Robert A. Gordon

Everyday life as an intelligence test: Effects of intelligence and intelligence context

Abstract To show why the importance of intelligence is often misperceived, an analogy between single test items and single nontest actions in everyday life is drawn. Three requirements of good test items are restated, and the analogy is employed to account for underrecognition of the importance of general intelligence in everyday actions, which often fail to meet the requirements and thus fail as intelligence measures for reasons that have little to do with their dependence on intelligence. A new perspective on the role of intelligence in nontest actions is introduced by considering its operation at three levels: that of the individual, that of the near context of the individual, and that of entire populations. Social scientists have misunderstood the operation and impact of IQ in populations by confining attention to the individual level. A population-IQ-outcome model is explained that tests for the pooled effects of intelligence at all three levels on differences between two populations in prevalences of certain outcomes. When the model fits, the difference between two populations in the outcome measured is found commensurate with the difference in their IQ or general intelligence distributions. The model is tested on and found to fit prevalences of juvenile delinquency, adult crime, single parenthood, HIV infection, poverty, belief in conspiracy rumors, and key opinions from polls about the O.J. Simpson trial and the earlier Tawana Brawley case. A deviance principle is extracted from empirical findings to indicate kinds of outcome the model will not fit. Implications for theories of practical and multiple intelligences are discussed. To understand the full policy implications of intelligence, such a fundamentally new perspective as that presented here will be needed.

The estimation of the prevalence of delinquency: Two approaches and a correction of the literature†

The index “the proportion of a cohort that have become delinquent by a given age,” here called the “prevalence of delinquency,” is an important social indicator. In the present paper, we indicate methods by which this index can be estimated from data, and correct errors in previous sex‐ and race‐specific prevalence estimates published by Monahan (1960) for the city of Philadelphia. The difference between the sexes and between the races shown by these corrected prevalence estimates are of sufficient magnitude to render suspect any comparisons of prevalences of delinquency among cohorts which do not take account of the sex and race compositions of the cohorts to be compared.

Can we count on muddling through the g crisis in employment

Abstract The fragmentation of academic disciplines handicaps efforts, educational and otherwise, to deal rationally with problems arising from group differences in general intelligence. In personnel psychology, eyes tend to be on the courtroom, but the classroom may prove the more telling arena. Perhaps equally serious is the failure of each discipline to reckon with strains outside of its immediate province in calculating how much latitude exists for errors of its own that might add to those strains within the polity. One such error would involve doing away with selection tests, thereby compromising unwittingly the fundamental principle of merit. Open discussions like those in this special issue are essential if such blunders are to be avoided. This article illustrates these points through comments on the moral, scientific, and legal concerns addressed by the contributors, and especially through a critique of Seymours (1988) new analyses, which purport to reveal unsuspected racial unfairness in tests.

An Explicit Estimation of the Prevalence of Commitment to a Training School, to Age 18, by Race and by Sex

Abstract Estimates of the race and sex specific prevalence to age 18 of commitment to a training school are derived. The obtained relative prevalence rates are found to be similar to another set for a different criterion, a different time, and a different, smaller jurisdiction. It is concluded that differences in relative rates of three to one make it impossible to interpret differences in absolute rates without taking into account the racial composition of the population. This viewpoint is in direct opposition to current proposals to suppress the use of racial and other ethnic identifications in the collection of social data.

Digits Backward and the Mercer-Kamin Law

Jensen (e.g., 1980a) has grouped the various criteria for detecting bias in ability tests under two broad headings: internal and external. Studies under the first heading concern themselves with relations among the components of a test; those under the second are concerned with predictive relations between the test as a whole and outside variables. Mercer organized her chapter in this book to parallel Jensen’s distinction, and my chapter parallels hers, but, because of space limitations, only the half of her chapter concerned with internal validity.

Book Reviews : Daniel N. Robinson (Ed.) Heredity and Achievement. New York: Oxford University Press, 1970. Pp. X + 441.

tition of symbols and their names, explanations of tables both in the text and at the tables themselves, in addition to isolation and review of concepts and formulas, are other important procedures used by the author to insure understanding and retention. The 12 chapters of the book include the usual topics in descriptive and inferential statistics at the elementary level: frequency distributions, central tendency, variability, percentiles, regression, correlation, hypothesis testing, and t tests. In addition, one-way and two-way analyses of variance are covered in two chapters; nonparametric techniques in a long, 41-page chapter; and further topics in probability in the final chapter. Almost twice as many pages are devoted to the last six chapters, which deal with statistical inference, as to the first six chapters on descriptive statistics. Chapters 7 and 8 on hypothesis testing are particularly well written.