Bias in Mental Testing, by Arthur R. Jensen, New York: Free Press, Macmillan. 1980. 786 pp. $29.95.
A distinguishing feature of libertarian social theory is its location of justice in the elementary acts between individuals and its consequent acceptance of any sort of overall "order" that arises as the result of just elementary actions. This is in contrast to the "social engineering" approach, which applies a normative standard to the aggregate order (for example, a "just" distribution of income) and manipulates its elements so as to yield the "right" overall result.
This same divergence of principle underlies the current flap over affirmative action. On one side are those who take the a priori normative position that any lack of strict proportionality in group ratios at any time at any level is naked proof of invidious discrimination; hence the various stages of selection we all pass through ought to be manipulated so that a "socially desirable" proportionality is maintained at all times. On the other side are those, like us, who argue that there is no way to define justice to groups that is distinct from the just treatment of the individuals that comprise them; hence, any distribution of group proportions that results from the fair treatment of individuals is by that itself just.
This is the framework in which to view the ongoing controversy about IQ tests. The point of departure for this debate is that IQ tests and all other standardized tests of achievement and professional aptitude (for example, the SAT, MCAT, LSAT, or GRE) do not generally yield the same proportions of various social groups all the way up the grade distribution, even though the identical test is given to each person and each is graded blindly. Specifically, the central tendency of blacks is lower than that of gentile whites, and that of the latter well below that of both Orientals and Jews; and the central tendency of higher socioeconomic groups is above that of lower ones, and that of southerners below that of northerners.
There is, of course, a wide overlap in all of these distributions, and it should be clearly understood that what is not at all at issue in the current debate is whether individuals should be evaluated in terms of the average scores of their group rather than their own manifest abilities (for example, a return to racially segregated public schools). No party to the central debate argues in favor of such measures.
What is at issue are two questions. One, a normative matter that need not be argued here, is whether we really want after all to accept the outcome of fair treatment of individuals, even when it is conceded to be fair at this level, if this has the result of generating large group disproportions. The other is a much more limited question, namely, whether standardized tests are, despite surface appearances, all that fair in the first place. Perhaps the average group differences in scores are caused by bias in the tests against some ethnic or economic groups, with the bias lying not at such a superficial level as unfair grading but in the sorts of questions it asks and the kinds of experiences and cultural knowledge it presupposes. It is the latter question that this massive book by Arthur Jensen addresses. It is a brilliant book that will inevitably be, if not the last word (faint hope!), the Mount Everest in this field. Anyone wanting to take intelligent part in this debate will have to be familiar with its contents.
I do not intend to review all the technical arguments set forth in this book but will merely indicate that it contains detailed and thorough discussions of all the usual questions laymen ask about IQ and aptitude tests. In brief, IQ tests do measure intelligence (that very thing whose reality we experience in such commonplace observations as the fact that kids get smarter as they get older); the rankings determined by mental tests do correspond rather well to later academic performance; such evaluations have virtually the same predictive validity for whites, blacks, "native" Americans, Japanese, Chinese, Hispanics, southerners, northerners, the rich, the poor, and even New Yorkers, as long as they are raised in this country and speak English; average group differences are not explainable in terms of the race or ethnicity of the test giver; and mental tests do have an uncanny power to "read through" social veneers and reveal significant differences in academic capability. Nothing currently exists in the way of an alternative means for making merit allocations of scarce academic opportunities that, at the level of individuals, is as fair, objective, and unbiased as mental tests.
This is not to say that tests are never misused or abused, and Jensen is quite circumspect in what he advocates as the proper uses of mental tests. But such misuses are largely rooted in the bureaucratic sloppiness, lethargy, and indifference that we well know typifies the system of public schooling and that can only be remedied by means of a radical shift toward educational voluntarism. The purely scientific question whether individual differences in intelligence can be ascertained by means of mental tests can certainly be answered in the affirmative.
If it is true that the evidence supporting the validity of mental tests is compelling (and this book should convince you that it is), then one might reasonably wonder why the same old canards about bias, or about IQ not reflecting "real" intelligence, get thrown up again and again in the public discussion. I cannot tell you what the reason is, beyond observing that it is very much the same reason why, in the face of the mass of evidence establishing the superiority of the free-enterprise system in speeding economic growth and fostering civil liberties, the likes of Galbraith, Commoner, Harrington, Heilbroner, Howe, and Hayden continue to drone on about the marvels of their Lorelei, socialism. It is a faith akin to religion, and like a religious conviction, it is impervious to all possible counter-demonstration. One might with justice call this attitude a "test bias."
Bill Havender has a Ph. D. in genetics from the University of California at Berkeley.
This article originally appeared in print under the headline "IQ (Intelligence Questions)".