples from three countries: The relative importance of actor, partner, andsimilarity effects. Journal of Personality and Social Psychology, 99, 690–702.http://dx.doi.org/10.1037/A0020385.
Edwards, J. R. (1994). Regression analysis as an alternative to difference scores.Journal of Management, 20, 683–689. http://dx.doi.org/10.1016/0149-2063(94)90011-6.
Edwards, J. R. (2001). Ten difference score myths. Organizational Research Methods,4, 265–287. http://dx.doi.org/10.1177/109442810143005.
Epstein, E., & Guttman, R. (1985). Mate selection in man: Evidence, theory, andoutcome. Social Biology, 31, 243–278. http://dx.doi.org/10.1080/19485565.1984.9988579.
Furr, R. M. (2008). A framework for profile similarity: Integrating similarity,normativeness, and distinctiveness. Journal of Personality, 76, 1267–1316.http://dx.doi.org/10.1111/J.1467-6494.2008.00521.x.
Furr, R. M. (2010). The double-entry intraclass correlation as an index of profilesimilarity: Meaning, limitations, and alternatives. Journal of PersonalityAssessment, 92, 1–15. http://dx.doi.org/10.1080/00223890903379134.
Gattis, K. S., Berns, S., Simpson, L. E., & Christensen, A. (2004). Birds of a feather orstrange birds? Ties among personality dimensions, similarity, and maritalquality. Journal of Family Psychology, 18, 564–574. http://dx.doi.org/10.1037/0893-3200.18.4.564.
Gaunt, R. (2006). Couple similarity and marital satisfaction: Are similar spouseshappier? Journal of Personality, 74, 1401–1420. http://dx.doi.org/10.1111/J.1467-6494.2006.00414.x.
Gonzaga, G. C., Campos, B., & Bradbury, T. (2007). Similarity, convergence, andrelationship satisfaction in dating and married couples. Journal of Personalityand Social Psychology, 93, 34–48. http://dx.doi.org/10.1037/0022-3514.93.1.34.
Gonzaga, G. C., Carter, S., & Buckwalter, J. G. (2010). Assortative mating,convergence, and satisfaction in married couples. Personal Relationships, 17,634–644. http://dx.doi.org/10.1111/j.1475-6811.2010.01309.x.
Griffin, D., Murray, S., & Gonzalez, R. (1999). Difference score correlations inrelationship research: A conceptual primer. Personal Relationships, 6, 505–518.http://dx.doi.org/10.1111/j.1475-6811.1999.tb00206.x.
Headey, B., Muffels, R., & Wagner, G. G. (2010). Long-running German panel surveyshows that personal and economic choices, not just genes, matter for happiness.Proceedings of the National Academy of Sciences of the United States of America,107, 17922–17926. http://dx.doi.org/10.1073/Pnas.1008612107.
Hoppmann, C. A., Gerstorf, D., Willis, S. L., & Schaie, K. W. (2011). Spousalinterrelations in happiness in the Seattle Longitudinal Study: Considerablesimilarities in levels and change over time. Developmental Psychology, 47, 1–8.http://dx.doi.org/10.1037/a0020788.
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 (pp. 102–138). New York: GuilfordPress.
Kenny, D. A., & Acitelli, L. K. (1994). Measuring similarity in couples. Journal ofFamily Psychology, 8, 417–431. http://dx.doi.org/10.1037/0893-3200.8.4.417.
Kenny, D. A., Kashy, D. A., & Cook, W. L. (2006). Dyadic data analysis. New York:Guilford Press.
Lucas, R. E., & Donnellan, M. B. (2012). Estimating the reliability of single-item lifesatisfaction measures: Results from four national panel studies. Social IndicatorsResearch, 105, 323–331. http://dx.doi.org/10.1007/s11205-011-9783-z.
Luo, S. H., & Klohnen, E. C. (2005). Assortative mating and marital quality innewlyweds: A couple-centered approach. Journal of Personality and SocialPsychology, 88, 304–326. http://dx.doi.org/10.1037/0022-3514.88.2.304.
Rammstedt, B., & John, O. P. (2007). Measuring personality in one minute or less: A10-item short version of the Big Five Inventory in English and German. Journalof Research in Personality, 41, 203–212. http://dx.doi.org/10.1016/J.Jrp.2006.02.001.
Robins, R. W., Caspi, A., & Moffitt, T. E. (2000). Two personalities, one relationship:Both partners’ personality traits shape the quality of their relationship. Journalof Personality and Social Psychology, 79, 251–259. http://dx.doi.org/10.1037/0022-3514.79.2.251.
Schimmack, U. (2008). The structure of subjective well-being. In M. Eid & R. J. Larsen(Eds.), The science of subjective well-being (pp. 97–123). New York: GuilfordPress.
Steel, P., Schmidt, J., & Shultz, J. (2008). Refining the relationship betweenpersonality and subjective well-being. Psychological Bulletin, 134, 138–161.http://dx.doi.org/10.1037/0033-2909.134.1.138.
Vandenberg, S. G. (1972). Assortative mating, or who marries whom? BehaviorGenetics, 2, 127–158. http://dx.doi.org/10.1007/BF01065686.
Voorpostel, M., Tillmann, R., Lebert, F., Weaver, B., Kuhn, U., Lipps, O., et al. (2010).Swiss Household Panel Userguide (1999–2009), Wave 11, October 2010. Lausanne:FORS.
Watson, D., Klohnen, E. C., Casillas, A., Simms, E. N., Haig, J., & Berry, D. S. (2004).Match makers and deal breakers: Analyses of assortative mating in newlywedcouples. Journal of Personality, 72, 1029–1068. http://dx.doi.org/10.1111/j.0022-3506.2004.00289.x.
