The Orphans of Davenport

by Marilyn Brookwood

  McNemar then tore apart the results of Wellman’s 1932 preschool study, Skeels and Wellman’s 1938 Davenport preschool study, Skodak’s 1938 and 1939 adoption studies, along with the Iowans’ other investigations that demonstrated the effects of environment. For reasons that are not clear, McNemar did not attack, or even mention, Skeels and Dye’s 1939 study of Davenport’s orphans at Woodward and Glenwood.

  In his critique of Davenport’s preschool investigation, McNemar claimed that Iowa’s statistical analyses of children’s intelligence were inaccurate because over three years “only a few” of the original children remained in the experimental and control groups due to the adoptions of the others. In fact, eleven of the original twenty-one experimental children left the study because they were adopted. Each time this occurred, another child, similar in age, sex, and IQ, had been substituted. McNemar criticized the inclusion of substituted children, writing that this made it difficult to control variables such as numbers of days of residence and school attendance.

  Because the preschool study occurred in a real-world setting, from its outset the psychologists knew that they might lose subjects to adoption. They noted that it would have been unethical to deny a child the opportunity to be raised in a family. The same problem of attrition by adoption occurred in every orphanage study of that time.26 In Skeels’s study, children whose socialization and intelligence improved as a result of preschool experience became those more likely to be adopted. A perfectly executed study would have denied the opportunity for adoption for the sake of statistical scrupulosity. McNemar, however, turned the substitution of children into a cudgel to damage the Iowans’ scholarly reputations.

  McNemar next faulted the Iowans for their failure to emphasize the factor of rapport, that is, the IQ test examiner’s ability to establish a setting of trust and comfort that enables children to perform at their best. Yet, McNemar supported this claim with an erroneous reference, suggesting that the children’s negative experiences with their cottage matrons led them to mistrust all adults. As Wellman noted, establishing testing rapport with these needy subjects was close to effortless.27

  McNemar then threatened to “explode” what the Iowans identified as a “leveling effect,”28 when long institutional residence with limited human contact and almost no cognitive stimulation appeared to lower children’s intelligence.29 Specifically, they found that after three years, control group children who began with IQ test scores in the range of 65 to 105 ended with scores between 70 and 79. Control group children who initially had IQs of 103, 98, and 73 ended with IQs of 60, 61, and 62 and had to be transferred to an institution for persons with low intelligence. Overall, the preschool children with the longest residence lost a mean of 4.5 points, while control children with the longest residence lost 28.5 points.30 Because McNemar rejected the notion that IQ could change, he attributed these shifts to the surfacing of a child’s heredity and claimed that the Iowans’ findings resulted from their improper substitution of children.31

  In their rebuttal, published in the same journal edition as McNemar’s attack, the Iowans’ reanalyzed their data, including only the original children. Their results were nearly identical.32 Further, McNemar inaccurately reported the amount of time the preschool children spent away from orphanage care as “five or six hours daily,” or 37 to 44 percent of their waking hours. In fact, the children attended preschool for about 8 hours a day, five days per week, or 60 percent of their waking hours each week. McNemar also criticized Marie Skodak’s study of 154 mostly illegitimate children of prostitutes who were adopted before the age of 6 months and after two years had intelligence in the superior range. To support his claims, McNemar asserted that the birth mothers’ and fathers’ educational levels were what they had claimed, in the tenth-grade range, which accounted for the children’s superior intelligence. However, as any educational statistician of that period would have known, and as Skodak’s paper pointed out, at the time a policy known as “non-failing” (actually, automatic promotion) was well established in the United States, and Iowa routinely promoted students based on chronological age.33

  McNemar also incorrectly reported the intelligence of the study’s birth fathers, asserting that their occupational levels, at the time a proxy for male intelligence, were higher than Skodak claimed. In a response, Skodak showed that if the known birth fathers were employed at all, and many had never held a job, those who worked fell into the lowest occupational levels.34 McNemar’s assumptions about the children’s fathers relied on a cultural surmise made by hereditarians, the “smart John” effect. Although that term is a modernism probably not then in use, it suggests that any prostitute’s child with good intelligence—many of Skodak’s subjects were such children—reflected the intelligence of the prostitute’s patron. Providing no evidence, McNemar asserted that 23 percent of the fathers in Skodak’s study were such patrons.35 In a rebuttal, Skodak wrote, “The pregnancy may have followed an assault by a stranger, may have been a chance pick-up . . . or the mother may be unable or unwilling to give information identifying him.”36

  In an outright error, McNemar reported the average IQ for 2-year-old children in Skodak’s study as 108. It was 114. When he challenged the validity of intelligence tests given to 1-year-old children, Skodak recomputed her calculations without those tests. She got the same result. Further, McNemar charged that Skodak had not considered selective placement as a factor in the children’s high IQs. However, from pages 63 to 141 of her paper, she discussed selective placement fifteen times.

  In their rebuttal, the Iowans also noted that McNemar had not read, or had ignored, the key chapter in Skodak’s report “that showed marked gains in IQ following residence in average and above-average adoptive homes.”37 Importantly, only those of McNemar’s readers who closely studied the Iowans’ rebuttals would have been aware of his inaccuracies.

  Another major salvo directed at the Iowa discoveries came from Florence Goodenough. Like McNemar, she chose to publish her paper in another journal, one available in January, weeks before the NSSE meeting. In it she listed nine investigations on the subject of nursery school education and intelligence outcomes, only one of which, she claimed, demonstrated a positive effect.38 Stoddard responded, “Since this is the kind of simple, straightforward statement that is likely to be cited in the years to come, it may be helpful to spend a few minutes in its analysis.”39 He showed that results from seven of the ten preschool studies that Goodenough cited demonstrated gains from 1.8 to 6.6 IQ points. The highest gains came from the University of Iowa’s Laboratory Preschool.

  Both McNemar and Goodenough challenged Skodak’s test results in children under 2 years of age. To answer them, Skodak removed from her data all IQ tests on children who were younger than 2 years. Her results held up. Because the evaluation of very young children was based on the attainment of motor skills rather than verbal responses, any test instrument designed to assess their intelligence naturally aroused skepticism. However, the Iowans, and literally all other 1930s psychologists who evaluated very young children’s intelligence, relied on such tests, notably the Kuhlman-Binet. In 1928, Goodenough herself had published a monograph recommending procedures for that test’s administration.

  Hoping to win support, the Iowa group revised their data according to methods McNemar suggested. In nearly every case, their original findings prevailed. Still not satisfied, McNemar accused them of making deceptive statements. In one example, he challenged their account that “more than one-third of the long-residence control children had final IQs within these limits [50s and 60s], while only one preschool child had a final IQ below 70.”40 McNemar said, “An exact statement would read: Initially four and finally eight of 22 control children were below 70; initially three, and finally one, of the 21 preschool children fell below 70.”41 Both statements were equally true accounts, despite McNemar’s needless complexity and invidious implication.

  McNemar’s slanted language, careless use of material, feigned or true ig
norance of known information, unsupported assertions, outright errors, along with his assumption of statistical superiority, all were known to the Iowans from their spring 1939 reading of his paper. That they could not—head to head—refute such faults before their colleagues undoubtedly drove Stoddard’s introductory comment that the Yearbook was only a “pale” remnant of what it might have been.

  Three Iowa papers answered McNemar’s attack: Stoddard had arranged for the first, a rebuttal by Wellman, Skeels, and Skodak, published in the same journal issue as McNemar’s critical paper;42 and a second, this from Stoddard, was published soon after the St. Louis meeting, in April 1940.43 Then, at a June 1940 meeting of her psychology honor society, Wellman presented a third.44 Taken together these counterarguments identified McNemar’s inaccuracies and included statistical revisions of the Iowa work according to McNemar’s suggestions, almost none of which altered their studies’ findings. Anyone wanting to understand both sides would have had no difficulty locating this information. But in the end, McNemar’s implication of the Iowans’ statistical incompetence aroused damning suspicion of their results, especially among readers who were already skeptical, and therefore Iowa’s rebuttals made almost no impression. McNemar’s paper, a proxy for the views of Lewis Terman and others, eclipsed the Iowans’ evidence and became decisive.

  Although Lewis Terman had not attended the St. Louis meeting, colleagues reported to him about how attendees regarded the Iowans and their supporters. Florence Goodenough remarked that Iowa supporter Paul Witty “had made very much an ass of himself.”45 Her comment was sharp but not specific. Witty had in fact spoken about research that showed African Americans were being disadvantaged in education because of assumptions that their intelligence was low. He also provided evidence that gifted African American children were denied enriched experiences that would allow them to fully develop their abilities.46 Moreover, Witty fiercely indicted the use of intelligence tests, which he said favored those with a “fortunate social heritage,” leading educators to assume that the less fortunate lacked potential.47 The use of the tests, Witty asserted, contradicted the founding ideas of democracy.

  With some pleasure, in a letter to Terman, Goodenough reported that Iowa’s ideas may have generated opposition within their own university. She related her “amusing experience” when the chair of the University of Iowa’s social work department introduced her to a colleague as “the person who had made Iowa infamous.” The social work head assured Goodenough that “our department is on the same side of the fence that you are” and told her “no other departments at Iowa . . . agreed with Stoddard’s and Wellman’s claims.” The social work chair also reported “there was a great deal of feeling . . . that the Research Station had done a good deal to discredit the University.”48

  A second report to Terman came from Benjamin Simpson, who wrote that because Stoddard did not back down when attacked, he was an effective spokesperson—so effective that an Ohio State colleague told Simpson, “George Stoddard is a dangerous man.” Simpson added that Stoddard “is out to win popular support . . . [and] a force to be reckoned with.” Yet the hot-tempered Simpson noticed that attendees maintained control of their responses and he admitted that he had been chastised because “his writings had been too emotional.”49 What neither Goodenough nor Simpson said, but what might have been evident to both, was that by the meeting’s end there was no clear victor in the contest between the environmentalist and hereditarian views of development.

  Only two months after the St. Louis meeting, in April 1940, California educators and Stanford faculty, along with the university’s president, Ray Lyman Wilbur, came together in Cubberly Auditorium for a second Terman-Stoddard debate. For the evening’s topic, the NSSE’s latest Yearbook, Terman arranged to speak for an hour and gave Stoddard 10 minutes. A panel discussion by others would follow. Although the moderator later reported that Stoddard had the audience cheering, Stoddard’s memory of that evening was the wound Terman inflicted on him. “Terman,” he remembered, “pronounced ‘Skeels’ and ‘Skodak’ explosively, lip curled, as if these two young Iowa researchers were a species of insect that had crept under his collar.”50 Stoddard said that he began to think that Terman’s invitation had been a setup to inflict more abuse, that “the performance was rigged . . . I had traveled 2000 miles for my bit part.” The moderator, Stoddard wrote, had confessed to a colleague that he “held Stoddard down while Terman beat him.”51 The two antagonists did not meet again.

  Only limited analysis exists of Terman’s bellicose attacks on Stoddard and his other critics. In one, his biographer, Henry L. Minton, has suggested that certain weaknesses in Terman’s own work may have contributed. Minton wrote, for example, that especially when he was younger, but even into the 1940s, Terman “tended to interpret the data to fit his preconceived notions [and] drew causal conclusions from correlational findings.” Such fundamental research errors would naturally have drawn peer disapproval to someone, Minton observed, who had an “intolerance of accepting personal criticism.” Minton concluded that Terman’s battles with his critics “seemed driven more by a need to prove himself than to settle a scientific issue.”52

  In another view, historian Hamilton Cravens suggested that Terman’s rage was defensive of his, and most psychologists’, social status; that for them, how child development was studied represented

  part of the tacit social and cultural understandings that knit together the social order, and as such . . . had explosive implications for the distribution of status, power, wealth, and justice in society. If the Iowans and their allies were . . . right . . . then the social network of groups and their relative positions . . . could be regarded as cruel, arbitrary, and oppressive.53

  Cravens speculated that Terman understood “only too well the implications of the Iowa work,” and suggested that the Iowans provoked his fury because they challenged assumptions of inherited privilege, that they “crossed some line . . . [and] betrayed their science, their profession, and their social class.”54

  Unbeknownst to Stoddard, in the spring of 1940 he faced further undermining unfolding in the corridors of the Carnegie Institution in Washington. From the station’s research, Stoddard appreciated the importance of early development and had hoped to promote a program of national preschools, something that other nations had already established. In over fifteen articles, he had reported on the economics of bringing his idea to scale: teacher salaries and teacher training; potential benefits for children’s health; positive effects on intelligence; the relation of early education to democratic principles. Alert to opportunities to capture public attention, he even gave advice about how to educate the world-famous Dionne quintuplets, in Ontario, the first quintuplets known to have survived infancy. In 1937, when the quints were 3 years old, Stoddard asked: “Are children today paying the price for the prolonged controversy on heredity versus environment? This ‘case for the quints,’ gives encouragement to continue the quest for the optimum environment.”55

  Stoddard’s advocacy had brought him national attention, and he served on a New Deal emergency committee for the Works Progress Administration that established nursery schools for children from 2 to 6 years old—a total of 3,000 schools in thirty-one states. The program hired thousands of unemployed teachers and supported children’s well-being with nutritious meals—essential during the Great Depression and Dust Bowl—and also guaranteed adequate health care.56 Finding widespread support, Stoddard had proposed to Henry A. Wallace, a longtime Iowa friend who was now Franklin Delano Roosevelt’s secretary of agriculture, that the program might be expanded to serve the entire nation. Wallace agreed.

  But just at the time of the post-NSSE downdraft, in May of 1940, Terman’s student Barbara Burks, at the Carnegie Institution, learned of Wallace’s interest. She told Terman that along with two colleagues, eugenicists Robert Cook and Harold Jones, she thought the plan absurd, and the three hoped to make “at least a dent on Secretary Wallace.”57 To that end, T
erman provided copies of his July 7, 1939, Stanford attack on the Iowa work, which Burks circulated at Carnegie.58 The next week she wrote again, now to say that although Iowa’s media publicity had influenced Wallace and “the Iowa ideology had filtered down to everyone in the Department . . . including the most far-flung field workers,” Burks and her colleagues had convinced Wallace that the Iowa ideas were mistaken. Wallace, she said, was no longer ready to “buy IQ’s . . . to manufacture them by given dosages of nursery school, [or] moron nursemaids.”59

  Later she told Terman that Wallace would certainly have accepted “the Iowa stuff in toto” had Cook, now her fiancé, not intervened and that “it is a safe . . . bet that Iowa-ism would have assumed the proportions of a national movement.”60 It may never be known whether the chilly reception of Iowa’s ideas after the St. Louis meeting influenced Wallace, but in the end, a Lewis Terman confederate subverted George Stoddard’s program for the kind of preschool experience that most industrialized nations now consider essential in children’s development.

  It is reasonable to wonder how the dispute between the Iowans and McNemar might have unfolded had today’s academic peer review requirements then been in place. Peer review, the process by which scientific claims are vetted, requires that prior to journal publication an article’s methods and findings are reviewed by scholars in the same field. While at times these reviews can be partisan, even brutal, in the main they bring more eyes to new material and allow for shared judgments. One can imagine, for example, that a peer reviewer might have flagged McNemar’s unsupported errors and assumptions. Likewise, the Iowans might have been asked to rework some of their data using different statistical methods. But until the post–World War II era, peer review was scattershot, with few established procedures,61 and played no role in the publication process of the Iowans’ work, nor in that of their critics, nor in much of psychology’s academic publishing.