Wednesday, September 30, 2015

Unstandard Deviations

"But the burden of whiteness is this: You can live in the world of myth and be taken seriously." Ta-Nehisi Coates 
"It was only by reading – and checking – the actual data in The Bell Curve... the undisputed data..." -- Andrew Sullivan
Ta-Nehisi Coates is a Genius!

In a discussion over at Angry Bear of Coates's recent Atlantic piece on mass incarceration, one of the commenters raised the spectre of the I.Q. gap between Whites and Blacks, which is allegedly 15 I.Q. points or roughly "one standard deviation" from the norm of the reference White population.

Where does this ubiquitous "15 I.Q points" come from? Arthur Jensen, in his 1969 Harvard Educational Review article, “How Much Can We Boost I.Q. and Scholastic Achievement?” cited a 1966 (first edition, 1958) book by Audrey Shuey, The Testing of Negro Intelligence:
Negroes test about 1 standard deviation (15 I.Q. points) below the average of the white population in I.Q. and this finding is fairly uniform across the 81 tests of intellectual ability used in the studies reviewed by Shuey.
So who, then, was "Shuey"?

Dr. Audrey M. Shuey was Chairman of the Department of Psychology in the Randolph-Macon College for Women, at Lynchburg, Virginia. Dr. Shuey's mentor at Columbia University was Henry E. Garrett, a militant segregationist and hereditarian who also contributed a foreword to her book. In his 1958 review of the first edition of The Testing of Negro Intelligence, Horace Mann Bond wrote:
Never before has the literature of psychology witnessed so determined an effort to establish, as a fact, the proposition that there are "native differences between Negroes and whites as determined by intelligence tests."
By the interesting device of discarding all "interpretations, criticisms, comments," and even conclusions in individual studies, and taking the statistical tables reporting differential scores as the only, bona-fide, "results," Dr. Shuey arrives at what she calls a "remarkable consistency in test results, whether they pertain to school or pre-school children, to high school or college students, to drafts of World War I or World War II, to the gifted or the mentally deficient, to the delinquent or criminal." This "remarkable consistency" becomes the foundation of her concluding inferences that "point, to the presence of some native differences between Negroes and whites as determined by intelligence tests."
Another reviewer, Ina C. Brown, concluded her review of the book with the observation:
The book really adds up to much ado about nothing. No informed person questions the fact that on the average Whites perform better than Negroes on the tests or that northerners perform better than southerners or urban subjects better than rural subjects. It is the why that is important and Dr. Shuey’s brushing aside the interpretations of most of the recent testers in favor of her own conclusions adds nothing to our knowledge. 
One can, however, predict wide use of the book by White Citizens’ Councils and others who are in search of material which they can interpret as “scientific” support for their point of view.
The prediction that White Citizens' Councils would widely use the book was either prescient or simply well informed. Publication of the book was funded by Wickliffe Draper's Pioneer Fund, which sponsored free distribution of the book throughout the South, assisted by the Citizens' Councils. After Shuey's death, two other Pioneer Fund grantees, R. Travis Osborne and Frank McGurk, published a follow-up volume to include data from 1966 to 1982.

Now about that "actual data" Andrew Sullivan claimed to have "checked" and found "undisputed." Bullhockey. The following chart summarizes the allegedly "undisputed" data about race and I.Q. relied upon by Murray and Herrnstein in The Bell Curve. Note the sources: Shuey, Osborne and McGurk, Jensen. Why not list the sponsors, too: Pioneer Fund... White Citizens' Council... Wickliffe Draper... Henry Garrett?

Undisputed data? Virtually every aspect of the race and I.Q. data cited by Murray and Herrnstein -- along with its hereditarian interpretation -- has been not only disputed but refuted. Voluminously.

The only aspect left "undisputed" is that there are differences in I.Q. scores. But by itself that "fact" is trivial and meaningless. It is like saying a dollar is worth 100 cents because each cent is worth one-hundredth of a dollar. The logic is circular.

How could Sullivan have "checked" the data in The Bell Curve and yet be unaware of the extensive critiques of the data by such authorities as Stephen Jay Gould, Richard Lewontin or Leon Kamin?

Undisputed data? Seriously? A better description for "the actual data in The Bell Curve" and the way in which it was presented would be fraudulent. In his 1995 review of the book, Michael Nunley called it "too smooth to be true":
Their fraud begins in their introduction with a series of astonishing assertions. They list a number of propositions they claim form the very foundation of their book, but for which they tell us they are not going to provide any evidence. These include such notions as that there is such a thing as general cognitive ability; that IQ is the most accurate measure of it; and that IQ tests are not biased against any social, economic, ethnic, or racial groups. They are excused from providing any evidence for these assertions because, as they put it, these statements are "beyond significant technical dispute." This struck me as odd because, you see, I’m an anthropologist with an interest in cross-cultural studies of cognition. And if I were trying to think of things that might be considered "beyond technical dispute" from cross-cultural studies in anthropology, they would be that human cognition is far too complex to be captured on a simple linear scale; that IQ measures only a very narrow band of all kinds of cognitive achievement; and that tests of cognitive achievement cannot avoid being culturally biased by the content, materials, and style of testing used. 
In an afterword to the 2010 edition of The Bell Curve, Charles Murray dismissed -- with a few well placed "never minds" -- criticism by Charles Lane and Leon Kamin of the book's extensive dependence on "tainted sources":
Never mind that The Bell Curve draws its evidence from more than a thousand scholars... Never mind that the relationship between the founder of the Pioneer Fund and today's Pioneer Fund is roughly analogous to that between Henry Ford and today's Ford Foundation.
It is, of course, impossible to refute a "never mind" that declines to substantiate the evidence for the claims that it sarcastically asserts will be ignored anyway.

Never mind that more than 995 of Murray's "more than a thousand scholars" are irrelevant to the five studies cited in The Bell Curve's analysis of differences between Black and White cognitive test scores, three of which come from books and articles by Pioneer Fund sponsored authors.

Never mind that Murray offered no rationale or evidence for his spurious analogy between the Ford Foundation and the Pioneer Fund.

Never mind that Murray's "refutation" of the charge that the book relied on "tainted sources" was a non sequitur and a red herring. Murray's coy "never minds" are, to use Nunley's description, "too smooth to be true":
When I first sat down to read Herrnstein and Murray’s book The Bell Curve (1994), I was predisposed to be skeptical of certain positions I’d heard it was promoting, but I was nevertheless expecting it to be an honest attempt to make a case for a wrongheaded but sincerely held scientific position.  But when I read the text of The Bell Curve, including its footnotes, which were stuck at the back of the book, forcing me to page back and then forward again every paragraph or two, with all the book’s references buried well back in the footnotes, and when I then tracked down and read some of the references to see what they actually said and what other scholars had said about them, I became convinced that The Bell Curve is something quite different from my initial expectation. I believe this book is a fraud, that its authors must have known it was a fraud when they were writing it, and that Charles Murray must still know it’s a fraud as he goes around defending it. By “fraud,” I mean a deliberate, self-conscious misrepresentation of the evidence. After careful reading, I cannot believe its authors were not acutely aware of what they were including and what they were leaving out, and of how they were distorting the material they did include. The book is, moreover, a very good fraud, a very cleverly constructed one. There aren’t enough "o"s in “smooth” to describe it.
Ta-Nehisi Coates is a Genius.

Coates's Atlantic feature story on mass incarceration commences with a discussion of Daniel Patrick Moynihan's 1965 report, "The Negro Family: The Case for National Action." Toward the end of his account of the Moynihan report, Coates mentions that "William Ryan, the psychologist who first articulated the concept of 'blaming the victim,' accused Moynihan’s report of doing just that."

Toward the end of his article, Coates returns to Moynihan and briefly alludes to his 'ominous" citation of "a 'rather pronounced revival -- in impeccably respectable circles -- of the proposition that there is a difference in genetic potential' between the two races." Moynihan noted the proposition in March 1969 report to Nixon regarding Arthur Jensen's Harvard Educational Review article.

Impeccably respectable circles.

Undisputed data.



Thornton Hall said...

Well done.

Over at the Reality Based Community, Mark Kleiman frequently makes the point that there's a difference between beliefs where the accuracy of the belief is important for day to day living and beliefs where error is of little direct consequence except to signal tribal affiliation.

A white, rural, Southern farmer believes all the science behind properly fertilizing his crops while simultaneously dismissing climate change as a hoax. The meaning of "belief" changes allowing individuals of sound mind and normal intelligence to tell pollsters that they "believe" the President is a Muslim.

But how far up does this go? How sharply does a Tea Bagger's "belief" transition to Murray's "fraud"? (And I agree it is fraud in his case).

Does Mike Huckabee believe Jesus stands for hate? My father uses the Moynahan Report as a touchstone, but when encountering an actual unmarried black father he is compassionate and understanding.

Do libertarians really think that competition causes firms to self-regulate? In the face of Ashley Madison, Target, and the Federal Givernment all failing to safeguard private personal data in the exact same fashion? There's a bit of a network effect, but the Ashley Madison App is pretty close to a perfect market: you need a computer and a software engineer to compete. Is Rand Paul a fraud or dumb?

Andrew Sullivan? Is there some part of him that admits the motivated reasoning on steroids that caused him to believe Murray's challenge to liberal TNR readers was worthy of the platform?

Thornton Hall said...

Re-reading Sullivan's post (I read it at the time he posted it, too) I'm struck by how one wrong idea runs thru his and Chait's responses to TNC. The belief not just in the right of free speech, but in the positive value of all speech. It came up in the press's unthinking "Charlie Hebdo=heroes" stance as well. Noah Smith also falls for it. And the thing is, I believed it too, well into my first year of law school. The response to bad speech is good speech. And good speech eventually wins on its own merits.

Only a white man can believe such nonsense.

But Chait and Sullivan really do believe it. It's an article of faith among all journalists, almost. Writers at Charlie Hebdo think it's funny to deliberately antagonize a persecuted minority in a country where it's illegal for that minority to wear certain clothes and that mocking supercilious sneer is a positive good for society?

America becomes a better place to live because someone has the courage to publish the claim that black people are ("only on average", Sullivan exclaims, "on average!") stupider than white people because they inherit stupid genes from their parents?

But, "the response to bad speech is good speech" is really easy for people on the business of speaking to belief.

Sandwichman said...

Jensen 1969 article was published in the April issue of HER. Moynihan reported to Nixon about it on March 19. That kind of impact factor is hard for "good speech" to match. Citizens United raises that outcome to a constitutional principle.

The bias of Jensen's article confirmed Nixon's own squalid racial prejudices. Moynihan elected to remain loftily agnostic. Nixon' crude ignorance is somehow more forgivable than Moynihan's cautious disavowal. The nature of the emperor's new clothes was not an "open question," as Moynihan caviled. Moynihan should have known better. Instead of "good speech" countering "bad speech" we have pandering abdication.

john c. halasz said...

Flynn effect.

anonymous said...

In your post, you say that Dr. Audrey M. Shuey was Chairman of the Department of Psychology in the Randolph-Macon College for Women, at Lynchburg, Virginia. This is incorrect. Dr. Shuey was a faculty member at Randolph-Macon College in Ashland.

Sandwichman said...

Thanks, anon. I'll check my sources again.

Galtonian said...

Hello Mr. Sandwichman,

I am curious.

Do you not believe that African-Americans tend to have IQ scores that average about 15 points (one Standard Deviation) lower than Euro-Americans?

Do you not believe that IQ (general cognitive ability) is a substantially heritable human trait?

Do you not believe that racial group IQ differences are related to racial group academic achievement gaps?

Sandwichman said...


1. No. 2. Irrelevant. 3. Irrelevant.

To elaborate, 1. the frequently reported "15 point, one standard deviation" difference is based on NO comprehensive comparable data. It is based on ad hoc compilations performed by people whose bias (and biased funding) was clearly expressed. The reported results are simply not credible. Most of the reported results are simply repeating Shuey.

2. Whether or not I.Q. is heritable has nothing to do with whether alleged differences between African-Americans and Euro-Americans. There are much larger differences in I.Q. scores WITHIN the poorly defined groups than there are BETWEEN them. The hypothesis that racial or ethnic group I.Q. test scores reflect some kind of group heritability is simply not testable. The hypothesis (like magic) CANNOT be refuted therefore it is not a scientifically valid hypothesis.

3. I.Q. testing and academic achievement are both historically specific institutions with deeply entrenched and ingrained cultural biases. To the extent there is a correlation between them that doesn't necessarily "validate" the test scores.

Now I ask you a question. Do you know of any IQ studies in which a test was constructed by African-American psychologists using African-American children as the primary reference group and then subsequently administered to various ethnic and racial groups students? If not, why not?