In the classic twin design, estimation of genetic andenvironmental effects is based on the assumptionthat environmental influences are shared to thesame extent by monozygotic and dizygotic twins(equal environment assumption, EEA). We explorethe conditions in which the EEA can be testedbased on multivariate phenotypic data. We focus onthe test whether the correlation between sharedenvironmental factors in dizygotic twins (rC) is lessthan 1. First, model identification was investigatedanalytically in Maple and Mx. Second, statisticalpower was examined in Mx. Third, the amount ofbias caused by violation of the EEA was evaluated.Finally, applications to empirical data concern spatialability in adolescents and aggression in children.Bivariate and trivariate models include severalinstances in which the EEA can be tested. Thenumber of twin pairs that is needed to detect viola-tion of the EEA with a statistical power of .80(α = .05) varied between 508 and 3576 pairs for thesituations considered. The bias in parameter esti-mates, given misspecification, ranged from 5% to34% for additive genetic effects, and from 4% to34% for shared environmental effects. Estimates ofthe nonshared environmental effects were notbiased. The EEA was not violated for spatial abilityor aggression. Multivariate data provide sufficientinformation to test the validity of the EEA. Thenumber of twin pairs that is needed is no greaterthan the number typically available in most twin reg-istries. The analysis of spatial ability and aggressionindicated no detectable violation of the EEA.Genetic and environmental contributions to individ-ual differences in phenotypic traits can be estimatedusing genetically informative data. In the case of twindata, the estimates are based on the monozygotic(MZ) and dizygotic (DZ) twin variances and covari-ances. With such data, a number of competing modelscan be considered. These models may include additivegenetic effects (A), dominant genetic effects (D),shared environmental effects (C), and nonshared envi-ronmental effects (E). Estimates of these effects in thetwin design are based, inter alia, on the followingthree assumptions: (i) the additive (dominant) geneticeffects correlate 1 (1) in MZ twin pairs and .5 (.25) inDZ twin pairs, (ii) the shared environmental effectscorrelate 1 in both MZ and DZ twin pairs, and (iii)the nonshared environmental effects do not correlatein MZ and DZ twin pairs. In this article, we explorethe possibilities of testing the second assumption,usually referred to as the equal environment assump-tion (EEA), given bivariate or trivariate data. TheEEA implies that the shared environmental influencesare equally important in MZ and DZ twin pairs. Thevalidity of the EEA has been debated (Faraone &Biederman, 2000; Joseph, 2000).It has been shown that MZ twins in childhoodmore often share playmates, share the same room,and dress more alike than same-sex DZ twins(Loehlin & Nichols, 1976). However, this does notnecessarily imply that the EEA is violated. First, thegreater environmental similarity in MZ than DZtwins does not have to be related to a greater pheno-typic similarity. Second, even if a greaterenvironmental similarity is related to a greater pheno-typic similarity, this association could be mediated bya greater genetic similarity in MZ than DZ twins(Scarr & Carter-Saltzman, 1979). The EEA is onlyviolated when the correlation between environmentalsimilarity and trait similarity is significantly greaterthan zero within zygosity groups. Eaves et al. (2003)concluded on the basis of simulation studies that theabsence of any association between environmentalsimilarity and trait similarity justifies the claim thatenvironmental similarity is not a factor in twin resem-blance. However, the counterclaim that the presenceof an association between environmental similarityand trait similarity falsifies the EEA is unfounded.403Twin Research and Human Genetics Volume 9 Number 3 pp. 403–411A Test of the Equal EnvironmentAssumption (EEA) in Multivariate Twin StudiesEske M. Derks,1Conor V. Dolan,2and Dorret I. Boomsma11 Department of Biological Psychology,Vrije Universiteit,Amsterdam, the Netherlands2 Department of Psychology, University of Amsterdam,Amsterdam, the NetherlandsReceived 16 December, 2005; accepted 27 January, 2006.Address for correspondence: E. M. Derks, Vrije Universiteit,Department of Biological Psychology, Van der Boechorststraat 1,1081 BT Amsterdam, the Netherlands. E-mail: [email protected] methods have been proposed to detect vio-lation of the EEA. One way of assessing the validity ofthe EEA is to see whether within zygosity groups, dif-ferences in environmental experiences are associatedwith differences in phenotypic traits (Loehlin &Nichols, 1976). The influence of several environmen-tal experiences have been studied, including ratings oftreatment similarity (Loehlin & Nichols, 1976; Roweet al., 2002; Wade et al., 2003), physical similarity(Hettema et al., 1995), and frequency of contact asadults (Kendler, Karkowski, et al., 2000). Loehlin andNichols (1976) reported low correlations between dif-ferential treatment scores and mental ability,personality traits, vocational interests, and interper-sonal relationships within zygosity groups. The rangeof correlations was –.15 to +.22, which is about whatone would expect on the basis of change fluctuation.This suggests that the EEA is not violated for a widevariety of phenotypic traits. The validity of the EEAwas also confirmed by more recent studies, whichincluded measures of the influence of specified envi-ronmental measures on psychiatric diseases, such asmajor depression, generalized anxiety disorder, panicdisorder, phobias, posttraumatic stress disorder, nico-tine dependence, marijuana dependence, alcoholdependence, and psychoactive substance use andabuse (Hettema et al., 1995; Kendler & Gardner,1998; Kendler, Karkowski, et al., 2000; Xian et al.,2000). The EEA has also been found to be tenable forsexual orientation (Kendler, Thornton, et al., 2000). Incontrast, possible violation of the EEA was reportedfor smoking initiation (Kendler & Gardner, 1998),and bulimia (Hettema et al., 1995; Kendler &Gardner, 1998; Rowe et al., 2002).Another test for the validity of the EEA is providedby twins whose genetic similarity is misperceived bythemselves and others (Scarr & Carter-Saltzman,1979). If beliefs about zygosity determine the extent towhich the two members of a twin pair are behav-iorally
View Full Document
Unlocking...