Transracial adoption studies are a powerful methodology for examining the relative contribution of genetics and environment to average differences in IQ across racial and ethnic groups. The best — and most famous — transracial adoption study was the Minnesota Transracial Adoption Study (MTAS) conducted by Sandra Scarr and her colleagues (Scarr & Weinberg, 1976; Weinberg et al., 1992). That study produced ambiguous results that supported both environmentalist and hereditarian hypothesis for the origin of IQ differences (Waldman et al., 1994).
In the 21st century, a different transracial adoption study has garnered interest from environmentalists: Elsie Moore’s study of black children adopted by black and white parents. Comparing Moore’s study to the MTAS, Nisbett (2009, p. 198) calls the former “A superior adoption study . . .” A few years later, Nisbett and a group of environmentalist coauthors called it, “One of the most telling studies” (Nisbett et al., 2012, p. 146). Other authors have cited the study approvingly as showing the cultural differences were more important than genetic differences in creating IQ gaps (e.g., Dickens, 2005; Flynn, 2019).
Citations to Moore’s work show that hereditarians are aware of the study, but they have done little analysis of it. Lee (2010) compared the average IQs of the children adopted by white parents in Moore’s studies to children in MTAS and Eyferth study. Rushton and Jensen (2005) summarized the study and merely commented that a later follow-up in the late teen years would be informative. Even when I have cited the study, I have done little beyond acknowledging its existence (Warne, 2019, 2021; Warne et al., 2018). Moore’s study is long overdue for in-depth examination.
Elsie Gloria Jean Moore Smith was born on November 16, 1949, and earned her PhD on August 29, 1980, from the University of Chicago. According to an obituary, Moore joined the faculty at Arizona State University (ASU) in 1981 and was affiliated with the university for over 40 years. By 1987, she was an associate professor in the Department of Counselor Education at ASU. Her university obituary refers to “Her vast body of research,” but Google Scholar has indexed only 9 articles, 1 (co-authored) book, and 1 book chapter that she published in her lengthy career. She seems to have spent most of her time on administrative duties. Moore passed away on February 21, 2022, at the age of 72.
Moore reported her results in her dissertation (Moore, 1980), a book chapter (Moore, 1985), and two journal articles (Moore, 1986, 1987). Moore’s dissertation is quite good; she reported a wide variety of data about biological mothers and adoptive parents, background variables for the adopted children, adoptive mothers’ childrearing values, the home environment, and the children’s problem-solving behavior. It is a study and worthy of attention.
Generally, the study as reported in the dissertation conforms with the reports published later. Changes tend to be minor or clearly typos in one document or another. For example, in the interpretation of the finding that the adoptive black children’s IQ scores scored were positively correlated with the proportion of their friend group that consisted of white children. In the dissertation, Moore calls this finding “Surprising” (1980, p. 337), but when this portion of the study was reported several years later, there was no trace of surprise in the report of this finding (Moore, 1987). All in all, that is an unimportant change. The information that matters — methodology, tables, analyses, results — are consistent (or easily reconcilable) across Moore’s publications.
Moore (1980, 1985, 1986, 1987) reported data from 46 adopted black children, half of whom were adopted by black parents (“traditionally adopted children”) and half of whom were adopted by white parents (“transracially adopted children”). All of the children lived in 2-parent, middle-class families in the Chicago area. At an average age of 8.61 years (range = 7 to 10), the traditionally adopted children had IQs of 103.6 (SD = 11.4), and the transracially adopted children had an average IQ of 117.1 (SD = 9.4), a difference of 13.5 points. With this difference being nearly as large the average difference between black and white Americans in the general population (about 15 points), some commentators have seen Moore’s study as showing that cultural differences explain most of the black-white difference in IQ (e.g., Flynn, 2018; Nisbett, 2009; Nisbett et al., 2012).
Moore (1980) interpreted this mean IQ difference between the two groups of adoptees as arising from cultural differences and childhood experiences. She heavily emphasized the black and white mothers’ differing childrearing goals, the mothers’ behavior when their child attempted to solve a difficult cognitive task in their presence, and the social milieu of the different families. Moore believed that there was no fundamental difference in the intelligence of the traditionally and transracially adopted children. Therefore, any differences would be due to the environment that the different families provided for their adopted children.
Moore (1980, 1987) believed that some of these differences were captured by traditional measures of socioeconomic status. However, she recognized (correctly) that they did not fully explain the 13.5-IQ point mean difference in her data. To explain this, she proposed that traditional metrics of socioeconomic status did not capture all of the important differences among families (Moore, 1980) and that the tests did not fully measure the capabilities of black children because the tests defined cognitive competence in white terms (Moore, 1985). To support this belief, Moore identified behaviors and values that differed between white and black mothers and attributed a causal influence of these variables on IQ. For example, when their child was solving a difficult problem, black mothers were more likely to make negative comments about their child’s mistakes, whereas white mothers made more positive remarks and were more encouraging (Moore, 1980, 1986).
The first point that is striking about the mean IQs in Moore’s study is how high they are. This is easily explained by her use of the original Wechsler Intelligence Scale for Children (WISC), published in 1949. Moore collected her data between 1977 and 1979, meaning that the Flynn effect had about 29 years to inflate the scores. Flynn (1984, Table 2) estimated that the WISC scores increased by an average of .343 points per year during that time. Therefore, the scores in Moore’s study were inflated by about 9.947 points. Applying this correction to Moore’s data results in average IQs of 93.7 for traditionally adopted black children and 107.2 for the transracially adopted children. Moore (1985, p. 109) was not aware of the Flynn effect’s impact on her data. Consequentially, her study does not show that black children raised in middle class black homes score as well as the general white population. However, this Flynn effect correction does not reduce the 13.5-IQ point difference between traditionally and transracially adopted children.
One of the most frequently overlooked pieces of data from Moore’s transracial adoption study is that she reported mean IQs of biracial children separately from black children, both for the entire sample and within the traditionally and transracially adopted groups (Moore, 1980, p. 130; 1985, p. 111; 1986, p. 320). Adoptees with two black biological parents had an average IQ of 108.1 (SD = 10.13, adjusted mean = 98.2), and adoptees with one black and one white biological parent had an average IQ of 113.2 (SD = 8.25, adjusted mean = 103.3). This is a 5.1-IQ point difference favoring the biracial children. Waldman et al. (1994) commented on this, stating that this difference was not statistically significant. That inference is true when based on a two-tailed test (p = .074). The hereditarian hypothesis, however, is directional (stating that biracial children should have higher scores than black children), making a one-tailed test appropriate. The p-value for this test is statistically significant (p = .037), though with the small sample size (n = 26 black children and 20 biracial children) and barely significant p-value, this is not strong support for the hereditarian hypothesis.
Digging deeper, there is something strange about these mean IQ scores. A 13.5-IQ point difference between adoptive groups is surprisingly large — only 1.5 points smaller than the typical black-white IQ difference. In other words, the differences between the home environments created by the white and black adoptive parents almost fully reproduced the IQ gap found in the general population. Comparing this difference to IQ differences in the MTAS provides important context. In the initial MTAS report (Scarr & Weinberg, 1976), there was an 10.8 IQ-point difference between black adoptees and the biological children of the white adoptive parents. Also in MTAS, the IQ difference between all black/interracial adoptees and white adoptees is 5.2 IQ points. These differences in mean IQ among groups with similar adoptive environments are much smaller than what is seen in Moore’s study. Why on earth is the difference in IQ between two groups of adopted black children so large?
Moore (1980) provides a lot of data about the adopted children’s background and their adoptive family environment. Examining these only deepens the mystery because the white and black families are similar on so many variables. All the families are middle class and live in the Chicago area, and the children had almost the exact same average age at testing. The parents had similar levels of education. (White fathers had more education that black fathers, but white mothers had less education than black mothers.) Children from both groups had similar Apgar scores and birth weights, and all attended preschool. Both groups of children had a similar probability of attending private schools. The academic achievement at the schools was similar, though slightly higher for the children adopted by white parents before grade 2 (Moore, 1980).
Where there were differences, they sometimes favored the traditionally adopted children raised by black parents. The black families had a higher income (mostly due to the white adoptive mothers being less likely to work) and smaller families. Traditionally adopted black children attended preschool for longer, had healthier biological mothers who were older at the time of birth, and were adopted earlier with fewer pre-adoption placements (Moore, 1980).
That being said, it is important to recognize that the white adoptive fathers did have more prestigious jobs, and the white adoptive mothers scored higher on the Information subtest of the Wechsler Adult Intelligence Scale than the black adoptive mothers. The white parents were also more civically involved and gave their children more opportunities for extracurricular activities and play outside of the home. The transracially adopted black children also had peer groups that were more white than the traditionally adopted children had (Moore, 1980, 1987).
Taken together, these families and the environments they provided are much more similar than black and white families in the general population. Moore recognized this fact:
Clearly, the two groups of adoptive parents in this sample are more similar on those variables which researchers have found to be related to children’s I.Q. scores than is found between black and white people in the larger populationMoore (1980, p. 216)
And yet, despite their similarities, the average IQ difference among the white and black parents’ (black) children is almost as large as what is found nationwide. This is a genuine puzzle that shows that this study’s data are anomalous.
One clue comes from Moore’s report that at age 3.5 to 4, the children took the Peabody Picture Vocabulary Test (PPVT) as part of an earlier study on children in different adoption placement situations. The PPVT scores and WISC scores correlated r = .64 (Moore, 1980, p. 136). In line with this positive correlation is a substantial mean IQ difference between the groups during at preschool age. Transracially adopted children had an average IQ of 111.3 points, while the traditionally adopted children had an average IQ of 92.9 points. This is a difference of 18.4 IQ points on the PPVT. If anything, this shows that the IQ gap between the two groups had narrowed from age 4 to 8.5. Clearly, the large IQ gaps are not due to a data collection artifact when the children were studied by Moore, because she had nothing to do with the PPVT testing years earlier.
I found the original report of the PPVT scores, and the means are different. In that report (Shireman & Johnson, 1980, p. 27), the mean for the transracially adopted children was 105.09, and the mean for the traditionally adopted children was 97.56, which is a difference of only 7.53 points. This discrepancy is almost certainly due to attrition. In the earlier report, there were 42 transracially adopted children and 45 traditionally adopted children, resulting in a total sample size of 87. Four years later, Moore found 23 in each group who would participate, giving an overall attrition rate of 47.1%. Apparently, attrition in the study when the children were about age 8 was related to their IQ. I believe that this shows that Moore’s sample is not representative of black adopted children — let alone the general population. Her study is not generalizable.
Confounded Independent Variables
Moore openly admitted in all her relevant publications that placement with families was not random. Disproportionately, the adoptive black parents expressed a wish to have a child who resembled them in appearance, whereas the white adoptive parents did not have a preference for the appearance of their child (Moore, 1980, p. 94; 1985, p. 107; 1986, pp. 318-319; 1987, p. 47). As a result, 6 of the 23 traditionally adopted children (26.1%) were biracial, whereas 14 of the 23 transracially adopted children (60.9%) were biracial. This difference is statistically significant (chi-squared = 5.66, df = 1, p = .017, phi = .351). As a result, a child’s race (black vs. biracial) and the race of their adoptive parents (black vs. white) are highly confounded. Therefore, any simply univariate comparisons of IQs — whether that is comparing adoptive groups or comparing black to biracial children — are uninformative. Thus, neither the 13.5-IQ point difference between adoptive groups, nor the 5.1-IQ point difference between racial groups in Moore’s study is conclusive evidence for either the environmentalist and hereditarian hypotheses.
Moore did not understand the severity of the confounding. She never conducted the chi-squared test to determine whether the child’s race and their adoptive parents’ race were correlated. Indeed, she incorrectly stated that her design controlled for genetic factors:
Because all of the children included in the sample were socially-classified black children, it was also assumed that potential genetic sources of variance were controlled. With these controls established in the sampling procedure, then the influence of the ethnicity of the rearing environment on any observed differences in the performance of the two groups of children could be assessed.Moore (1980, p. 322; see also 1985, p. 106)
Poor Statistical Choices
Perhaps because she did not grasp that her two most important independent variables were confounded, Moore did not conduct any multiple regression analyses, which would attempt to control for one independent variable while showing the strength of the other. As a result all her analyses are uninformative and cannot shed light on whether genes or environment are more important for explaining the average IQ difference between American racial groups.
Additionally, Moore (1980, 1986) relied heavily on multivariate analysis of variance (MANOVA) to examine the relationship between background variables (including IQ) and maternal behavior and the child’s problem-solving behaviors. There were a few problems with this choice. First, the dependent variables in the MANOVAs are supposed to all be manifested variables of the same underlying latent variable (Warne, 2014). Moore never justifies that the dependent variables she lumps together in her MANOVA are all measuring the same construct. Indeed, it is highly doubtful that they are. For example, in the maternal behavior MANOVAs, Moore combines approval, positive tension (i.e., laughing, joking), positive evaluation (i.e., indicating that the child has performed well), enthusiasm, and giving hints into the same MANOVA. These are definitely not different manifestations of the same behavior (Moore, 1986, p. 324).
Moore’s lack of adeptness with MANOVA also includes her treatment of independent variables. Moore routinely dichotomized continuous variables (especially age, IQ, and the adoptive mother’s years of education), thereby reducing variance and making it harder for these to obtain statistical significance.
As if these errors were not severe enough, Moore (1980, 1986) follows-up all of her statistically significant MANOVAs with MANOVAs that include only a single statistically significant independent variable. This is the incorrect post hoc procedure; she should have used a descriptive discriminant analysis (Warne, 2014). Because of a series of errors, none of Moore’s MANOVAs provide useful information regarding her — or any — hypotheses.
Moore’s transracial adoption study has received attention from both hereditarians and environmentalists. That attention is deserved. Moore’s study is still the second-largest transracial adoption study, and she collected a wide variety of data on adoptive children, their biological mothers, their adoptive parents, their problem-solving behaviors, and their neighborhoods. Her work is very impressive, and I admire her for tackling a challenging problem.
Despite all her work, the study has little citation value today. The results cannot generalize to any meaningful wider population, and the children’s racial heritage is confounded by the race of the adoptive parents. Moore’s statistical analyses are inherently flawed, and their results are useless.
I wish Moore had reported her raw data in her dissertation, but the recruitment letter sent to the adoptive parents promised that no one outside of the research staff would have access to the data (Moore, 1980, pp. 346-347). If she had published the data as an appendix, it would have been an ethical breach. That is unfortunate, because a re-analysis of her data could mitigate some of the flaws in her research. Likewise, it is disappointing that she never conducted another follow-up of the children and measured their intelligence in adolescence or adulthood. It is another example of how environmentalists have squandered forty years instead of conducting better research.
Dickens, W. T. (2005). Genetic differences and school readiness. The Future of Children, 15(1), 55-69. https://doi.org/10.1353/foc.2005.0003
Flynn, J. R. (1984). The mean IQ of Americans: Massive gains 1932 to 1978. Psychological Bulletin, 95(1), 29-51. https://doi.org/10.1037/0033-2909.95.1.29
Flynn, J. R. (2018). Academic freedom and race: You ought not to believe what you think may be true. Journal of Criminal Justice, 59, 127-131. https://doi.org/10.1016/j.jcrimjus.2017.05.010
Flynn, J. R. (2019). Reservations about Rushton. Psych, 1(1), 35-43. https://doi.org/10.3390/psych1010003
Lee, J. J. (2010). Review of Intelligence and how to get it: Why schools and cultures count, R. E. Nisbett, Norton, New York, NY (2009). ISBN: 9780393065053. Personality and Individual Differences, 48(2), 247-255. https://doi.org/10.1016/j.paid.2009.09.015
Moore, E. G. J. (1980). The effects of cultural style on black children’s intelligence test achievement [Unpublished doctoral dissertation]. University of Chicago.
Moore, E. G. J. (1985). Ethnicity as a variable in child development. In M. B. Spencer, G. K. Brookins, & W. R. Allen (Eds.), Beginnings: The social and affective development of black children (pp. 101-115). Erlbaum.
Moore, E. G. J. (1986). Family socialization and the IQ test performance of traditionally and transracially adopted Black children. Developmental Psychology, 22(3), 317-326. https://doi.org/10.1037/0012-16126.96.36.1997
Moore, E. G. J. (1987). Ethnic social milieu and black children’s intelligence test achievement. The Journal of Negro Education, 56(1), 44-52. https://doi.org/10.2307/2295382
Nisbett, R. E. (2009). Intelligence and how to get it: Why schools and cultures count. W. W. Norton & Company.
Nisbett, R. E., Aronson, J., Blair, C., Dickens, W., Flynn, J., Halpern, D. F., & Turkheimer, E. (2012). Intelligence: New findings and theoretical developments. American Psychologist, 67(2), 130-159. https://doi.org/10.1037/a0026699
Rushton, J. P., & Jensen, A. R. (2005). Thirty years of research on race differences in cognitive ability. Psychology, Public Policy, and Law, 11(2), 235-294. https://doi.org/10.1037/1076-89188.8.131.52
Scarr, S., & Weinberg, R. A. (1976). IQ test performance of Black children adopted by White families. American Psychologist, 31(10), 726-739. https://doi.org/10.1037/0003-066x.31.10.726
Shireman, J. F., & Johnson, P. R. (1980). Adoption: Three alternatives (ED197811). ERIC. https://eric.ed.gov/?id=ED197811
Waldman, I. D., Weinberg, R. A., & Scarr, S. (1994). Racial-group differences in IQ in the Minnesota Transracial Adoption Study: A reply to Levin and Lynn. Intelligence, 19(1), 29-44. https://doi.org/10.1016/0160-2896(94)90051-5
Warne, R. T. (2014). A primer on multivariate analysis of variance (MANOVA) for behavioral scientists. Practical Assessment, Research & Evaluation, 19(17), 1-10. http://pareonline.net/getvn.asp?v=19&n=17
Warne, R. T. (2019). An evaluation (and vindication?) of Lewis Terman: What the father of gifted education can teach the 21st century. Gifted Child Quarterly, 63(1), 3-21. https://doi.org/10.1177/0016986218799433
Warne, R. T. (2021). Between-group mean differences in intelligence in the United States are >0% genetically caused: Five converging lines of evidence. The American Journal of Psychology, 134(4), 480-501. https://doi.org/10.5406/amerjpsyc.134.4.0479
Warne, R. T., Astle, M. C., & Hill, J. C. (2018). What do undergraduates learn about human intelligence? An analysis of introductory psychology textbooks. Archives of Scientific Psychology, 6(1), 32-50. https://doi.org/10.1037/arc0000038
Weinberg, R. A., Scarr, S., & Waldman, I. D. (1992). The Minnesota Transracial Adoption Study: A follow-up of IQ test performance at adolescence. Intelligence, 16(1), 117-135. https://doi.org/10.1016/0160-2896(92)90028-p