Stability and Change in AdultPersonality: Genetic andEnvironmental ComponentsNANCY L. PEDERSEN1,2* and CHANDRA A. REYNOLDS21Division of Genetic Epidemiology, Institute of EnvironmentalMedicine, The Karolinska Institute, Stockholm, Sweden2Department of Psychology, University of Southern California,Los Angeles, USAAbstractThe enduring nature of personality, particularly in adulthood, has been demonstratedin numerous cross-sectional and longitudinal studies where high stability has beenobserved even with decades between testing intervals. Biometrical studies, reportinggenetic eects as the primary cause of familial resemblance, have been interpreted aslending further support to theories concerning an inherent stability of personality.Heterogeneity in heritability estimates across age cohorts may, however, alter thisnotion. Furthermore, recent phenotypic studies report some evidence for change over thelife course for characteristics such as `outgoingness'. The purpose of the present analysiswas to examine longitudinal stability and change in the sources of variation in personalityin the latter half of the life-span using a twin/adoption design with up to four times ofmeasurement. Data from the Swedish Adoption/Twin Study of Aging (SATSA) are usedboth to demonstrate how genetic and environmental eects can contribute to phenotypicstability, and to estimate the extent to which these in¯uenc es are, themselves, stable.Particularly intriguing are ®ndings of increasing variability in rate of change despiterelative mean level stability as well as genetic stability.#1998 John Wiley & Sons, Ltd.INTRODUCTIONModels of adult personalitySeveral theorists have posited models of personality ranging from numerous factors(e.g. Catell, 1947) to as few as three major factors (e.g . extraversion (E), neuroticism(N), and psychoticism (P); Eysenck, 1991). A model that is recently enjoying somefavour is one which orders theoretically posed personality traits as well as naturallanguage trait adjectives into a framework of ®ve dimensions: neuroticism (N),CCC 0890±2070/98/050365±22$17.50 Received 30 June 1998#1998 John Wiley & Sons, Ltd. Accepted 30 July 1998European Journal of PersonalityEur. J. Pers. 12, 365±386 (1998)*Correspondence to: Nancy L. Pedersen, Division of Genetic Epidemiology, Institute of EnvironmentalMedicine, The Karolinska Institute, Box 210, 17177 Stockholm. E-mail: [email protected] grant sponsor: MacArthur Foundation Network in Successful Aging; National Institute ofAging; Swedish Council for Social Research.extraversion (E), intellect or openness to experience (O), agreeableness (A) and con-scientiousness (C) (see Goldberg, 1993). Within these broad factors a multitude oftraits is contained. We shall focus our analysis and discussion on three of the Big-Fivetraits: neuroticism (N), extraversion (E) and openness to experience (O).Longitudinal Studies of the Big-Five dimensions and related traitsKogan (1990) reviewed four major longitudinal studies of personality: (i) TheBaltimore Longitudinal Study of Aging (Costa, McCrae and Arenberg, 1983), (ii)Kelly's (1955) longitudinal study of engaged couples initiated in the 1930s with twolater follow-up studies (Conley, 1984; 1985), (iii) the Cardiovascular Disease Project(Finn, 1986), and (iv) the Oakland Growth Study (Haan, Millsap and Hartka, 1986).Across a variety of measures and methods the studies demonstrated substantialstability of personality across long intervals of testing. However, conclusions regard-ing mean-level stability have not been consistent across studies, partly re¯ectingdierences in the initial ages of longitudinal samples (e.g. college samples versus post-college adults). It may be that less change is found with increasing age. For example,Haan et al. (1986) found stability to be lowest from late adolescence to early adult-hood, whereas Finn's (1986) ®ndings were supportive of high stability for someMMPI scales across a 30 year period in college-age and middle-aged men.Recent evidence has been presented for change across 30 or 40 year spans forself-con®dence, cognitive commitment, outgoingness, and dependability (Jonesand Meredith, 1996) as well as change in stimulation-seeking, both cross-sectionallyand longitudinally (Giambra, Camp and Grodsky, 1992). The study of Jones andMeredith (1996) applied latent curve methods to longitudinal data from the BerkeleyGuidance Study and the Oakland Growth Study. Relevant results indicated increasesin outgoingness over time for both men and women from ages 18 to 62 years of age.Furthermore, nine per cent of the women showed particularly striking increases inoutgoingness between 30 and 40 years of age. The study of Giambra et al. (1992)found signi ®cant decreases in average scores on two sensation-seeking measures(boredom susceptibility, need for external stimulation) utilizing both cross-sectionaland longitudinal information from the Baltimore Longitudinal Study of Aging.However, results of gender were less clear, given that longitudinal decreases werefound for women only on the boredom scale and decreases on the need for externalstimulation scale were found for men only.Regardless of the extent of mean-level stability, there is considerable variability inbehavior at all ages (Nelson and Dannefer, 1992; Neugarten, 1977). Behavioral geneticstudies can ad dress issues concerning the source of these individual dierences, and,more importantly, evaluate the extent to which phenotypic stability re¯ects geneticand environmental in¯uences (for a detailed description of various concepts andtechniques, see Pedersen, 1993). Genetic factors can contribute to change as well as tostability. Demonstrating geneti c in¯uence implies neither immutability nor stability;some genes may be `turned on' only at speci®c stages during the life span while othersmay be inactivated. Temporal genes may be involved in the timing of speci®c age-related events (Farrar, 1987; Paigen, 1980). Presumed examples of genetic changeoccur during puberty and menopause, when dierent genes governing hormoneproduction become active and later inactive. Even moderately stable traits may bein¯uenced by dierent sets of genes at dierent periods in the life span (Plomin, 1986).366 N. L. Pedersen and C. A. Reynolds#1998 John Wiley & Sons, Ltd. Eur. J. Pers. 12, 365±386 (1998)Behavioral genetic studies of personalityThat genetic eects are important for individual dierences in