Letters to and from the Editor

LETTERS TO AND FROM THE EDITOR Dear Sir: We feel it is necessary to correct a misleading impression given by Professor Jensen in his article, "The Phylogeny and Ontogeny of Intelligence" [I]. Jensen writes, "It is now generally accepted by geneticists, psychologists, and sociologists who have reviewed the evidence that social class differences in mental abilities have a substantial genetic component." This statement should be considered in rela- ' tion to the following facts which at least many geneticists would accept: (i) The proportion of the IQ variance that arises from genetic variety in a population is less than 100 percent, (ii) IQ is a continuous variable, and it is not yet possible to determine gene frequencies, (iii) It follows "that we cannot at present answer the question whether there is any genetic component of social class or race differences in mean IQ" [2]. It is also impossible to say whether there is any environmental component, for this is, in fact, the same question. Thus, as it is impossible to quantify either the genetic or the environmental component of social class differences in mean IQ, we feel it is unwise to say that there is "a substantial genetic component" in these differences. We find it disturbing that Jensen chooses to ignore these facts. REFERENCES 1.A. R.Jensen. Perspect. Biol. Med., 15:37, 1971. 2.J. M. Thoday. J. Biosoc. Sci., 1 (suppl.):3, 1969. John B. Gibson C. G. Nicholas Mascie-Taylor James N. Thompson, Jr. Department of Genetics University of Cambridge Cambridge, England Dear Sir: The first two statements listed by Gibson, Mascie-Taylor, and Thompson are certainly true. I disagree, however, with their third point. It is not a necessary condition either that there be 100 percent heritability of IQ or that we must be able to count single gene frequencies in order statistically to infer with a high degree of probability the existence of a genetic component in the mean IQ difference among socioeconomic classes. (I am here referring to social class differences within a given racial group. The question of intelligence differences between racial groups involves other methodological problems and types of evidence not discussed here.) 154 I Letters to and from the Editor Several lines of evidence support with a high level of confidence the conclusions that social classes, on the average, differ to some degree in the genetic factors involved in intellectual development. Social classes may be viewed as Mendelian populations that have diverged genetically. When the population is stratified into five or six socioeconomic status (SES) categories, mainly according to occupational criteria, the mean IQs of the adults so classified, from the highest SES category (professional and managerial) to the lowest (unskilled labor), span a range of some 30-40 points. The standard deviation of IQs within SES groups averages about 9 or 10 points for the adult population, as compared with SD = 15 for the whole population. Children born into these SES groups, on the other hand, show a mean IQ difference, from the lowest to the highest class, of only 20-30 points; and the SD within classes for children is about 13 or 14 IQ points, which means there is almost as much IQ variation among children within social classes as we find in the total population. The cause of the higher degree of correlation between SES and IQ among adults than among children is the high level of social mobility in each generation. In England and in the United States more than 30 percent of the adult generation are found to be of a different SES than that of their own parents [1-3]. In each generation some individuals move up in SES and some move down. Those who move up have higher IQs, on the average, than those who move down. Since the heritability, h2 (i.e., the proportion of genetic variance), of IQ in the total population is between .70 and .80, and since the correlation between phenotypes and genotypes is the square root of the heritability, it follows that IQ estimates genotypic intelligence with a reliability of between -y/_jÖ and vT8Ö (i.e., between about .84 and .89 [4, 5]). Conversely, the reliability with which...