PERSONALITY PROCESSES AND INDIVIDUAL DIFFERENCESIGenetic and Environmental Influences on Observed Personality:Evidence From the German Observational Study of Adult TwinsPeter BorkenauMartin-Luther-UniversitatRainer RiemannFriedrich-Schiller-UniversitatAlois Angleitner and Frank M. SpinathUniversitat BielefeldPrevious behavior-genetic research on adult personality relied primarily on self-reports or peer reportsthat may be subject to contrast effects, resulting in biased estimates of genetic and environmentalinfluences. In the German Observational Study of Adult Twins (GOSAT), personality traits of 168monozygotic (MZ).and 132 dizygotic (DZ) twin pairs were rated on 35 adjective scales, largely markersof the Big 5. The ratings were provided by 120 judges who never met the twins but observed videotapedbehaviors of 1 twin of each pair in 1 of 15 different settings. The aggregated video-based trait ratingswere highly reliable, and substantial correlations were obtained between MZ as well as DZ twins.Model-fit analyses suggested about 40% genetic, 25% shared environmental, and 35% nonsharedenvironmental influence. Extraversion was the only trait that seemed not to be influenced by sharedenvironment.Numerous behavior-genetic studies suggest that individual dif-ferences in adult personality are almost exclusively accounted forby genetic and nonshared environmental influences. In a meta-analysis of behavior-genetic studies on personality, Loehlin (1992)concluded that additive effects of genes accounted for 22-46% ofthe phenotypic variance, that nonshared environment accountedfor another 44-55%, and that shared environmental influencesPeter Borkenau, Department of Psychology, Martin-Luther-Universitat,Halle, Germany; Rainer Riemann, Department of Psychology, Friedrich-Schiller-Universitat, Jena, Germany; Alois Angleitner and Frank M.Spinath, Department of Psychology, Universitat Bielefeld, Bielefeld,Germany.This article is based on a keynote lecture given by Rainer Riemann at ajoint meeting of the Behavior Genetics Association and the InternationalSociety for the Study of Individual Differences, July 1999, Vancouver,Canada.The research reported in this article was supported by a grant from theGerman Research Foundation (Deutsche Forschungsgemeinschaft).We are indebted to the twins and to the judges for their participation; tothe experimenters, Susanne Hempel, Veronika Koch, Holger Lorenz,Conny Post, Beatrice Rammstedt, Birgit Schlangen, and Robert Weiss, forcollecting the data; to Holger Lorenz and Wolfgang Thiel for their help inthe data analysis; and to Jeremy Miles for power calculations.Correspondence concerning this article should be addressed to PeterBorkenau, Department of Psychology, Martin-Luther-Universitat, D-06099Halle, Germany. Electronic mail may be sent to [email protected] weak, accounting for 0-11% of individual differences inpersonality. More recently, Plomin, DeFries, McClearn, and Rutter(1997) suggested that genes accounted for about 40% of thevariance in personality, that nonshared environment accounted forthe other 60%, and that there were no effects of the sharedenvironment.Whereas it is now generally accepted that genes have a substan-tial influence on individual differences in personality, it is still apuzzle why almost all environmental influences on personalityseem to be of the nonshared variety. Shared environment is definedas environmental factors that contribute to twin and sibling simi-larity, whereas nonshared environment is defined as environmentalfactors that do not contribute to sibling similarity. Sometimes onereads that the nonshared environment makes siblings differentfrom one another, but that definition is ambiguous: Nonsharedenvironmental influence attenuates the otherwise positive correla-tions between all kinds of relatives, but it does not result innegative correlations. If negative correlations between relativesare observed, they indicate contrast effects that make relatives, orthe personality descriptions of relatives, different from oneanother (Heath, Neale, Kessler, Eaves, & Kendler, 1992; Spinath& Angleitner, 1998).Lack of importance of the shared environment is inferred fromthree findings. First, the correlations between adoptive siblings andthose between adoptees and their adoptive parents tend to be small,usually about .05 (Loehlin, Willerman, & Horn, 1987; Plomin,Corley, Caspi, Fulker, & DeFries, 1998; Scarr, Webber, Weinberg,Journal of Personality and Social Psychology, 2001, Vol. 80, No. 4, 655-668Copyright 2001 by the American Psychological Association, Inc. 0022-3514/01/$5.00 DOI: 10.1037//0022-3514.80.4.655655656 BORKENAU, RIEMANN, ANGLEITNER, AND SPINATH& Wittig, 1981). Second, twins reared together are not systemat-ically more similar in personality than are twins reared apart(Loehlin, 1992). Finally, the correlations between monozygotic(MZ) twins tend to be twice or even more than twice as high as arethe correlations between dizygotic (DZ) twins (Loehlin, 1989,1992; Plomin, DeFries, McClearn, & Rutter, 1997), which sug-gests genetic and nonshared environmental but no shared environ-mental influence.Psychology has reacted to these findings in several ways. Someauthors (Rowe, 1994; Harris, 1995, 1998) have suggested theoriesof peer socialization to explain why children in the same family areso different from one another. Other authors have set out todemonstrate that there are shared environmental influences on atleast some traits, such as religious orthodoxy (Beer, Arnold, &Loehlin, 1998). Finally, theorists have suggested that the impor-tance of the shared environment may be systematically underesti-mated in adoption studies, because of range restriction (Stool-miller, 1999), as well as in studies that rely on self-reports orratings by knowledgeable informants, because of contrast effects(Miles & Carey, 1997; Rose, 1995; Saudino & Eaton, 1991). Thepresent article focuses on the latter hypothesis.Problems With Self-Reports inBehavior-Genetic ResearchAlmost all behavior-genetic studies on adult personality rely onself-reports (Brody, 1993). One problem with self-reports inbehavior-genetic research is that they do not allow for appropriateestimates of those sources of measurement error that actuallyattenuate the correlations between relatives. This problem haspreviously been dealt with in two ways. The first is that errorvariance and nonshared environment have not been disentangled,and their combined effects have