I am now going on my fifth year blogging. In that time, my views have considerably shifted to what I would term HBD racial realism (reductionism of the Neo-Darwinian type which is refuted by a holistic perspective of the organism) to a more holistic, systems approach of the organism and how it interacts with its environment—the gene-environment system.
Many long-time readers may know that I used to be a staunch hereditarian especially when it came to IQ. However, back in the Spring of 2017, I read DNA is Not Destiny (Heine, 2017) and Genes, Brains, and Human Potential (Richardson, 2017a) (in the same month, no less). Heine had me questioning my views while Richardson completely changed them. I would say that the biggest catalysts were chapters 4 and 5 on genes, what they are and how they work in concert with the physiological system were imperative to my view changes. Further, learning more about the history of IQ testing also further lead to these view changes. (See my article “Why Did I Change My Views?” for more information.)
This then leads me to someone on Twitter by the name of “ModernHeresy” who, back in October, asked me which books best represent my views on IQ:
I replied, Genes, Brains, and Human Potential (Ken Richardson), On Intelligence (Stephen Ceci) and Inventing Intelligence (Elaine Castles). He then said that he thinks that Jensen et al are right about IQ, but that he will give Richardson’s book an honest chance. Well, I was heavily biased against anti-hereditarian arguments before I read Richardson’s book almost 3 years ago, and now look at me.
In any case, ModernHeresy (MH) had responded to some of Richardson’s arguments in his latest book in a video titled “A Critique of Ken Richardson: Initial Impressions and Social Class“. It seems like a well-researched video with four topics that I will also cover today. MH covers Goddard’s use of the Binet-Simon scales in turning away prospective immigrants who scored lower; the construct validity argument; IQ as a measure of social class; and IQ ‘predicts’ only through test construction. I will respond to each point per section.
Goddard was the man who translated Binet’s original test and brought it to America, translating it to English in 1910. He was the director of the Vineland Training School of Feebleminded Boys and Girls in Vineland, New Jersey and he believed that one’s intellectual potential was biologically determined. Goddard used his translated-Binet to attempt to turn away those who he deemed “feebleminded” or “morons” (indeed, he was the one to coin the term; see Castles, 2012; Wilson, 2017; Dolmage, 2018). Goddard is of Kallikak family fame—a pseudonymous name for a family of “feebleminded people”, see Smith and Wehmeyer (2014) for an exposition on how Goddard was wrong about the Kallikaks and telling Deoborah Kallikaks true identity. To Goddard’s credit, though, he did recant some of his views in 1928 stating that “feeblemindedness” was not incurable, as he once thought.
MH then cites Snyderman and Herrnstein (1983) stating that they “thoroughly review the congressional record and testimony is almost no evidence that intelligence tests had any influence over the content or the passage of the 1924 immigration act.” MH then goes on to say that the claim that IQ testing had anything to do with the 1924 immigration act had its roots in the 70s, specifically in Leon Kamin’s The Science and Politics of IQ, which Gould then reiterated in both versions of Mismeasure of Man. (See here for a defense from Kamin and also see Dorfman.) MH then says that
Richardson’s book was published in 2017 this is completely inexcusable and I would argue an indication that Richardson’s work has a lot of its roots and arguments that originated in the 1970s and the formulation of these arguments have basically ignored or at best extremely selectively referenced any work in the almost 50 years since that have challenged them.
This is ridiculous. Snyderman and Herrnstein did nothing of the sort. Gelb et al (1986) write:
The historical record clearly documents that mental testing played a part in the national immigration debate between 1921 and 1924, though certainly in a less direct manner than Snyderman and Herrnstein purportedly sought to uncover.
In their distorted and simplistic account of the period, Snyderman and Herrnstein failed to account for the interconnections between psychometric, eugenic and political communities. While some historians of psychology have exxagerated the influence of the mental testers on the passage of the Immigration Act of 1924, Snyderman and Herrnstein’s attempt to exonerate the early testers contains flaws at least as serious as any of those they criticize. Important mental testers of the 1910s and 1920s were willing to use their fledgling science to promote immigration restriction. One cannot examine the relevant historical material without concluding that prominent testers promoted eugenic and racist interests and sought to, and in some degree succeeded in, providing those interests with a mantle of scientific respectability.
While Ford (1985) writes that “If the long-standing acceptance of racial, ethnic, and sexual bias with intellectual circles prior to 1924 is considered, Snyderman and Herrnstein’s conclusion becomes invalid.” We know that there is racial, ethnic, and sex bias which are built into the test to get the score distributions the researchers want (Mensh and Mensh, 1991; Hilliard, 2012).
Dolmage (2018: 119) states that “Whenever [Henry Laughlin] testified [to the U.S. Congress], he brought charts, graphs, pedigree charts, and the results of hundreds of IQ tests as evidence of “the immigrant menace. Laughlin plastered the Congress committee room with charts and graphs showing ethnic differences in rates of institutionalization for various degenerative conditions, and he presented data about the mental and physical inferiority of recent immigrant groups.” So, IQ tests were, quite clearly, used to stifle immigration from Eastern and Southern Europe (though this was not specifically on Goddard, this was due directly to his bringing the Binet-Simon test to America and translating it into English).
MH then cites Richardson’s (2002) paper What IQ Tests Test, stating that Richardson cited Leila Zenderland’s (1998) book Measuring Minds, a biography of Goddard. MH cites a passage from Zenderland on Goddard:
While Goddard believed that most of these immigrants were indeed mentally weak, he wondered about the cause. “Are these immigrants of low mentality cases of hereditary defect”, Goddard now asked pointedly, “or cases of apparent mental defect by deprivation?” If the former, they still posed a threat to posterity; if the latter, then Americans need have no fears about the succeeding generations. While Goddard knew of no data to settle this “vital question”, he himself believed it “far more probably that their condition is due to environment than it is due to heredity. Their “environment has been poor” and “seems to account for the result,” he decided.
Such conclusions could hardly be said to support those calling for more restrictive legislation.
MH then says “As we will see later, Richardson cites sources that if read in their entirety frequently contradict his claims.” This is ridiculous. In his 2002 paper, he does indeed cite Zenderland 6 times, but here’s the thing: five of the citations are about Binet; one for the claim that IQ tests are ‘intelligence’ tests like Galton claimed. As I showed above, IQ testing was indeed used to attempt to curtail the number of immigrants into America.
MH then claims that, due to a quote with ellipses in Richardson’s 2002 paper that he was being deceptive not giving the whole quote and that he was
trying to dig up stuff where spearman or Charles Murray or somebody is admitting that something he’s arguing against has major weaknesses. So he finds that quote and thinks ‘Hm pervasive. That makes it sound as if there is a lot of evidence for this, I don’t like that. But I like the part where he says the evidence is circumstantial and the reality remains arguable. So I’ll just cut that part out. Who’s actually going to check this? The vast majority of my readers wouldn’t be caught dead owning The Bell Curve, much less actually reading it in any detail. Besides, I put ellipses, it’s all legal and above board.’
I personally have read The Bell Curve a few times and I’m familiar with the quote; I don’t think that the ellipses, in any way, diminishes Richardson’s point.
I’ve written in-depth on this subject so I will be quick here. MH states that “it cannot be claimed that IQ tests have construct validity in the strict definitional sense.” He “partially agrees with the criticism” but he only “partially agrees” due to the “correlations” with regard to job performance and scholastic achievement.
Back in September, I wrote an article on test construction, item bias and item analysis. More recently, I wrote on the history of IQ testing and how tests are constructed with the presuppositions of the test’s constructors. Finally, in my most recent article on the ‘meaasurment’ of ‘intelligence’ I noted that first, IQ-ists need to provide a definition for intelligence, then they need to prove that IQ tests measure intelligence (they assume the tests measure what needs to be defined); then, after all is said and done, can IQ-ists then posit about “genetic” causes of intelligence and other psychological traits and variation between racial and ethnic groups. I have also created a syllogism in the modus tollens form showing that IQ tests cannot be construct valid:
Premise 1: If the claim “IQ tests test intelligence” is true, then IQ tests must be construct valid.
Premise 2: IQ tests are not construct valid.
Conclusion: Therefore, the claim “IQ tests test intelligence” is false. (modus tollens, P1, P2)
IQ ‘predicts’ things through test construction; it’s not really a ‘prediction’, in any case. Since IQ tests are related to other kinds of achievement tests—indeed, they are different versions of the same test—the claim that IQ is a predictor of future success is therefore circular (Richardson, 2017b). Indeed, all of the claims that IQ specifically are predictive can be explained by other, less ‘mystical’ ways.
Social class and IQ
MH states that a problem for the “IQ as a measure of social class” argument is the fact that “most of the IQ variation in society is within families … about 70 percent of IQ variation is due to with-in family differences.” MH then quotes Richardson stating that correlations between .6 and .7 have been reported between IQ and maternal encouragement, for example, then stating that Richardson did “not mention the strong caveats Mackintosh presents following his summaries of these studies.” MH then quotes Mackintosh stating that while the correlations between a developing child’s IQ and variables like parental involvement and attitudes and the presence of books, toys and games in the home “the establishment of these correlations alone will never prove that one is direct cause of the other.” MH then states that there are two possibilities: how the child acts can influence elicits certain responses from the parent or that parents influence child development at least as much through their actions toward their children along with the genes they pass on to them.
MH then invokes the “sociologists fallacy” which is the tendency to think of a correlation between a social variable and a phenotype as causal without thinking that genetics mediates the relationship between the social variable and the phenotype in question—which is known as “genetic confounding”, where genes confound the relationship between two variables. However, for the “genetic confounding” claim to have any weight, there must be a mechanism that produces psychological variation, so in lieu of that, the “genetic confounding” claim, and along with it the “sociologist’s fallacy” charge are irrelevant until a mechanism is identified.
Other aspects of social class can, as well, differ between siblings such as teacher quality, teacher treatment, school quality and so on—all of which influence IQ (Ceci, 1990). Furthermore, Richardson never claimed that social class accounts for all of the variations in IQ. Richardson (2002) writes:
It suggests that all of the population variance in IQ scores can be described in terms of a nexus of sociocognitive-affective factors that differentially prepares individuals for the cognitive, affective and performance demands of the test—in effect that the test is a measure of social class background, and not one of the ability for complex cognition as such.
Richardson’s main claim (and which he successfully argues for) is that variation in the sociocognitive affective preparedness nexus accounts for the variation in IQ. IQ is “in effect” (to use Richardson’s words) a measure of social class since social class is a significant determinant of the variables that make up the sociocognitive affective preparedness nexus.
MH then cites Korenman and Winship (1995) who write that:
incredible as it may seem, our sibling analysis suggest that, even though Herrnstein and Murray’s parental SES index is poorly measured and narrowly conceived, it appears in most cases adequate for producing unbiased estimates of the effect of AFQT scores on socioeconomic outcomes.
MH then states that the AFQT (Armed Forces Qualifying Test) “is really just an IQ test” but, as Mensh and Mensh (1991) note, such tests were biased from their beginnings due to how they were constructed and how items were chosen to go along with the presupposed biases of the test’s constructors.
MH then brings up the Wilson Effect, which “is the observation that the heritability of IQ increases by age and by adulthood, the effect of the home environment has almost zero contribution to individual differences in IQ on average” (MH). The Wilson Effect, too, is an artifact of test construction. Richardson (2000: 36) writes:
Another assumption adopted in the construction of tests for IQ is that, as a supposed physical measure like height, it will steadily “grow” with age, tailing off at around late puberty. This property was duly built into the tests by selecting items which a steady proportion of subjects in each age group passed. Of course, there are many reasons why intelligence, however we definne it, may not develop like this. More embarrassing, though, has been the undesired, and unrealistic, side effect in which intelligence appeared to improve steadily up to the age of around eighteen years, and then start to decline. Again, this is all a matter of item selection, the effect easily being reversed by adding items on which older people perform better and reducing those on which younger people perform better. […] That [IQ score differences] are allowed to persist is a matter of prior assumption, not scientific fact. In all these ways, then, we find that the IQ-testing movement is not merely describing properties of people: rather, the IQ test has largely created them.”
In response to the claim that Richardson has never “operationalized” social class, this claim is false. In his most recent paper, Richardson and Jones (2019) cite a whole slew of more recent research to buttress Richardson’s (2002) sociocognitive affective nexus, noting that social class is more about money, cars and things, but also is how we think and feel. Richardson and Jones (2019: 39) write:
Finally, different social conditions also lead to different affective orientations, such as self-confidence and achievement expectancies, that impact on school learning and test performances (Frankenhuis & de Weerth, 2013; Odgers, 2015; Schmader, Johns, & Forbes, 2008). The effects of test anxiety on cognitive performance are well known, and have been estimated to affect up to 15%–20% of school children (Chin, Williams, Taylor, & Harvey, 2017). In addition, feelings of social rejection effect test performances and self-regulation (Stillman & Baumeister, 2013).
In sum, whatever else CA and EA scores measure, they at least partly reflect a socio-psychological population structure in ways probably unrelated to any general cognitive or learning ability.
MH then quotes Richardson citing Hoge and Coladarci (1989) who states that teacher judgments have a higher correlation between teacher’s assessment and future success in life. MH states that since the teachers were presumably well-acquainted with the children and their academic aptitudes that this explains the higher correlation than IQ tests have with future success of students in their life.
… the marginal time cost is small, nearly every child is already in school, but if you’re a parent being told your child needs to be placed in remedial classes, what are you more likely to trust? The judgment of a single random teacher or an IQ test standardized on thousands of children from a representative sample of the population with a test-retest reliability of .9?
The claim that teacher’s judgments can be done in a “fraction of the time” compared to IQ tests is indeed true. I have noted that this is how these tests were constructed originally in the early 1900s, and early test constructors related teacher’s judgments on ‘intelligence’ to their subjective presuppositions, constructing the test on the basis of teacher’s judgments and their own biases.
What explains professional success? IQ or social class? Ceci (1990: 87) notes that “the effects of IQ as a predictor of adult income were totally eliminated … when we entered parental social status, and years of schooling as covariates.” Ceci goes on to write that since education and social class were signficant and positive indicators of adult income “this indicates that the relationship between IQ and adult income is illusory … Thus, it appears that the IQ-income relationship is really the result of schooling and family background, not IQ.” (pg 87). So it one’s social standing (access to schooling and family background) that mediates the IQ-income relationship.
Mensh and Mensh (1991) note that Gould held contradictory views on IQ testing. He noted the racist and social origins of the testing movement, but accepted IQ tests for their utility for certain uses—most likely because they helped to identify his son that had a learning disability. IQ tests are not objective scientific instruments; indeed, how can a human mind (in all of its subjectivity) create an unbiased test? That IQ tests are standardized on thousands of people are irrelevant; the IQ test constructors can build what they want into and out of the test, so claiming that a parent should trust a (biased) IQ test over the judgment of “a single teacher” who has had years of teaching experience is superior—as Hoge and Coladeri (1989) do indeed show.
Lastly, MH cites brain imaging/head measuring studies showing correlations between IQ and the measures (Rushton and Ankney, 2009), while also purportedly showing that this holds among siblings as well (Lee et al, 2019). Schonemann et al (2000) show that brain size does not predict general cognitive ability within families, while pre-registered studies show lower correlations between .12 and .24 (Pietschnig et al, 2015; Nave et al, 2018).
Indeed, a parent’s belief about their child’s GPA (grade point average) remain even “after controlling for siblings’ average grades and prior differences in performance, parents’ beliefs about sibling differences in academic ability predicted differences in performance such that youth rated by parents as relatively more competent than their sibling earned relatively higher grades the following year” (Jensen and McHale, 2015: 469). More arguments showing why these things would differ within families can be found in Richardson and Jones (2019). MH then cites a table of motor vehicle fatalities in Australian army personnel under 40, noting that the death rate in motor vehicle accidents sharply increased the lower one’s IQ score (O’Toole, 1990). I don’t contest the data, I contest MH’s interpreation of it: am I supposed to accept IQ as causal in regard to motor vehicle fatalities? That one is just dumber than average which then causes such fatalities? Or is the social class explanation much stronger—in that one’s access to resources and education influences their IQ scores? MH finally discusses reaction time (RT) in the context of its relationship to IQ. But Richardson’s (2002: 34) sociocognitive affective nexus, too, explains the relationship:
… low-IQ subjects regularly produce RTs equal to those of high-IQ subjects, but with less consistency over trials. This lack of consistency may well reflect poor self-confidence and high test anxiety and their effects on information processing, incursions of extraneous cognitions, sensory distractions and so on.
All in all, MH is implying that IQ’s correlations with brain imaging/skull measurement, the relationship between motor vehicle fatalities and the relationship between RT and IQ all point to the claim that IQ measures intelligence and not social class. This is a strange claim. For the structure and items on IQ (and similar) tests reflect that of the middle class. Indeed, the Flynn Effect rising as the middle-class increases is yet more evidence that IQ is a measure of social class. MH then claims that assuming that IQ=intelligence explains these things better than the assumption that IQ=social class. However, there has been much sociological research into how social class affects health and, along with it would affect scores on achievement tests (which are inherently biased by race, class, and sex; Mensh and Mensh, 1991; Au, 2007, 2008). IQ tests do not measure learning (what many IQ-ists use as a stand-in for ‘intelligence’); what IQ tests do is “sort human populations along socially, culturally, and economically determined lines” (Au, 2008: 151; c.f., Mensh and Mensh, 1991).
I think the video was well-researched and well-cited (to a point, he didn’t discuss all of the critiques that Snyderman and Herrnstein received on their Immigration Act paper), but he failed to prove his ultimate claim: that IQ tests measure intelligence and not social class. Goddard was one of the most well-known eugenicists in the 19th century, and his views had a devastating social impact, not only on European immigrants vying to emigrate to America, on the populace of ‘morons’ and those who were ‘feebleminded’ in America: they were sterilized as they were deemed ‘unfit’ to have and care for children (Wilson, 2017). IQ tests are not construct valid (which MH agrees with) but he still is possessed by the delusion that success at jobs is causally related to IQ (see Richardson and Norgate, 2015). The ‘sociologist’s fallacy’ claim and the genetic confounding claim both fail as you need to identify a causal (genetic) mechanism that is responsible for variation in psychological traits. The observation that IQ score heritability increases as children age is, too, built into the test through item selection. The claim that Richardson does not operationalize social class is false (see Richardson and Jones, 2019). Neuroimaging analyses show lower relationships between brain size and IQ when they are pre-registered; his citation to vehicle fatalities and IQ is irrelevant as is the part about RT and IQ—as social class, too explains the outcomes.
IQ most definitely is a measure of social class, as an analysis of the items on the test will show (see Mensh and Mensh, 1991; Richardson, 2002; Castles, 2012) and not a ‘measure’ of ‘intelligence.’