Psychology Today

Misleading Treatments: Attempts to Validate the EEA in Twin Research.

By Ken Richardson and Jay Joseph

For those who want to "disentangle" the effects of genes and environments in the development of individual differences, the biological accident of twins seems to present a perfect natural experiment. From the 1920’s to the 1950s, the results of twin research were accepted without much criticism. As claims have become more strident, however, critics have looked more closely at the validity of the method. If we once thought that the twin method was too good to be true, we now know that it is.

As regards the study of separated (“reared-apart”) MZ twins, the problems have been widely recognized (Joseph, 2004, 2010; Sternberg, 2007). Such pairs are relatively scarce; separation turns out to be incomplete (e.g. usually well after birth; merely into different branches of the same family); they remain in regular contact with each other; and there’s the small problem of sharing of intrauterine environment before birth, as well as the epigenetic effects we now know to be so important (Meaney, 2010).

In addition, the behavioral resemblance of reared-apart MZ twin pairs is strongly influenced by non-genetic cohort effects, which account for similarities in people’s behaviors and preferences that arise from the characteristics of the historical periods and cultural milieu in which they experience stages of life at the same time. As Farber (1981, p. 77) observed, reared-apart MZs are “not so much similar to each other as they are similar to people of their eras and SES [socioeconomic statuses].” According to behavior geneticist Richard Rose,

"A colleague suggests that we cannot know [the importance of reared-apart MZ resemblance] without necessary control data on similarities found in pairs of age-matched strangers….Were one to capitalize on cohort effects by sampling unrelated but age-matched pairs, born, say, over a half-century period, the observed similarities in interests, habits, and attitudes might,indeed, be 'astonishing.'” (Rose, 1982, p. 960)

Far from experiencing different environments, as twin researchers often claim, in most cases reared-apart MZ pairs share at least seven different cultural influences at the same time: national, regional, ethnic, religious, economic class, birth cohort, and gender cohort.

So investigators have relied heavily on the reared-together MZ-DZ comparisons, duly called the “classical twin method.” The problems with that approach are just as severe, as detailed in many places (e.g. Joseph, 2004, 2006, 2010; Norgate & Richardson, 2005). But they hinge most critically on what is called the “equal environments assumption” (EEA); twin researchers' claim that reared-together MZ and DZ pairs experience roughly equal environments, and that the greater trait resemblance of MZ versus DZ pairs is caused by genetic factors. This is what we focus on here.

On the face of it, the EEA seems a far-fetched expectation, and is flatly contradicted on a number of empirical grounds. In one recent review, Evans and Martin (2000, p. 77) state that “There is overwhelming evidence that MZ twins are treated more similarly than their DZ counterparts. As children, MZ twins are more likely to have the same playmates, share the same room and dress alike. As adults, MZ twins are more likely to keep in contact, than same-sex DZ twins.” In another review, Joseph (2004) cites questionnaire studies revealing very large differences between MZ and DZ pairs in experiences such as identity confusion (91% vs.10%); being brought up as a unit (72% vs. 19%); being inseparable as children (73% vs.19%); and having an extremely strong level of closeness (65% vs. 19%). It is also known that parents “hold more similar expectations for their MZ than DZ twins with respect to social responsibility and independence” (Scarr & Carter-Saltzman, 1979, p. 528).

Most scientists would have thought this challenge to the classical twin method to be rather devastating. On the contrary, human behavior geneticists have been dismissive of it, citing various studies that, they say, support the EEA and renders the method (and their conclusions about variance attributions) secure. For example, Segal et al., in their commentary on this site, say “numerous twin studies have examined this challenge and have found it to be wanting in most cases.” Likewise, in defending the large TEDS twin study (Twins’ Early Development Study), Kovas et al. (2007) simply say “The equal environments assumption has been tested in several ways and appears reasonable for most traits” (p. 4). For support, most defenders cite the review of Evans & Martin (2000), so we focus on some of the studies reviewed therein.

During the past fifty years, most twin researchers have defended the validity of the EEA and the twin method on the basis of their claim that MZ pairs create more similar environments for themselves compared to DZ pairs because they are more similar genetically. The main supporter of this position has been Kenneth Kendler (1983), and Segal has also defended the twin method on this basis: “It is important to note that if MZ twins are treated more alike than DZ twins, it is most likely associated with their genetically based behavioral similarities” (Segal & Johnson, 2009, p. 82).

However, proponents of this “twins create their own environment” position use circular reasoning in that they assume the very thing they need to demonstrate. In other words, modern twin researchers circularly rely on the twin method to validate the twin method, and in the process they assume that twins’ behavioral resemblance is caused by genetics, in order to circularly conclude the very same thing (Joseph, 2010). Many researchers performing EEA test studies also defend the validity of the EEA on the basis of circular arguments of this type (Joseph, 2006).

The first test of the EEA was that of Scarr (1968) applied to cognitive ability and personality tests. It compared the effects of perceived zygosity, as reported by twins and their parents, and true zygosity, as assessed by the researchers, on score differences. Basically, these were more in accord with true than perceived zygosities. Scarr concluded that parental treatment is more a function of true zygosity than their personal beliefs, and that it is not important in producing twin differences. Of course, the limitations of using perceived zygosity as a “measure” of similar environments in general, for all traits, has been pointed out by several critics. This study also suffers from small samples, there being only eleven misidentified pairs; and parents who persist with incorrect zygosity classification may not be typical of parents of twins.

A similar approach was used by Scarr and Carter-Saltzman (1979). Twins’ perceived zygosities and similarities of appearance were explored through questions such as, “Do you and your twin look as alike as carbon copies?” Other people’s perceptions of twins’ resemblances were estimated from ratings of black and white photographs of faces by eight independent psychologists. The question was whether co-twins were found to resemble each other on four different cognitive tests according to their true or perceived zygosities, so indicating whether different (assumed) treatment was important.

The authors concluded that it wasn’t. But what is revealing about this study is the riskiness of using either perceived zygosity or ratings of physical similarity as guides to treatment. Twins’ judgments of their own zygosity correlated only 0.55 with self-ratings of resemblance. Also, ratings of twins’ zygosity by psychologists using only black and white photographs turned out to be more accurate than those of the twins themselves (correlation of .66 with blood group differences compared with only .57). The psychologists’ ratings and twins’ judgments of similarity also correlated only moderately at .51. Thus, twins’ perceptions of their similarity are not very reliable guides to either their mutual resemblance, or to others’ perceptions of their resemblance (and, hence, others’ treatment effects).

Perhaps it is not surprising, therefore, that Scarr and Carter-Saltzman reported “a highly inconsistent pattern of results across tests” (p. 532). For the MZ twins, it was found that “The Raven Matrices score differences were significantly predicted by differences in appearance as rated by eight psychologists” (p. 535) – a strong negation of the EEA in that differences in appearance (presumed to lead to differences in treatment) lead to differences in test scores. For the sample as a whole, larger differences in Raven’s scores were associated with smaller differences in appearance ratings (see Richardson & Norgate, 2005, for a fuller assessment).

Kendler (1993) reported that similarity of treatment did not actually make twins more similar in personality, vocational interests and cognitive abilities. How did he show this? By identifying a number of twins’ parents who were mistaken about the twins’ zygosities and then showing that this made no difference to the development of one or other of five psychiatric disorders. Again, though, use of perceived zygosity as a measure of trait-relevant equal environments and treatments is very limited. It may be the case that parents’ current perceptions of their twins’ resemblances may influence recall of their approach to child-rearing. And the direction of effect is unknown: similar treatment of twins may influence twin resemblance, or vice versa (Mitchell et al, 2007). Finally, the claim that treatment has no effect on twins' personalities must generalize to the untenable position that treatment has no effect on anyone's personality.

Segal et al. refer to the “landmark” study of Loehlin and Nichols (1976) based on twins’ scores on the National Merit Scholarship Test. These authors assumed that similarity of treatment can be measured from mailed self-reporting questionnaires. But it is well known that such instruments have weaknesses of biased impressions, poor memory, and poor reliabilities of only 0.3–0.5 (see Richardson & Norgate, 2005, for references). These problems were clearly reflected in the pattern of results. Loehlin and Nichols’s parent questionnaire contained 56 treatment related items to which parents could respond: (1) “Twin one only,” (2) “Twin two only,” (3) “Both twins,” and (4) “Neither twin.” It appears that parents treated their twins remarkably alike, reporting that they had treated (1) their male MZ twins the same on 99.14% of the questions, (2) their female MZ twins the same on 98.98% of the questions, (3) their male DZ twins the same on 97.44% of the questions, and (4) their female DZ twins the same on 97.7% of the questions. Combining all responses, parents answered “neither twin” or “both twins” 98.3% of the time (Loehlin & Nichols, 1976, p. 55. Percentages calculated from Table 5-8). According to Loehlin and Nichols (p. 56), “it was exceptional for a parent to mark any treatment item as different for the twins; for the identicals it was, in fact, positively rare.” Clearly, rather than accurate memory, parents answered questions on the basis of an idealized image of their child rearing practices (Joseph, 2004). Loehlin and Nichols (1976) themselves wrote that their results do “not altogether exclude a completely environmentalist position,” and show that it “operates in remarkably mysterious ways” (p. 94).This highly flawed study hardly constitutes a valid defense of the EEA.

Another study usually cited in support of the EEA is that of Matheny et al. (1976). They reported no association between differences of appearance and IQ among their sample of three to thirteen year old twins. Again, though, scores of physical similarity were obtained from parents’ responses to a ten item questionnaire (including facial appearance, height, whether or not the twins were ever confused for one another, etc.). Responses were then added together to create a 20-point scale of twin similarity. From the design, acute restrictions of range were inevitable. Indeed, 90% of MZs had similarity scores ranging from 12 to 20, while 80% of the DZs had scores ranging from 0 to 8. A rank order correlation was used, but with large numbers of pairs ranked on scores from 0 to 8 or 12 to 20, there are, inevitably, many tied ranks, which further reduces the possibility of significant association.

These studies are frequently cited in support of the EEA. However, they are flawed in various respects. They have the common flaw of poor assessment of the factor in question, namely equality/differences of environments relevant to the traits in question. Apart from the unreliability of questionnaire reports, it seems naïve even to suppose that answers to a few such questions added together could capture multifaceted, interactive, nonlinear environments, equally relevant to all psychological traits. As Mitchell et al. (2007, p. 844) say, “Measures of the similarity of twins’ environments have inherent limitations, due to the richness and diversity of the construct of environmental influences.”

The validity of the EEA is, of course, implicitly tested in other ways that behavior geneticists tend to overlook. Claims about the presence of additive, independently deterministic, genes producing direct genotype-phenotype correlations, arising from twin studies, has inspired genome-wide association (GWA) searches for such genes at the molecular level. Decades of searches, however, have failed to discover them for the major psychiatric disorders (Gershon, Alliey-Rodriguez, & Liu, 2011), for personality traits (Verweij et al., 2010), or for IQ (Deary, Penke, & Johnson, 2010). This provides additional evidence that standard interpretations of twin studies, and the EEA, are incorrect, that such simple genes do not exist, and that we need to consider the roles of non-genetic and “systemic” factors in individual differences.

Indeed, psychologists are still unsure of the environments causally related to differences in mental traits, and the traditional behavior genetic approach has also failed to identify them, just as it has failed (through the GWA studies it has inspired) to identify specific genes. In commenting on this, Turkheimer (2011) says, “individual differences in complex human characteristics do not, in general, have causes, neither genetic nor environmental,” and that they “are not going to be found in individual genes or environments.” Instead he suggests that “some new paradigm, unglimpsed at present, will be required” (p.600). One of us has suggested that such a paradigm has been around for some time, and has emerged in the form of dynamic systems theory, through which complex systems, such as human psychology, make and change their own differences in the process of lifelong development (Richardson, 2010). Designing a true test of the EEA in such circumstances seems truly formidable.

So we are genuinely concerned that investigators persist in defending a key assumption of classical twin studies as if it had been amply supported. Major decisions are made about the apportionment of variances to genetic or environmental sources, on the basis of twin correlations, when they may be completely wrong. Not only that, but other more exotic inferences – such as environments being "genetic," or children making their own environments according to their genes – are drawn and described as "surprising," as if the initial variance attributions, and methods expressing them, were unquestionable. Until the twin method is validated more thoroughly, the potential for misleading scientists, policy makers, and the general public may be enormous.

References

Deary, I. J., Penke, L., & Johnson, W. (2010). The neuroscience of human intelligence differences. Nature Reviews Neuroscience 11, 201-211.

Evans, D. M., & Martin, N. G. (2000). The validity of twin studies. GeneScreen, 1, 77–79.

Farber, S. L. (1981). Identical twins reared apart: A reanalysis. New York: Basic Books.

Gershon, E. S., Alliey-Rodriguez, N., & Liu, C. (2011). After GWAS: Searching for genetic risk for schizophrenia and bipolar disorder. American Journal of Psychiatry, 168, 253-256.

Joseph, J. (2004). The gene illusion: Genetic research in psychiatry and psychology under the microscope. New York: Algora. (2003 United Kingdom Edition by PCCS Books)

Joseph, J. (2006). The missing gene: Psychiatry, heredity, and the fruitless search for genes. New York: Algora.

Joseph, J. (2010). Genetic research in pshychology and psychiatry: a critical overview. In K.E.Hood, C.T Halpern, G. Greenberg & Lerner, R.M. (Eds.). Handbook of developmental science, behavior and genetics (pp. 557-625). New York: Wiley-Blackwell.

Kendler, K. S. (1983). Overview: A current perspective on twin studies of schizophrenia. American Journal of Psychiatry, 140, 1413-1425.

Kendler, K.S., Neale, M.C., Kessler, R.C., Heath, A.C. & Eaves, L.J. (1993). A test of the equal-environment assumption in twin studies of psychiatric illness. Behavior Genetics, 23, 21-27.

Kovas, Y., Claire M.A. Haworth, C.M.A., Dale, P.S. & Plomin, R. (2007). The genetic and environmental origins of learning abilities and disabilities in the early school years. Monographs of the Society for Research in Child Development, 72, vii-160.

Loehlin, J. C., & Nichols, R. C. (1976). Heredity, environment and personality. Austin: University of Texas Press.

Meaney, M.J. (2010). Epigenetics and the biological definition of gene × environment Interactions. Child Development, 81, 41-79.

Mitchell, K.S., Mazzeo, S.E., Bulik, C.M., Aggen, S.H., Kendler, K.S. & Neale, M.C. (2007). An Investigation of a Measure of Twins’ Equal Environments. Twin Research and Human Genetics,10, 840–847.

Richardson, K. & Norgate, S. (2005). The equal environments assumption of classical twin studies may not hold. British Journal of Educational Psychology, 75, 1–13.

Richardson, K. (2010). The evolution of intelligent systems: how molecules became minds. Basinstoke: Palgrave Macmillan.
Rose, R. J. (1982). Separated twins: Data and their limits. Science, 215, 959-960.

Segal, N. L., & Johnson, W. (2009). Twin studies of general mental ability. In Y. Kim (Ed.), Handbook of behavior genetics (pp. 81-99). New York: Springer.

Scarr S. (1968). Environmental bias in twin studies. Eugenics Quarterly, 15, 34-40.

Sternberg, R. J. (2007). Critical thinking in psychology is really critical. In R. Sternberg, H. Roediger III, & D. Halpern (Eds.), Critical thinking in psychology (pp. 289-296). New York: Cambridge University Press.

Turkheimer, E. (2011). Commentary: variation and causation in the environment and genome. International Journal of Epidemiology, 40, 598–601. doi:10.1093/ije/dyq147

Verweij, K. J. H., Zietsch, B. P., Medland, S. E., Gordon, S. D., Benyamin, B., Nyholt, D. R., McEvoy, B. P., Sullivan, P. F., Heath, A. C., Madden, P. A. F., Henders, A. K., Montgomery, G. W., Martin, N. G., & Wray, N. R. (2010). A genome-wide association study of Cloninger’s temperament scales: Implications for the evolutionary genetics of personality. Biological Psychology, 85, 306-317.

The fact that studies of MZA twins have shown remarkably similar results in relation to the relative contributions of genetics and environment to IQ (and other traits) somewhat shows that twin studies are useful.

Yemen, One of the earth's weakest and very inferior america's, Provides an opportune redoubt. traveling aircraft remain weak. assistant is actually strange sunday remark that the protection equipment functioned affected person airline ticket 253 doesn't promote self confidence.