Behavioral genetics of the higher-order factors of the Big FiveIntroductionMethodParticipantsMeasuresStatistical analysesExploratory phenotypic analysesExploratory biometric analysesConfirmatory biometric analysesResultsDiscussionReferencesBehavioral genetics of the higher-order factors of the Big FiveKerry L. Janga,*, W. John Livesleya, Juko Andob,Shinji Yamagatac, Atsunobu Suzukic, Alois Angleitnerd,Fritz Ostendorfd, Rainer Riemanne, Frank SpinathfaDepartment of Psychiatry, University of British Columbia, 2255 Wesbrook Mall, Vancouver, BC, Canada V6T 2A1bKeio University, Tokyo, JapancUniversity of Tokyo, Tokyo, JapandUniversity of Bielefeld, GermanyeUniversity of Jena, GermanyfUniversity of Sa¨arucken, GermanyReceived 23 May 2005; accepted 1 November 2005Available online 27 March 2006AbstractThere is empirical evidence that underlying the Big Five personality factors are two higher-order factorswhich have come to be known as ‘‘alpha’’ (a) and ‘‘beta’’ ( b). The a factor is defined by the agreeableness,conscientiousness, and emotional stability domains; whereas b is delineated by extraversion and intellect. Ithas been argued that a and b are important constructs because they bridge the gap between psychometricstudies of personality and theories of personality development. However, it is unclear if a and b are con-structs that can be reliably reproduced across a diverse range of independent samples. In a sample of1209 MZ and 701 DZ twin pairs from Canada, Germany, and Japan who completed the NEO-PI-R, fac-torial analyses of the five NEO-PI-R domains extracted two factors resembling a and b. Subsequent mul-tivariate genetic analyses revealed that this factor structure was a clear reflection of the organizing effects ofmultiple genetic influences, providing evidence for a and b as stable heuristic devices that can be used tointegrate personality measurement and developmental theory. 2006 Elsevier Ltd. All rights reserved.0191-8869/$ - see front matter  2006 Elsevier Ltd. All rights reserved.doi:10.1016/j.paid.2005.11.033*Corresponding author. Tel.: +1 604 822 7895; fax: +1 604 822 7756.E-mail address: [email protected] (K.L. Jang).www.elsevier.com/locate/paidPersonality and Individual Differences 41 (2006) 261–272Keywords: Twins; Heritability; Big Five; Alpha; Beta1. IntroductionThere is growing evidence that the Big Five personality factors agreeableness (A), conscien-tiousness (C), emotional stability (ES), extraversion (Ex), and intellect (I) – considered to repre-sent the basic descriptive units of personality – can be further reduced to two super-ordinatefactors. For example, Digman (1997) consistently extracted two higher-order factors he called ‘‘al-pha’’ (a) and ‘‘beta’’ (b) from several sets of published intercorrelations between these factors. Al-pha was typically defined by factor loadings from A, C and ES whereas beta was delineated byloadings from Ex and I.Digman (1997) argued that these higher-order traits are important for personality research be-cause they provide a tangible link between psychometric models of personality used for the devel-opment of reliable taxonomies and measures for theories of personality development and humannature in general. For example, he wrote that complex ideas such as Freud’s theories on psycho-social development or Adler’s ‘‘social interest’’ which encompasses several concepts includingaggression, hostility, impulse restraint and neurotic defence could only be captured by a broadconstruct like a, unlike A, C or ES could alone. Similarly, the fusion of Ex and I as b is a far morecomprehensive reflection of constructs such as Roger’s ‘‘personal growth’’, Adler’s ‘‘superioritystriving’’, or Maslow’s ‘‘self-actualization’’ than is possible by either Ex or I individually.Despite the apparent increase in conceptual clarity afforded by a and b, their actual relationshipto developmental concepts has not been evaluated empirically. Such work cannot be conducteduntil the stability of these higher-order traits is better established. Although Digman’s (1997) anal-yses typically extracted two higher-order factors, their composition was often defined by differentBig Five factors in different samples. For example, in his analysis of Graziano and Ward’s (1992)and John et al.’s (1984) data, a was defined only by A and C. In John, Goldberg, and Angleitner’s(1984) data, b is defined solely by Ex. However, in Digman’s (1963, 1994) own data, b was definedby Ex, I, and to some extent by ES (factor loadings just above threshold values of .40). Moreover,despite the differences in the composition of a and b, other reports have suggested the presence ofa third higher-order factor. From an examination of several cross-national datasets, De Raad andPeabody (2005) reported evidence for higher-order factors that simply resembled broader versionsof extraversion, agreeableness, and conscientiousness. Most surprisingly, neuroticism (a variant ofemotional stability) was completely missing from their factor solutions which represents a funda-mental departure from Digman’s (1997) findings.One possible reason for these differences is that the above reports assessed the Big Fiveusing different scales and measures. Any variation in the item content of the measures could havean appreciable effect on the patterns of covariance between them. This source of error could beaddressed by the use of a common instrument. Another explanation is that factorial variance isan unavoidable consequence of lexical models of personality. This could be addressed by exam-ining the factor structure of a common measure across linguistically distinctive cultures to assessthe effect of language differences and comprehension of concepts underlying a commonmeasure.262 K.L. Jang et al. / Personality and Individual Differences 41 (2006) 261–272This raises an associated issue of systematic differences in personality perception across differentcultures as described by ‘‘implicit personality theory’’. However, McCrae, Jang, Angleitner, Rie-mann, and Livesley (2001) found that observed covariance structure of traits was indeed a reflec-tion of the organizing effects of multiple genetic influences, such as the gene controlling serotonintransporter (5-HTTLPR) which accounts for some of the covariance between Costa and McC-rae’s (1992) NEO-PI-R Agreeableness and