Studies of Reared-Apart (Separated) Twins: Facts and Fallacies

Twin studies supply the most frequently cited evidence in favor of important genetic influences on human behavioral differences. Examples of human behavior include IQ, personality, socially disapproved behavior such as criminality, political behavior and ideology, and psychiatric disorders. Almost all twin studies are based on classical twin method comparisons between MZ (monozygotic, identical) and DZ (dizygotic, fraternal) twin pairs reared together in the same family home. MZ pairs are said to share 100% of their segregating genes, whereas (like ordinary siblings) DZ pairs are said to share only 50% on average.¹

In an extremely small yet influential handful of studies, twin pairs were said to have been reared apart in different families. Twin researchers and others view this occurrence as the ultimate test of the relative influences of nature (genes) and nurture (environment). According to this view all behavioral resemblance between reared-apart MZ twin pairs (known as “MZA” pairs) must be the result of their 100% genetic similarity, because such pairs share no environmental similarity. The most celebrated and widely cited investigation is the “Minnesota Study of Twins Reared Apart” (MISTRA), carried out by behavioral geneticist Thomas J. Bouchard, Jr. and his colleagues at the University of Minnesota, whose most frequently cited publications appeared in the 1980s and 1990s.² In a 2012 book describing and defending the study, entitled Born Together—Reared Apart: The Landmark Minnesota Twin Study, MISTRA researcher Nancy Segal endorsed the view that MZA pairs share no environmental similarities: “The MZA intraclass correlation of 0.50 for a personality trait directly estimates the heritability of the trait because MZA twins share only their genes.”³

The authors of the six “twins reared-apart” (or “TRA”) studies calculated mean correlations in a sample of MZA pairs, and sometimes also studied reared-apart DZ pairs (known as “DZA” pairs). These studies focused mainly on IQ, personality, and other behavioral characteristics. Apart from anecdotal single case reports, there are too few pairs of this type for use in psychiatric twin research, which must rely on twin method comparisons of reared-together twin pairs.⁴

Prior to the publication of the Minnesota studies, three “classical” TRA studies were published between 1937 and 1965 by Horatio Newman and colleagues in the United States, James Shields in Great Britain, and Niels Juel-Nielsen in Denmark.⁵ These studies included detailed case history information for most studied pairs. An IQ study by British psychologist Cyril Burt was discredited by the late 1970s on suspicion of fraud. Despite the acclaim TRA studies have received (other than Burt’s study), critics have argued that they contain many biases and depend on many questionable assumptions. Some critics also have questioned the validity of core psychometric/behavioral genetic concepts such as “IQ,” “personality,” “heritability,” and “model fitting” statistical formulas.⁶ These issues are covered in detail in my 2015 book, The Trouble with Twin Studies: A Reassessment of Twin Research in the Social and Behavioral Sciences.⁷ The following is a partial list of TRA study problems and biases as discussed by the critics:

• Many twin pairs experienced late separation, and many pairs were reared together in the same home for several years
• Most twin pairs grew up in similar socioeconomic and cultural environments
• MZA correlations were inflated by non-genetic cohort effects, based on common age, common sex, and other factors
• Twins share a common pre-natal (intrauterine) environment, and the MZA pre-natal environment is more similar than the DZA pre-natal environment
• TRA study findings might not be (or are not) generalizable to the non-twin population
• In studies based on volunteer twins, a bias was introduced because pairs had to have known of each other’s existence to be able to participate in the study
• MZA samples were biased in favor of more similar pairs, meaning that studied MZA pairs are not representative of MZAs as a population
• The similar physical appearance and level of attractiveness of MZAs will elicit more similar behavior-influencing treatment by their social environments
• Twins sometimes had financial and other types of incentives to exaggerate or lie about their degree of separation and behavioral similarity, and their accounts are not always reliable
• There were several questionable or false assumptions underlying the statistical procedures used in the studies
• MZA pairs were not assigned to random environments
• There was researcher bias in favor of genetic interpretations of the data
• There were problems with the IQ and personality tests used
• The validity of concepts such as IQ, personality, and heritability are questionable
• In cases where evaluations and testing were performed by the same person, there was a potential for experimenter bias in favor of twin similarity
• A registry should be established to house raw TRA study data, which should be made available for independent inspection and analysis

Although the authors of many authoritative social and behavioral science texts have accepted the original researchers’ claims that these were studies of “reared-apart” twin pairs, most MZA pairs were reared together for periods of time, had frequent or regular contact, and/or had a close emotional bond with each other. In Shields’ 1962 study of 44 MZA pairs, for example, twins separated as late as age 9-years-old, or for only 5 years during childhood, were counted as “separated twin pairs.” According to Shields, “Without exception [MZAs] were brought up in different homes for at least five years during childhood.”⁸ A pair “living next door to each other, brought up by different aunts” was also counted by Shields as a “separated” twin pair. ⁹Two of many instances of twins’ emotional closeness include the members of a Shields MZA pair who “have always been closely attached to one another,” and another pair who “formed an extremely close association.”¹⁰ Each of the case histories in Juel-Nielsen’s study of 12 MZA pairs contained a section called “The Twin Relationship.” This should not be found in a study of “reared-apart” twins, where the common perception is that twins were separated at birth and had never met, and therefore had no relationship with each other.

TRA researchers usually assume that above-zero MZA IQ or personality test score correlations are the result of genetic influences. However, although theoretically pairs are raised in different family environments, as critics have noted, MZA behavioral resemblance can result from the fact that pairs are the same age and sex, are very similar in physical appearance, and usually grow up in very similar cultural and socioeconomic environments in the same historical eras. Collectively, these are known as “cohort effects.” Travelers to several Middle Eastern countries, for example, will encounter the remarkable (non-genetic) behavioral similarity that most women wear head covering, even though almost all were “reared apart” from each other in different family environments.

Far from being separated at birth and reared apart in randomly selected homes representing the full range of potential behavior-influencing environments, and meeting each other for the first time when studied, most MZA pairs were only partially reared apart, and grew up in similar cultural and socioeconomic environments at the same time.¹¹

The Minnesota (MISTRA) researchers concluded that their “landmark” study decisively confirmed the existence of strong genetic influences on IQ, personality, and behavior in general.¹² However, in addition to the TRA research problems and biases described by the earlier commentators, the Minnesota researchers’ conclusions in favor of genetics were based on many questionable assumptions. The researchers themselves recognized that some of these assumptions are “likely not to hold,”¹³ and are “generally oversimplifications of the actual situation, and their violation can introduce systematic distortions in the estimates.”¹⁴ Despite the researchers’ attempts to improve TRA study procedures, it is unlikely that the MISTRA volunteer twin sample achieved higher levels of separation and environmental dissimilarity when compared with the earlier classical studies.

Importantly, the MISTRA researchers decided to bypass the critical first step of TRA studies using MZA and DZA pairs, which requires a preliminary determination that the MZA correlation is significantly higher than the DZA correlation.¹⁵ According to Segal, “The simple comparison of the MZ (or MZA) and DZ (or DZA) intraclass correlations is an important first step in behavioral-genetic analysis because this demonstrates whether or not there is genetic influence on the trait.”¹⁶ Because MZA pairs are more similar genetically than are DZA pairs, a mean MZA sample correlation not higher than the corresponding DZA sample correlation at a statistically significant level suggests that non-genetic (environmental) factors alone are responsible for raising both correlations above zero.

And yet, in practice, Segal, Bouchard and their colleagues decided against making this preliminary “important first step” determination, and based their conclusions on behavioral genetic “model-fitting” statistical analyses, which depend on the clearly false assumption that “all resemblance between reared apart relatives is because of genetic factors.”¹⁷ In Born Together—Reared Apart, Segal wrote that the MISTRA personality studies found “that personality similarity between relatives seems to come mostly from their shared genes,”¹⁸ a conclusion arrived at circularly on the basis of the MISTRA assumption that, as Segal wrote elsewhere in her book, “shared genes underlie similarity between relatives.”¹⁹

Although rarely mentioned in (usually positive) academic and popular accounts of the study, the Minnesota researchers failed to publish and evaluate their full-sample mean DZA IQ score correlations, even though they reported full-sample DZA correlations for non-IQ behavioral characteristics in various MISTRA publications.²⁰ This occurred despite the fact that IQ (cognitive ability) was the main focus area of the study and of the field of behavioral genetics in general, and that the researchers had designated DZA twins as the MISTRA control group.²¹ Bouchard, Segal, and colleagues placed more emphasis on the importance of their DZA control group correlations at an earlier stage of the study in 1986, writing, “DZA twins allow us to test the two most common competing hypotheses proposed as alternatives to the genetic hypothesis as an explanation of the similarity between MZA twins.”²² Yet by the time of their first major peer-reviewed publication in 1988, with the data at hand, the researchers decided against performing the preliminary MZA versus DZA first step “test” of these “competing hypotheses.”

When encountering and commenting on MISTRA DZA correlations that did not fit their genetic models, the researchers developed various explanations that were consistent with these models. When they chose to publish it, the researchers had a genetic explanation for their DZA control group correlation regardless of its value—an approach that guarantees that genetic explanations will prevail. As the psychologist Ken Richardson pointed out, behavioral genetic researchers make “certain assumptions about ‘what to expect’ in the patterns of scores, and adjusted their analytical equations accordingly: not surprisingly, that pattern emerges!”²⁵

Nevertheless, based on the incomplete and unanalyzed DZA IQ correlations that have been published, there does not appear to be a statistically significant difference between the MISTRA MZA and DZA groups on either the Wechsler (WAIS) IQ test (MZA correlation = .62, versus DZA = .50), or the Raven Progressive Matrices IQ test (MZA correlation =.55, versus DZA = .42).²⁶ The failure to find a significant difference between the MZA and DZA groups runs counter to genetic predictions and theories, and suggests that environmental factors alone account for both the MZA and the DZA above-zero IQ correlations.²⁷

Another important issue has been the Minnesota researchers’ decision to deny access to critically minded reviewers seeking to perform independent inspection and analysis of the MISTRA raw data.²⁸ Unlike the authors of the classical studies, for the most part the MISTRA researchers failed to include case histories for the pairs they studied, and also failed to provide basic information and test score results for each pair. Granting access to independent reviewers is especially important in this case because, unlike other types of human behavioral research, TRA studies are extremely difficult to reproduce due to the rarity of reared-apart twin pairs. In Born Together—Reared Apart, Segal wrote that the “burden of proof lies with the critics” to show that the study contained invalidating biases. “Bias must be demonstrated, not assumed,” she wrote.²⁹ This has placed critics in a difficult “Catch-22”-like position because, if they were known to be inclined to look for bias, the Minnesota researchers denied them access to the raw data.

A committee of Dutch psychologists assessing previous and potential fraud in their discipline concluded in 2012, “It must always remain possible for the conclusions to be traced back to the original data. Journals should only accept articles if the data concerned has been made accessible in this way.”³⁰ In direct contrast, the MISTRA researchers have kept their original data secret, thereby rendering it inaccessible.

Although his subsequent work with the MISTRA suggested a shift in his views, twin researcher Irving Gottesman evaluated the TRA study approach in 1982 as follows:

After a quarter century of experience with twins reared together and twins reared apart, it is my conviction that twins reared apart are a wonderful source of hypothesis generation, but not a useful source for hypothesis testing.³¹

Twins reared-apart studies have certainly helped generate some interesting hypotheses, but they have completely failed to provide scientifically acceptable evidence in support of genetic influences on human behavioral differences, which include IQ, personality, and psychiatric disorders—a conclusion consistent with the ongoing decades-old failure to uncover genes for behavioral characteristics at the molecular genetic level. As Bouchard acknowledged in 2014, “In spite of numerous studies with sufficient power to detect rather small effects,” the results of attempts to uncover genes for general intelligence (g) “have been dismal in comparison with expectation.”³²

Adding the general problems of twins reared-apart studies to the unique flaws and biases of the MISTRA leads to the conclusion that genetic interpretations of the MISTRA and all other TRA study IQ and personality results are invalid, and should be removed from the scientific record at least until complete access to the raw data is granted, and independent critical analyses are performed and published.

* * * * *

References:

1. Research performed in the 21rst century has called into question the basic twin study assumption that both members of an MZA pair are genetically identical throughout their lives. See Charney, E. (2012), Behavior Genetics and Postgenomics, Behavioral and Brain Sciences, 35, 331-358.

2. The most cited MISTRA study is Bouchard, T. J., Jr., Lykken, D. T., McGue, M., Segal, N. L., & Tellegen, A. (1990), Sources of Human Psychological Differences: The Minnesota Study of Twins Reared Apart, Science, 250, 223-228.

3. Segal, N. L. (2012), Born Together—Reared Apart: The Landmark Minnesota Twin Study, Cambridge, MA: Harvard University Press, p. 334. The MISTRA researchers did attempt, albeit inadequately, to correct correlations for the confounding effects of common age and common sex.

4. The only twins reared-apart study that assessed psychiatric disorders was a MISTRA study of substance abuse/dependence and antisocial personality. See Grove, W. M., Eckert, E. D., Heston, L. L., Bouchard, T. J., Jr., Segal, N. L., & Lykken, D. T. (1990), Heritability of Substance Abuse and Antisocial Behavior: A Study of Monozygotic Twins Reared Apart, Biological Psychiatry, 27, 1293-1304.

5. Juel-Nielsen, N. (1965/1980), Individual and Environment: Monozygotic Twins Reared Apart (rev. ed.), New York: International Universities Press; Newman, H. H., Freeman, F. N., & Holzinger, K. J. (1937), Twins: A Study of Heredity and Environment, Chicago: The University of Chicago Press; Shields, J. (1962), Monozygotic Twins Brought up Apart and Brought up Together, London: Oxford University Press. The other two TRA studies were the Swedish Adoption/Twin Study of Aging (SATSA), and a study from Finland.

6. Critical works include Farber, S. L. (1981), Identical Twins Reared Apart: A Reanalysis, New York: Basic Books; Joseph, J. (2004), The Gene Illusion: Genetic Research in Psychiatry and Psychology under the Microscope, New York: Algora; Joseph, J. (2015), The Trouble with Twin Studies: A Reassessment of Twin Research in the Social and Behavioral Sciences, New York: Routledge; Kamin, L. J. (1974), The Science and Politics of I.Q., Potomac, MD: Erlbaum; Kamin, L. J., & Goldberger, A. S. (2002), Twin Studies in Behavioral Research: A Skeptical View, Theoretical Population Biology, 61, 83-95; Lewontin, R. C., Rose, S., & Kamin, L. J. (1984), Not in Our Genes, New York: Pantheon; Richardson, K. (1998), The Origins of Human Potential, London: Routledge; Taylor, H. F. (1980), The IQ Game: A Methodological Inquiry into the Heredity-Environment Controversy, New Brunswick, NJ: Rutgers University Press. 

7. Joseph, 2015.

8. Shields, 1962, p. 27.

9. Shields, 1962, p. 48.

10. Shields, 1962, p. 165, 50.

11. See Joseph, 2015, Chapter 2, and Tables 2.1, 2.2, & 2.3. See also Farber, 1981.

12. Bouchard et al., 1990; Segal, 2012.

13. McGue, M., & Bouchard, T. J., Jr. (1989), Genetic and Environmental Determinants of Information Processing and Special Mental Abilities: A Twin Analysis, In R. Sternberg (Ed.), Advances in the Psychology of Human Intelligence (Vol. 5, pp. 7-45), Hillsdale, NJ: Erlbaum, p. 23.

14. Johnson, W., Bouchard, T. J., Jr., McGue, M., Segal, N. L., Tellegen, A., Keyes, M., & Gottesman, I. I. (2007), Genetic and Environmental Influences on the Verbal-Perceptual-Image Rotation (VPR) Model of the Structure of Mental Abilities in the Minnesota Study of Twins Reared Apart, Intelligence, 35, 542-562, p. 548.

15. See Joseph, 2015 for a detailed analysis of this point.

16. Segal, 2012, p. 62.

17. McGue & Bouchard, 1989, p. 22.

18. Segal, 2012, p. 102.

19. Segal, 2012, p. 63.

20. For example, compare the famous 1990 Bouchard et al. Science publication, where the researchers reached strong conclusions about genetic influences on IQ but where no DZA correlations were reported, with a concurrent 1990 MISTRA study in which the researchers published their full-sample control group DZA mean personality correlation (.18). See Bouchard, T. J., Jr., & McGue, M. (1990), Genetic and Rearing Environmental Influences on Adult Personality: An Analysis of Adopted Twins Reared Apart, Journal of Personality, 58, 263-292, p. 278.

21. Segal, 2012, p. 12 (see also Footnote 69 on Segal’s p. 343).

22.Bouchard, T. J., Jr., Lykken, D. T., Segal, N. L., & Wilcox, K. J. (1986), Development in Twins Reared Apart: A Test of the Chronogenetic Hypothesis, in A. Demirjian (Ed.), Human Growth: A Multidisciplinary Review (pp. 299-310), London: Taylor & Francis, p. 300.

23. The first major peer reviewed MISTRA publication was the 1988 personality study by Tellegen and colleagues. See Tellegen, A., Lykken, D. T., Bouchard, T. J., Jr., Wilcox, K. J., Segal, N. L., & Rich, S. (1988), Personality Similarity in Twins Reared Apart and Together, Journal of Personality and Social Psychology, 54, 1031-1039.

24. Joseph, 2015, Chapters 5 & 6.

25. Richardson, 1998, p. 121.

26. The Wechsler (WAIS) IQ correlations are taken from Segal, 2012, p. 286, based on unpublished figures given to her by Bouchard. Segal did not state the number of MZA and DZA twin pairs, but the full MISTRA sample consisted of 81 MZA and 56 DZA pairs. The Raven IQ correlations are taken from Johnson et al., 2007, p. 552, Table 3, and were based on 74 MZA and 52 DZA pairs. The VassarStats website provides a test of statistical significance between two independent sample correlation coefficients. This test shows that both the MISTRA Wechsler and Raven MZA and DZA correlations fail to differ below the conventional .05 level of statistical significance, meaning that the difference between these correlations is assumed to have occurred by chance (the null hypothesis stating that the correlations do not differ is not rejected).

27. See Joseph, 2015, Chapter 6.

28. See Joseph, 2015, Chapter 5.

29. Segal, 2012, p. 299.

30. Levelt Committee, Noort Committee, Drenth Committee, (2012), Flawed Science: The Fraudulent Research Practices of Social Psychologist Diederik Stapel, p. 58.

31. Gottesman, I. I. (1982), [Review of the Book Identical Twins Reared Apart: A Reanalysis, by S. Farber], American Journal of Psychology, 95, 350-352, p. 351.

32. Bouchard, T. J., Jr. (2014), Genes, Evolution and Intelligence, Behavior Genetics, published online March 7, 2014. DOI 10.1007/s10519-014-9646-x