A Critical Analysis of IQ Studies of Adopted Children

The pattern of parent-child correlations in adoption studies has long been interpreted to suggest substantial additive genetic variance underlying variance in IQ. The studies have frequently been criticized on methodological grounds, but those criticisms have not refl ected recent perspectives in genetics and developmental theory. Here we apply those perspectives to recent IQ adoption studies and show how they further question two sets of problems: fi rst, the assumption of additive gene and environmental effects; second, the assumption that the adoption situation approximates a randomizedeff ects design. We show how a number of possible factors having systematic eff ects in breach of those assumptions can produce the received pattern of correlations without appealing to unusual amounts of additive gene variance. Copyright © 2006 S. Karger AG, Basel Studying adopted children has long been viewed as a well-controlled, quasi-experimental method for estimating genetic and environmental components of variation in human intelligence which can then be used in models of human development. Because biological mothers and their adopted away children share genes but not environments, covariance (correlation) in traits like IQ is considered to be a direct estimate of genetic eff ects on trait variance. Because adoptive caregivers and their adopted children share environments but not genes, IQ correlations between them can indicate the magnitude of environmental sources of variance. Correlations between unrelated children reared in the same home are taken to indicate what proportion of the environmental variance is ‘shared’ (varies between families) or ‘non-shared’ (varies within families). For a review of all these aspects of behaviour genetic methods see Plomin, DeFries, McClearn and Rutter [1997]. Sarah H. Norgate, Psychology Directorate Faculty of Health and Social Care, University of Salford Allerton Building, Frederick Rd. Campus Salford, M6 6PU (UK), Tel. +44 161 295 2324 Fax +44 161 295 2427, E-Mail [email protected] © 2006 S. Karger AG, Basel 0018–716X/06/0496–0319$23.50/0 Accessible online at: www.karger.com/hde Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com Human Development 2006;49:319–335 320 Richardson /Norgate Since the 1920s a now-familiar pattern of results has emerged from such studies: IQs of adopted children appear to correlate more with those of biological mothers than with adoptive mothers or adoptive fathers. Table 1 shows typical data from two of the more recent studies: the fi rst sweep of the Texas Adoption Project (TAP1) as reported by Horn, Loehlin and Willerman [1979], and the fi rst Minnesota Adoption Study (MAS1, a transracial study) as reported by Scarr and Carter-Saltzman [1989], involving children covering a wide age range, from 3 to 18 years. Th e seemingly diminishing correlations between adopted children and their adoptive parents in later years are shown in table 2 (results from TAP follow-up aft er about 10 years and second MAS study of a new sample of youths, aged 16–22 years). Th e more recent Colorado Adoption Project (CAP) replicated this general picture, as shown in table 3 [Plomin, Fulker, Corley, & DeFries, 1997]. Th is study included a control group of normal families matched to the adoptive families for a number of demographic factors. As can be seen, the correlation between the adopted children and their biological parents almost doubled, from less than 0.2 to nearly 0.4 in the period when the children were 14–16 years old, while the adoptive parents-adopted children correlation remained around 0. Finally, all of these studies report IQ correlations between adopted children and other children in the same family signifi cantly lower than those between natural siblings, even though (it is assumed) they are sharing the same environment. Table 1. Adoptive child’s IQ correlations with parents Biological father Biological mother Adoptive father Adoptive mother TAP1a – 0.31 0.14 0.12 MAS1b 0.43c 0.33c 0.27 0.21 Dash indicates that no correlation was obtained. a Mean age 8 years (range 3–16+) in the TAP; b mean age 7 years (range 4–18 years) in the MAS; c biological parents not tested, IQs were estimated from education levels. Table 2. Adopted child-adoptive parent correlations in the TAP2 and MAS2 studies Adoptive father Adoptive mother TAP2a 0.10 0.02 MAS2b 0.16 0.09 a Mean age 18 years (means of various test combinations, see table 4.3 in Loehlin et al., 1997). b Age range 16–22 years.