The Bell Curve Debate
A Lucy Horwitz review of
1. The Bell Curve by Richard J. Herrnstein & Charles Murray
2. The Bell Curve Debate- Russell Jacoby & Naomi Glauberman, editors
Surely no reader of this column needs to be told what The Bell Curve, probably the most reviewed and discussed book of this or any recent year, has to say. But for those of you who have been visiting other planets during the last six months, here is the basic argument:
General intelligence, called "g," and defined as "a person's capacity for complex mental work," constitutes the "broadest conception of intelligence." IQ is "what people mean when they use the word intelligent or smart." G, IQ or smarts can be accurately (and without cultural bias) measured by standard intelligence tests. IQ is 40%-80% heritable and is relatively stable over a person's lifetime. Blacks' IQ scores are significantly lower than whites'. Low IQ is the cause of social problems such as poverty, crime, unemployment, illegitimacy, welfare dependency, etc. High IQ is the only passport to success. The world is rapidly separating into a cognitive elite and a cognitively deficient underclass. Nothing can be done to raise IQs, and social programs such as Head Start, Affirmative Action, AFDC, etc. are useless, counterproductive or both. Since nothing can be done to change these inequalities, the only answer is "letting people find valued places in society." My first reaction when I read The Bell Curve some months ago was regret that this book was not available twenty years ago. At that time I was writing an elementary statistics text, half of which covered the fundamentals, while the other half used examples to demonstrate how statistical fallacies can distort the truth beyond recognition-inadvertently or purposefully. I spent long hours in the library, collecting examples of such fallacies from many different sources-newspaper and journal articles, advertisements, political speeches, etc. And now here is a book which could have saved me all that trouble. Every fallacy I wanted to demonstrate, available in one convenient volume!
My second reaction, of course, was regret that this insult to our intelligence had been published at all. As the weeks went by, I was happy to see that no one seemed to be taken in. Experts and laymen alike were quick to point out the dishonesty of the book, uncovering flaw after flaw in its words and numbers, its logic and statistics, its charts and graphs. It was variously called "atrocious science" (Scientific American, February, 1995), "pseudoscholarly" (San Francisco Chronicle, November 6, 1994, "sleazy" (The New York Review, November 17, 1994),"an unusually lengthy promotional brochure for a rather unattractive political package." (The Washington Monthly) "a house of cards constructed to push a political agenda" (Business Week, November 7, 1994), "a fable masquerading as social science" (San Francisco Chronicle, November 6, 1994), etc., etc.
So why was it getting all this publicity? I decided it might be what I had come to think of as the Archie Bunker Effect. Why was All in the Family so popular? Because the strongest human emotion is not love or hate, but the need to feel superior to someone. And Archie Bunker, by expressing his need to feel superior to everyone different from himself, provided a double thrill-the thrill of having others put down, yes, but also the thrill of feeling superior to Archie. And so it seemed with The Bell Curve. I began to see it as providing the guilty thrill of feeling superior to a whole race, plus the virtuous thrill of feeling superior to its racist authors.
Was that too cynical? (My friend Frank says we are always afraid of being too cynical, whereas the world keeps reminding us that we are not cynical enough.) And what was I doing clipping every review, making my own lists of flaws and fallacies, clearly preparing to add my two cents worth?
The answers to these questions appeared in the form of a manuscript entitled The Bell Curve Debate. Just as the authors of The Bell Curve (hereafter TBC) could have saved me a lot of time in l974, the editors of The Bell Curve Debate (hereafter TBCD) could have saved me a lot of trouble in l994. For in 1995 they are coming out with a fat volume which gathers together everything that has been said about TBC and more. Perusing the manuscript resolved my conflicts about writing this column.
Although I could not see my way clear to urging people to go out and plunk down $30 just to see for themselves what trash can be published under the guise of "scholarship," I have no hesitancy at all in urging people to beg, borrow or buy a copy of The Bell Curve Debate, which raises a host of interesting questions and provides some very illuminating answers.
TBC clearly generates two areas of dispute-the scientific and the political. But TBCD raises many additional questions. There are the questions about the scientific data on race, IQ and heritability, and questions about how this data is to be interpreted, yes, but also about the meta-issue of whether science should investigate these questions at all. Similarly the political debate about the programmatic implications also leads to the question of whether these issues should be publicly debated.
After reading TBCD, I no longer doubt that the answer to this last question is a resounding "yes." Two things convinced me of the importance of having such a discussion. The first was the surprising number of people who were taken in by the bogus arguments of TBC. The second was my almost being taken in by "Going Public" by Father Neuhaus, in which he pleads for no public discussion of these issues. I read it with the same pleasure I often derive from reading Miss Manners, whose concern for people's feelings I find highly commendable. Father Neuhaus argues that "Society depends upon taboos and interdictions. Kindness is no limp or expendable virtue." He then goes on to chide the authors for speaking about racial differences in cognitive functioning "especially when they conclude that there is little or nothing that can be done about it."
There's something terribly wrong here, and it's not too difficult to see what it is. The problem is a failure to make the distinction between taking people to task for telling vicious lies about an issue and saying that the issue shouldn't be discussed at all. As long as some people secretly believe in the inferiority of some segment of the human race, we shall have to continue to publicly expose their errors.
This leads us to the related meta-issue of the social responsibility of scientists. Are there certain things that scientists should not study because they are potentially damaging to society? This debate has raged for as long as I can remember, beginning with arguments about the atomic bomb. (Remember Tom Lehrer's "When the missiles go up who cares where they come down. That's not my department, said Werner von Braun"?) Well, it's a big question, and one we won't attempt to answer here-it will probably always remain between the consciences of individual scientists and their funding agencies.
But we can say something about this particular case. Charles Murray makes much of the alleged courage he and his co-author, recently deceased Richard Herrnstein, showed in breaking the "taboo" of talking about race and intelligence. Perhaps in some circles there is such a taboo. But among most responsible scientists today, the only taboo is bad science. And that taboo they have certainly broken, though I'd hardly call their act courageous. Ellen Willis, in "The Median is the Message" has an excellent discussion of this aspect of the debate.
It is true that there is very little responsible work being done in the field of racial differences, but as several writers point out, this is not because of any "taboo," but rather because modern science has moved beyond the crude notion of human races. Read "Defining Race" by Steven A. Holmes (or better yet the article it refers to by Jared Diamond in the November 1994 issue of Discovery). Learn why it makes as much sense to group Eastern European Jews and French Canadians together, or Nigerians and Norwegians as it does to define the races we generally accept. It's all a matter of which genetic traits we pick. We happen to use skin color. But one could just as well use blood type, presence or absence of the enzymes lactase, the sickle-cell gene or any other sets of genes that vary from group to group.
The other supposed taboo is to talk about intelligence. And here again we run into not a taboo, but just plain bad science. As Howard Gardner puts it, "When it comes to science, the book could have been written a hundred years ago." Herrnstein and Murray (hereafter H&M) prefer to ignore all the advances that have been made in neurobiology, embryology, cognitive psychology, developmental psychology, evolutionary biology and genetics. No serious scientist now speaks of intelligence as a single faculty like g that is set at conception and remains unchanged throughout the lifespan. Intelligence is now generally regarded as context-specific, and whether you want to accept Howard Gardner's multiple intelligences or some other system, g is decidedly out.
There are many good articles about the question of what constitutes intelligence and what intelligence tests really measure. The best is Leon Kamin's "The Pioneers of I.Q. Testing," a real eye opener. For instance Binet, who designed the first intelligence test to predict school performance, did not believe that his tests measured a genetic immutable trait. He actually "prescribed therapeutic courses in 'mental orthopedics' for those with low test scores." Read also Brigitte Berger, who in "Methodological Fetishism" points out that what standardized intelligence tests measure is not intelligence, but something she calls "modern consciousness." In Joe Chidley's "The Heart of the Matter," Camille Paglia is quoted as saying "What they're calling IQ is Apollonian logic-cause and effect-that the West invented...to identify that narrow thing with all human intelligence is madness." Also look at Horace M. Bond's "What the Army 'Intelligence' Tests Measured."
So now we have two questionable constructs, race and intelligence, for which H&M analyze realms of data. To judge the quality of any data, one looks, among other things, at sample source and size. Leon Kamin, in "Lies, Damn Lies and Statistics," goes back to the original studies that H&M draw on. He sums up the result of his research by stating that "the caliber of the data cited by Herrnstein and Murray is, at many critical points, pathetic-and their citations of those weak data are often inaccurate."
As one example, the longest section in the book depends entirely on data from the "National Longitudinal Survey of Labor Market Experience of Youth,"(henceforth referred to as the NLSY). There are two main problems with this data. The first is that all the data is self-reported. That is, it comes from answers the subjects gave to questions about themselves. Not all self- reported data is worthless, but it certainly raises a red flag of warning to any experienced statistician. The second problem is sample size. The survey included 12,000 subjects. That seems a lot. But when they get sorted into dozens of different categories, and then only the extremes of the categories are used to draw conclusions, you end up with very small numbers on which to hang very large generalizations.
And now we come to the use of statistics. Admit it, you don't want to hear about it. And that's just what they're counting on. If Samuel Johnson were alive today, he might well decide that the very last refuge of the scoundrel is statistics. I will briefly mention three of the most serious sins committed here, and point you toward the articles that do an excellent job of explaining the issues in non-technical terms.
Probably the most egregious crime is the constant confusion of correlation with causation. Correlation simply means that a connection exists between two sets of data. For example, there is generally a correlation between height and shoe size. Taller people tend to have larger feet than shorter people. The fact that a correlation exists, tells you nothing about cause and effect. A correlation between A and B might be due to A causing B, B causing A, C causing both A and B, sheer coincidence, or experimental error. Yet over and over again correlation is treated as cause. In particular, the cause of all social ills is either stated or implied to be low IQ. See Kamin on this point, too. Many of the articles comment on this fallacy, and David Suzuki focusses on it in an excellent piece, "Correlation as Causation"
Closely related to the concept of correlation is that of regression analysis, and the longest section in the book (that using the NLSY data) is filled with graphs of regression lines. Here's how they work. You plot points on a graph such that each point represents two pieces of data, such as income and education or, if you are an author of this book, IQ and illegitimacy. Then you use a mathematical formula to calculate a "line of best fit." The angle of this line tells you whether there is a positive or negative correlation between the two sets of data. The guilty little secret that most non-statisticians do not know is that you can always draw such a line, whether there is a real correlation or not. Then, if you are an honest statistician you report the strength of the correlation. But as Gould puts it,"...in violation of all statistical norms..they plot only the regression curve and do not show the scatter of variation around the curve, so their graphs do not show anything about the strength of the relationships... Indeed, almost all their relationships are weak."
Finally there is the fallacy of confusing statistics about populations and individuals. This may seem too obvious to mention. Even H&M say in their disingenuous fashion that just because blacks as a group are dumber than whites, this says nothing about any individual black. However, when it comes to the concept of heritability, they completely lose it.
Heritability says nothing about the extent to which a trait is inherited by an individual. Rather it is a measure of the extent to which genotype (the genetic basis of a trait) matches phenotype (the actual expression of that trait) in a given population. If eye color were uniquely determined by one gene, then genotype would always match phenotype, and heritability would be 1.0. Speaking English (as opposed to some other language) has a heritability of 0. There are no genes for English-it is entirely environment-dependant. Traits that psychologists, geneticists and others study have heritabilities between 0 and 1. But because the distribution of genes as well as the interaction between genetic composition and environmental influences varies from one population to another, heritability does also. So to claim that any one number can represent the heritability of IQ for all populations is nonsense. Applying that number to individuals (as in 60% of a person's IQ is inherited) constitutes a complete departure from reality. Pat Shipman has a good discussion in her article "Legacy of Racism," though in other respects I feel she gives the devil rather more than his due.
There are many, many more flaws of this nature, but you get the picture. The final, and to me most damning, point that is made about TBC in TBCD is about the hypocrisy of the whole enterprise. Stephen Jay Gould calls it disingenuousness. As he puts it, "The authors omit facts, misuse statistical methods, and seem unwilling to admit the consequences of their own words." Howard Gardner calls it scholarly brinkmanship, and shows how this "encourages the reader to draw the strongest conclusions, while allowing the authors to disavow this intention." What, me racist? In "Curveball," Bob Herbert quotes a New York Times Magazine article in which Murray talks about how he and some friends burned a cross on a hill when he was a teenager. He is quoted as saying "It never crossed our minds that this had any larger significance." It is clear that Murray has learned nothing in the intervening years.
There are many more treasures in TBCD than I have room to describe. But don't miss Gregg Easterbrook's account of how he suddenly acquired basketball genes when he had the opportunity for daily practice in a black neighborhood, and just as suddenly lost them when he left, or Mike Walter's wonderful comparison of TBC with cold fusion, or Bruce McCall's "Ethnicity, Genetics and Cuteness (Addendum to Recent Fearless Findings)" from the New Yorker.
There are also a number of pieces that you can easily skip. I might note that I found the pieces supporting TBC almost uniformly stodgy, unimaginative rehashes of the book with nothing to recommend them in the way of new insight or new ideas. But what would you expect of supporters of 19th-century science and the politics of Attila the Hun?
If the human race is to survive on this planet, we will have to get beyond the tribalism of the '90s. We are going to have to recognize that we are all members of the same species, with much more to unite than to divide us. We must learn to celebrate our diversity, not because it is the politically correct thing to do, but because we truly appreciate that in it lies the power of the human race, not only to survive, but to evolve.
Lucy Horwitz is a Contributing Editor at the Boston Book Review.
BACK TO CONTENTS OF Special Articles
Since opening 5/25/97, visitors to
MATHEMATICIANS OF THE AFRICAN DIASPORA
SEARCH the site
CONTACT Dr. Williams