Universal Features of Personality Traits From the Observer ...

1 Personality PROCESSES AND INDIVIDUAL DIFFERENCESU niversal Features of Personality Traits From the Observer s Perspective:Data From 50 CulturesRobert R. McCrae and Antonio TerraccianoNational Institute on Aging, National Institutes of Health,Department of Health and Human Services78 Members of the Personality Profilesof Cultures ProjectTo test hypotheses about the universality of Personality Traits , college students in 50 cultures identifiedan adult or college-aged man or woman whom they knew well and rated the 11,985 targets using the3rd-person version of the Revised NEO Personality Inventory. Factor analyses within cultures showedthat the normative American self-report structure was clearly replicated in most cultures and wasrecognizable in all. Sex differences replicated earlier self-report results, with the most pronounceddifferences in Western cultures. Cross-sectional age differences for 3 factors followed the patternidentified in self-reports, with moderate rates of change during college age and slower changes after age40.

2 With a few exceptions, these data support the hypothesis that Features of Personality Traits arecommon to all human claims have recently been made about the universality ofpersonality Traits . McCrae and Costa (1997) argued that the five-factor model (FFM) of Personality is found in all cultures,1ahypothesis subsequently supported in a wider range of cultures(Rolland, 2002). McCrae et al. (1999) reported that cross-sectionalage differences were similar in different cultures whose cohortshad experienced very different life histories, and Costa, Terrac-ciano, and McCrae (2001) reported pancultural patterns of genderdifferences. McCrae, Costa, Martin, et al. (2004) provided data oncross- Observer agreement suggesting that even in collectivisticcultures, where there is purportedly a greater emphasis on rela-tionships than on Traits , people accurately perceive their own andothers Traits . These recurring regularities despite differences inlanguage, history, religion, and culture suggest that personalityRobert R.

3 McCrae and Antonio Terracciano, Gerontology ResearchCenter, National Institute on Aging; the 78 contributing members of thePersonality Profiles of Cultures 78 members of the Personality Profiles of Cultures Project are listedin order of data submission in the R. McCrae receives royalties from the Revised NEO PersonalityInventory (NEO-PI-R). Portions of these data were presented at the SecondWorld Congress on Women s Mental Health, Washington, DC, March2004. Czech participation was supported by Grant 406/01/1507 from theGrant Agency of the Czech Republic and is related to Research Plan AV0Z7025918 of the Institute of Psychology, Academy of Sciences of theCzech Republic. Sami Gu lgo z s participation was supported by the TurkishAcademy of Sciences. Burkinabe` and French Swiss participation wassupported by a grant from the Swiss National Science Foundation to The data collection in Hong Kong was supported by RGC DirectAllocation Grants DAG02 and DAG03 awarded toM.

4 Yik. Data collection in Malaysia was supported by UKM FundamentalResearch Grant 11JD/015 thank Dave Schmitt for contact information on potential collabora-tors, Bob Smith for authorizing use of the NEO-PI-R, and the translatorswhose earlier work on Form S of the NEO-PI-R made this project assistance on this project we thank Herbert Biggs, Luciana de Almeida,Hudson W. Carvalho, Marco Montarroyos Calegaro, Andre ia da Silva Bez,Zheng Li, Ana Butkovic , Ole Dreyer, Susy Ball, Anna Gramberg, Ho-nathan Harrow, V. S. Bose, Suguna Kannan, K. Sarita, K. Madhavi,Lidwina Dominica R, Vina Bunyamin, Hiromi Imuta, Kenji Sugiyama,Midori Takayama, Rozita Kamis, Rosmaini Ismail, Anna Nedtwig, Zach-ary Smith, Aaron Wolen, Maya Tamir, Christie Napa Scollon, Valery , Ivan G. Senin, J. C. Munene, Silvo Kozelj, Manca Jakic, SimonaZbac nik, Nadia Messoulam, Facundo Abal, Fernanda Molina, DaianaBion, Sebastia n Mosquera, Ludmila Firpo, Lorena Etcheverry, FernandoVera, Catherine Currell, Richard Chan, Christopher Paik, Herbert , Andreas Fink, and Cornelia , Russian, and Czech data were taken from earlier studies (Mc-Crae, Costa, Hr eb c kova , et al.)

5 , 2004; Ostendorf & Angleitner, 2004), andportions of the Brazilian, Lebanese, and Thai data are also reported inchapters in McCrae (in press); McCrae, Terracciano, and Khoury (inpress); and Costa and McCrae (in press), concerning this article should be addressed to Robert , Gerontology Research Center, National Institute on Aging, Na-tional Institutes of Health, Department of Health and Human Services, Box#03, 5600 Nathan Shock Drive, Baltimore, MD 21224-6825. this article, we use the termculturesloosely to refer tonations or subgroups within nations. We are aware that nations do not havemonolithic cultures and that our samples do not necessarily reflect the fullcultural diversity seen within of Personality and Social Psychology, 2005, Vol. 88, No. 3, 547 561In the public domainDOI: are basic Features of the human species (Allik & McCrae,2002). The present study offers new tests of these hypotheses obvious limitation to prior claims is that cultures andsubcultures have not been exhaustively studied.

6 Only a few Afri-can and no Arabic cultures have been included in previousstudies using the Revised NEO Personality Inventory (NEO-PI-R;Costa & McCrae, 1992a) or other measures of the FFM ( ,Heaven, Connors, & Stones, 1994). No preliterate cultures havebeen examined; in fact, most studies have used college studentsamples, whose members may be relatively second limitation is that most studies have relied exclusivelyon self-report methods, leaving the possibility that method artifactsmay be responsible for some or all of the findings. Observerratings form an alternative method of Personality measurement,known to be convergent but not wholly redundant with self-reports(McCrae, Costa, Martin, et al., 2004). In American studies, ob-server ratings typically yield similar conclusions about structureand about age and gender differences ( , Costa & McCrae,1992b), but this is not invariably the case in cross-cultural re-search.

7 For example, Extraversion and Openness to Experienceboth appeared to decline cross-sectionally in German adults whenself-reports were analyzed but not when Observer ratings wereanalyzed (McCrae et al., 2000). In a Czech sample, age associa-tions found in self-reports were replicated in Observer ratings forExtraversion and Openness but not for Neuroticism or Agreeable-ness (McCrae, Costa, Hr eb c kova , et al., 2004).The NEO-PI-R offers two versions: a self-report Form S and anobserver rating Form R, with the same items rephrased in the thirdperson. The factor structure of Form R in American samplesclosely resembles that of Form S ( , Piedmont, 1994), and thesame is true in German-, Russian-, and Czech-language versions(McCrae, Costa, Martin, et al., 2004; Ostendorf & Angleitner,2004). However, there appear to be no published studies of thefactor structure of Observer rating measures of the FFM in non-Western cultures. The present article includes data from more thana and Present DesignsMost previous cross-cultural studies of the FFM were based onsecondary analyses of data collected for a variety of purposes(Costa et al.)

8 , 2001; McCrae, 2002; Rolland, 2002). Samples variedin size and composition (although only normal volunteer data wereused) and in the time period of data collection. In many cases, onlysummary statistics were available, and demographic data weregenerally not available. No attempt was made to assess or controlthe quality of the data. It is testimony to the robustness of theunderlying effects that clear regularities emerged despite the present study, we collected data from college studentswho were asked to identify an individual from one of four targetgroups college-aged men, college-aged women, adult men, andadult women and provide ratings of that target on Form R of theNEO-PI-R. Because a uniform approach was taken to data collec-tion, results are more likely to be comparable across cultures ( , 1992). Samples are similar in size, age and sex oftargets, and time period in which data were collected. In addition,item-level data and basic demographics are available for use of college student raters also offers advantages.

9 Collegestudents are not, in general, representative of their national popu-lations, and this is particularly likely to be true in less affluentcultures. However, this fact is less problematic in Observer ratingstudies than in self-report studies: Raters could choose anyone theyknew well as a target, yielding a wider age and educational rangethan would normally be obtained in self-report studies. For exam-ple, about 11% of the targets in the present study had fewer than9 years of addition, college students may be more familiar and comfort-able with questionnaire methods than members of the generalpopulation, especially in non-Western cultures (cf. Marsella, Du-banoski, Hamada, & Morse, 2000), yielding more meaningful , even college students may differ across cultures intest-taking experience and attitudes; in particular, cultural differ-ences in acquiescence have been noted (Smith, 2004). In thepresent study, we attempted to assess the quality of the data in eachsample, to compare cultures on data quality, and to take qualityinto account in interpreting results.

10 It must be stressed that qualityis primarily a function of the fit between an imported Westernpersonality measure and the experiences and attitudes of eachculture; poorer data quality should not be seen as evidence ofproblems with either the instrument or the respondents but ratherof their , the use of Observer ratings permits an analysis of certainaspects of person perception and assessment. When self-reports areexamined, target and rater are completely confounded, making itimpossible to know whether ratings are a function of the personbeing rated or the person making the ratings. It is possible, forexample, that women everywhere score higher on Neuroticism notbecause they are less emotionally stable but merely because theyare better able to perceive negative affect (cf. Feldman Barrett,Lane, Sechrest, & Schwartz, 2000; Terracciano, Merritt, Zonder-man, & Evans, 2003) or more willing to attribute it to a target thanmen are. In the present design, both men and women rate men andwomen, so it is possible to estimate sex differences in rating stylesor recruited collaborators from a wide range of cultures, subject to therequirement that prospective participants would be fluent in English or oneof the other languages for which an authorized NEO-PI-R translation wasavailable.