|
Upstream: Issues: Bell Curve: SarleEditor's Note: the following material was extracted from Usenet postings.
Newsgroups: sci.stat.math,sci.stat.consult I never said that _all_ the reviews were idiotic, only _most_ of them. I have repeatedly cited Lane's review in the NYRB as highly informative. To the list of highly informative reviews, I now add Richard Nisbett's, published in _The Bell Curve Wars_; his original review in _The New Republic_ was just unsubtantiated claims, but the expanded version in TBCW has references. He argues that most of the evidence is "consistent with a zero genetic contribition to the [B/W] gap" in IQ, meaning that the differences are nonsignificant (the editor of the TBCW, Steven Fraser, accepts the null hypothesis, saying that the studies "point to a zero genetic contribution"). Nisbett's section on "The Effects of Intervention on Cognitive Skills" is the most interesting. He cites a 1992 article in _Pediatrics_ that "showed a nine point IQ advantage at age three as a consequence of early intervention for at-rsik infants." Perhaps the most interesting is a study that shows that interventions directed at study habits can dramatically improve the performance of black students in college math courses (Triesman, U. (1992), "Studying students studying calculus: A look at the lives of minority mathematics students in college," The College Mathematics Journal, 23, 362-372). To balance Nisbett and maintain the average level of idiocy, TBCW includes a review by Jacqueline Jones, professor of history at Brandeis University, who wins the prize for Political Correctness. She castigates H&M for sexism as well as racism, and accuses them of supporting the sterilization of all poor women, the disenfranchisement of people with low IQs, lynching, slavery, etc. ad nauseam. She concludes by saying that TBC is "hate literature with footnotes." Kamin's review from _Scientific American_ has been revised and published in _The Bell Curve Debate_ (not to be confused with TBCW). Kamin has corrected one statistical gaffe (where he claimed that scores on Raven's Progressive Matrices cannot be converted to IQ equivalents because RPM has a skewed distribution) but has introduced another. In numerous figures in TBC, H&M present regression lines or predicted probabilities adjusted for various factors. One factor that appears in most of the analyses is age, so when H&M plot a regression line, they plug in the average age of the NLSY sample, 29. Since there were rarely any significant interactions in the analyses, this is a reasonable thing to do. But Kamin thinks that these analyses are based only on the subset of the data for which the respondents are 29 years old, and he rebukes H&M for using an excessively small sample! He further ridicules (his interpretation) of their analyses for failing to make predictions in exact integers: "The regression analysis informs us that fully 26% of those 2.78 blacks (0.72 of a black) are predicted to be in the professions. Murray is right; we are losing ground. Before the days of affirmative action, an entire token black was par for the course." _The Mismeasure of Man_ had a good discussion of factor analysis but grievously misrepresented several works, especially Jensen's _Bias in Mental Testing_; Gould's treatment of Jensen is almost as scurrilous as his trashing of H&M. While I have read and enjoyed some of Gould's work in the past, I now consider him thoroughly unreliable. Gould vs. _The Bell Curve_: A Review of a Review Stephen Jay Gould's book _The Mismeasure of Man_ (TMM) established the current conventional wisdom on intelligence testing--basically, that IQ is a sham. But TMM is a work of history--it does not address the last 50 years of research except to caricature Arthur Jensen's work. Herrnstein and Murray's (H&M) book _The Bell Curve_ (TBC) is a direct threat to TMM, showing that IQ is relevant to a wide variety of social problems, based on a massive review of research (over 800 items in the bibliography) and analysis of a large data base, the National Longitudinal Survey of Youth (NLSY), of exceptionally high quality. So how does Gould react to TBC in his review published in _The New Yorker_? He staunchly refuses to consider current research and merely recycles arguments from TMM attacking research from the first few decades of the century. Gould has improved his old tactic of caricaturing his victims--now he simply states blatant lies about TBC. Attacks against TBC have centered on the alleged claim that difference in mean intelligence between blacks and whites in the U.S. are genetic in origin. Gould says (p.139): "The Bell Curve," with its claims and supposed documentation that race and class differences are largely caused by genetic factors and are therefore essentially immutable, ...and (p.139): Hernnstein and Murray's second claim, the lightning rod for most commentary, extends the argument for innate cognitive stratification to a claim that racial differences in I.Q. are mostly determined by genetic causes ...and (p.141): Hernnstein and Murray violate fairness by converting a complex case that can yield only agnosticism into a biased brief for permanent and heritable difference.It is curious that Gould should use the word "agnosticism", since what H&M actually say in summarizing arguments regarding racial differences in IQ is (p. 311): It seems highly likely to us that both genes and environment have something to do with racial differences. What might the mix be? We are resolutely agnostic on that issue; as far as we can determine, the evidence does not yet justify an estimate.Perhaps Gould would argue that the above paragraph is merely a disclaimer tacked on to a biased argument (pp. 141-142): They impose this spin by turning every straw on their side into an oak, while mentioning but downplaying the strong circumstantial case for malleability and little average genetic difference.But in fact, no competent scientist could read TBC and conclude that the evidence presented therein demonstrated that racial differences in mean IQ are "mostly" genetic. There is circumstantial evidence for some genetic difference, but the experiments necessary to reach anything resembling a firm conclusion have not been done and are impossible to do in contemporary society--how can you shield a black child from all racial prejudice? There is simply no evidence in TBC that racial differences in mean IQ are "mostly" genetic, as Gould claims there is. Gould shows a remarkable talent for packing a large number of lies into a single sentence. For example, he says (p. 139): Intelligence, in their [H&M's] formulation, must be depictable as a single number, capable of ranking people in linear order, genetically based, and effectively immutable.Nowhere do H&M say that "Intelligence ... must be depictable as a single number." Intelligence tests contain multiple subtests measuring different aspects of intelligence. H&M discuss these subtests and various combinations thereof in several places, notably pp. 299-303 where they argue that profile difference among ethnic groups provide circumstantial evidence of genetic differences. It is ironic that critics such as Gould accuse H&M both of claiming that intelligence is reducible to a single number and of claimimg that racial differences are mostly genetic when in fact the multidimensionality of intelligence is essential to one of the few arguments that H&M present in favor of genetic differences among races. Nowhere do H&M say that IQ is "capable of ranking people in linear order". They state repeatedly that IQ is of limited value for comparing individuals. For example, on p. 19: Suppose that the question at issue involves individuals: "Given two 11 year olds, one with an IQ of 100 and one with an IQ of 90, what can you tell us about the differences between these two children?" ... On many important topics, the answer must be, "We can tell you nothing with any confidence." It is well worth a guidance counselor's time to know what these individual scores are, but only in combination with a variety of other information about the child's personality, talents, and background. The individuals IQ score all by itself is a useful tool but a limited one.Again on pp. 82-85, H&M point out that, although IQ is generally the best predictor of job performance, the correlation between IQ and job performance is modest and of little value for choosing among a few job applicants. Nowhere do H&M say that intelligence is based exclusively on genetics. Based on a review of the literature, they say (p. 23) that: Cognitive ability is substantially heritable, apparently no less than 40 percent and no more than 80 percent.Nowhere do H&M say that intelligence is effectively immutable. They are careful to make the point that some degree of genetic heritability of IQ does not necessarily impose limits on the ability to change IQ environmentally. For example, on p. 390, they say: Even the highest estimates of heritability leave 20 to 30% of cognitive ability to be shaped by the environment.and: Limitless possibilities for improving intelligence environmentally wait to be uncovered by science: [examples deleted]. In principle, intelligence can be raised environmentally to unknown limits.What H&M do say, based on a review of the literature, is that raising the IQs of disadvantaged children is difficult and the effects diminish over time. The conclusion of chapter 17, p. 416: An inexpensive, reliable method of raising IQ is not available. The wish that it were is understandable, and to pursue the development of such methods is worthwhile. But to think that the available repertoire of social interventions can do the job if only the nation spends more money on them is illusory.Gould apparently finds his own lies so convincing that he is angered by them (p. 143): Nothing in [TBC] angered me more than the authors' failure to supply any justification for their central claim, the sine qua non of their entire argument, that the number known as _g_, the celebrated "general factor" of intelligence, ... captures a real property in the head.The reason that H&M supply no justification for this claim is that they do not make any such claim. The central claim of TBC is not that g or IQ is "a real property in the head" but that IQ is among the best predictors of a wide variety of social ills, including poverty, crime, divorce, unemployment, illegitimacy, etc. In particular, as shown by many statistical analyses of the NLSY data, IQ is a better predictor of most social ills than is parents' socioeconomic status (SES), controlling each of IQ and parents' SES for the other variable. While it is customary to blame poverty as the major contributor to these social ills, H&M show that to the extent that we consider poverty to be a serious social problem, we are obliged to consider low IQ a _more_ serious problem. There is absolutely no argument in TBC that relies in any way on IQ being "a real property in the head". -- Warren S. Sarle SAS Institute Inc. The opinions expressed here saswss@unx.sas.com SAS Campus Drive are mine and not necessarily (919) 677-8000 Cary, NC 27513, USA those of SAS Institute. |
| Microsoft Internet Explorer 5.01 | 10/12/100 13:32:05 | User: 1 | Last visit: Sat Dec 9 23:52:33 2000 |