The Fatal Flaws in the Minnesota Twin Study: or Why Nurture is More Powerful than Nature

A new book, Born Together-Reared Apart by Nancy L. Segal and improbably published by Harvard Press is perpetuating the bad science and fatal flaws of the most influential study purporting to prove that genes cause personality. The authors of the Minnesota Twin Study (“Sources of Human Psychological Differences: The Minnesota Study of Twins Reared Apart” [Science, 12 October, p. 223] Bouchard et al.) report that 70% of the variance in IQ in a sample of monozygotic twins “reared apart,” was “associated with genetic variation.” They also attribute similarities in a host of other personality traits to nature, not nurture. However, these findings are thoroughly undercut by the authors’ problematic assumptions, which they never discuss. When the authors’ conceptual and methodological presuppositions are scrutinized, it becomes clear that the authors’ findings more convincingly support the diametrically opposed conclusion that early childhood experiences in fact exert strong effects on IQ scores and psychological test scores.
Most damaging to the study’s findings, the integrity of the research rests on the startling premise that the formative years of childhood occur after age four. This dubious assumption allows the authors to characterize their sample as twins “separated very early in life [and] reared apart during their formative years,” in spite of the fact that the twins lived together after birth for a mean period of 5.1 months with a standard deviation of 8.5 months, and two twins characterized as “reared apart” remained together until they were over four years of age!
But the four years after birth is obviously a formative period (see e.g., Smart Love: The Comprehensive Guide to Understanding, Regulating, and Enjoying Your Child; The Scientist in the Crib). To give but one illustration, by five months sufficient learning and relationship pleasure will have transpired that the child will know and prefer the face of the important persons in her or his life. And either at this point or within a month or two, the introduction of a stranger’s face can cause the infant to break into tears. The fact that the IQ scores of twins who have remained together for a mean period of five months, and for as long as four years, are strikingly similar can be used more compellingly to argue for the powerful and lasting effects of early childhood parenting on IQ scores than to demonstrate the power of hereditary influences. In other words, this study begs the question by assuming what it purports to establish, namely, that the first months and years of life are not formative.
The definition of the twins in the sample as “reared apart” is also inaccurate due to the variability of the length of time the twins remained apart once separated. Unfortunately, the distribution is not provided, but at the low end of the range, two of the twins were apart only half a year before their first reunion.
In summary, as a result of the authors’ mislabeling of the experimental sample of twins as “reared apart during their formative years,” the samples of “reared apart” and “reared together” twins do not differ significantly in their exposure to the influences of their birth families and of each other. Therefore, it is arguable that the most likely explanation for the findings of “remarkable similarity” between samples of twins characterized as “reared apart” and “reared together” is that the twins in both samples shared formative early experiences.
The authors’ description of the experimental sample of twins as “reared apart” also is misleading in that it implies that, once separated, the twins had little contact with each other until testing occurred. However, regardless of the age at which they were separated initially, the twins labeled “reared apart” spent considerable time together before assessment took place. The mean total contact time was over two years (112.5 weeks) with a standard deviation of 4.4 years (230.7 weeks), and the twins with the most contact time were together for 23.7 years (1233 weeks). The mean time apart before assessment was only 30 years, in spite of the fact that the mean age of the twins at assessment was 40 years. Thus, even if the authors’ improbable premise that the formative years of life do not begin until age four is accepted, most of the twins actually had significant contact during later years. The “striking similarity” between the IQ scores and psychological tests of the two samples of twins can be explained not only by the contact the twins in the sample designated “reared apart” had during infancy and early childhood, but also by the contact that occurred between these twins after infancy but prior to testing.
Like the authors’ general method of determining the relative contributions of environmental and hereditary “sources of human psychological differences,” the authors’ conclusion that the twins’ years of contact with each other after childhood did not affect the twins’ similarity in scores rests on the shaky, yet unexamined, methodological assumption that the analysis of variance is equivalent to the analysis of causality. But the authors never discuss the devastating critiques of this use of analysis of variance, especially as found in the seminal, directly relevant article, “Analysis of Variance and Analysis of Causes,” by Richard Lewontin.
We would note in passing that a further detraction from the findings is that, in spite of the extensive literature critiquing this practice, the authors naively identify IQ scores with “general intelligence.” In the very least, the authors should discuss and attempt to rebut the arguments that have been levied against this questionable equation.
Perhaps of most concern, while the authors report their findings with the caveat that they should not be used to “detract from the value or importance of parenting, education, and other propaedeutic interventions,” they must have been aware that their conclusions would easily lend themselves to this destructive purpose. For example, in the Science editorial written to accompany this article, Koshland uses the study as a springboard to assert that when confronted with “homicidal maniacs,” society has wrongheadedly made “a diligent search of their childhoods, as though understanding their upbringing and circumstances would explain their aberrant action.” To be sure, Koshland inserts qualifications in an attempt to avoid the ideological taint of making hereditary rather than environmental influences hegemonic in the development of cognitive and psychological traits, but the conflict between his qualifications and his view of childhood experience as nonexplanatory is neither addressed nor resolved. Reports of this study in the media have uncritically drawn the same conclusions about the relative inefficacy of environmental inputs and the relative strength of hereditary factors. In the current climate of scarce resources and the American predilection to blame the victim, we can easily imagine the funding cuts for ameliorative social, clinical and educational services that will result from this line of reasoning. In this way, the study’s conceptual and methodological flaws not only vitiate its results, but will also be the cause of significant social harm.