Exploding the “Separated-at-Birth” Twin Study Myth

“The reader whose knowledge of separated twin studies comes only from the secondary accounts provided by textbooks can have little idea of what, in the eyes of the original investigators, constitutes a pair of ‘separated’ twins”—Evolutionary geneticist Richard Lewontin, neurobiologist Steven Rose, and psychologist Leon Kamin in Not in Our Genes, 1984¹

“The Myth of the Separated Identical Twins”—Chapter title in sociologist Howard Taylor’s The IQ Game, 1980²

Supporters of the nature (genetic) side of the “nature versus nurture” debate often cite studies of “reared-apart” or “separated” MZ twin pairs (identical, monozygotic) in support of their positions.³ In this article I present evidence that, in fact, most studied pairs of this type do not qualify as reared-apart or separated twins.

Other than several single-case and small multiple-case reports that have appeared since the 1920s, there have been only six published “twins reared apart” (TRA) studies. (The IQ TRA study by British psychologist Cyril Burt was discredited in the late 1970s on suspicions of fraud, and is no longer part of the TRA study literature.) The authors of these six studies assessed twin resemblance and calculated correlations for “intelligence” (IQ), “personality,” and other aspects of human behavior. In the first three studies—by Horatio Newman and colleagues in 1937 (United States, 29 MZ pairs), James Shields in 1962 (Great Britain, 44 MZ pairs), and Niels Juel-Nielsen in 1965 (Denmark, 12 MZ pairs)—the authors provided over 500 pages of detailed case-history information for the combined 75 MZ pairs they studied.

The three subsequent TRA studies were published in the 1980s and 1990s, and included Thomas J. Bouchard, Jr. and colleagues’ widely cited “Minnesota Study of Twins Reared Apart” (MISTRA), and studies performed in Sweden and Finland. In the Swedish study, the researchers defined twin pairs as “reared apart” if they had been “separated by the age of 11.”⁴ In the Finnish study, the average age at separation was 4.3 years, and 12 of the 30 “reared-apart” MZ pairs were separated between the ages of 6 and 10.⁵ In contrast to the original three studies, the authors of these more recent studies did not provide case-history information for the pairs they investigated. (The MISTRA researchers did publish a few selected case histories, some of which, like the famous “Three Identical Strangers” triplets, had already been publicized in the media.)

The Newman et al. and Shields studies were based on twins who had volunteered to participate after responding to media or researcher appeals to do so in the interest of scientific research. As Leon Kamin and other analysts pointed out long ago, however, TRA studies based on volunteer twins are plagued by similarity biases, in part because twins had to have known of each other’s existence to be able to participate in the study. Like the famous MISTRA “Firefighter Pair,” some twins discovered each other because of their behavioral similarities. The MISTRA researchers arrived at their conclusions in favor of genetics on the basis of a similarity-biased volunteer twin sample.

The three tables I present below are adapted from tables in my 2015 book The Trouble with Twin Studies: A Reassessment of Twin Research in the Social and Behavioral Sciences.⁶ They are based on original-source information found in the above-mentioned three TRA studies whose authors provided detailed case histories.

Table 1 contains information provided by Newman and colleagues for all 19 “reared-apart” MZ pairs they studied, along with some passages from their case descriptions.

Table 2 contains information on all 44 “reared-apart” MZ pairs studied by Shields, accompanied by excerpts from Shields’ case history material for each pair.

Table 3 lists all 12 “reared-apart” MZ pairs studied by Juel-Nielsen, along with excerpts from the detailed case histories he provided. Eight of the 12 pairs were identified through Danish population registries.

The information on these 75 MZ twin pairs can be read in Tables 1-3 pair-by-pair from left to right, but another approach would be to read the right-hand column “From the Case Descriptions” straight down from the top to the bottom of each table (pairs are separated from other pairs by horizontal lines).

Contrary to the common contemporary claim that twin pairs found in TRA studies were “separated at birth”—which should mean that twins did not know each other or interact with each other between their near-birth separation and the time they were reunited for the study—the information provided by the original researchers shows that few if any MZ pairs fit this description. This is even more obvious in the 1962 Shields study. As seen in the tables below and in the case descriptions:

• Some pairs were separated well after birth
• Some pairs grew up nearby to each other and attended school together
• Most pairs grew up in similar cultural and socioeconomic environments
• Many pairs were raised by different members of the same family
• Most pairs had varying degrees of contact while growing up
• Some pairs had a close relationship as adults
• Some pairs were reunited and lived together for periods of time

In other words, in addition to sharing a common prenatal environment and many similar postnatal environmental influences (described here), twin pairs found in volunteer-based TRA study samples were not “separated at birth” in the way that most people understand this term. The best way to describe this sample is to say that it consisted of partially reared-apart MZ twin pairs.

The Minnesota researchers have always denied access to independent researchers who wanted to inspect the unpublished MISTRA raw data and case history information, and we can safely assume that the volunteer MISTRA MZ twin pairs were no more “reared apart” than were the MZ pairs seen in Tables 1-3 below.⁷