The Darwinian tradition has entered psychology in two different ways. The first is a conservative, deterministic approach, based upon the theory of natural selection and the idea of the survival of the fittest. The second emphasizes development and flexibility, treating variability itself as an evolutionary asset. We argue here that Jensen's is closer to the first, conservative and somewhat rigid approach. This is apparent in a general view of our species that is focused on the correlations of g. Although the one-factor theory of intelligence is obviously a cumulative scientific program, reducing data and treating mental ability in a unitary way, as an applied theory it leads to a conservative, impatient and one-sided view of society.
2. This divergence is part of a general problem about whether to consider human variability as (1) adaptation to an ever-changing and open environment, with randomness a part of normal development or as (2) a distribution of traits in a stable world, where different quantities in the distribution are connected to positive or negative values according to their adaptedness to the conditions of this stable environment: The outcome of (2) would be a social distribution of individuals. Spearman's tradition, continued by Jensen, is a clear example of explaining psychological phenomena in terms of a single underlying factor.
3. Jensen argues that scientific psychology should discard the term "intelligence" altogether, just as it discarded "animal magnetism," and as chemistry discarded "phlogiston" (p. 48). Jensen states that it is possible to avoid all the trouble with intelligence by simply dropping the term, and replacing it with g. The hope of inding consensus between paradigms and discordant groups of psychologists through a new definition seems illusory (as the current controversy in this very forum illustrates). Nor does it end the debate on defining the term intelligence: it simply gives a new, implicit definition of mental abilities. According to Jensen, mental ability as a whole (previously called intelligence), is nothing but g itself. (This is quite obvious from the subtitle of his book: "the science of mental ability.") Hence, as g is the principal factor underlying diverse cognitive tests, what was formerly called intelligence corresponds to what all kinds of mental abilities have in common. If an indiviual's g is measured, and they are much better in, say, math than in verbal skills, their skill will not be represented sufficiently in their g, but will form a group factor, and hence will not be part of what was previously called intelligence.
4. Everyday experience shows that there are different kinds of talent. The so-called "group factors" (e.g., different types of mental abilities) exist, and despite their overall correlation, as found in various kinds of factor analysis, their practical importance is underestimated. In arguing about the practical validity of g, Jensen admits that
"Since these are statistical trends to which there are many exceptions, prediction based on a g measure for a given individual is only probabilistic, with a wide margin of error" (p. 290.).
The main problem with applying g-factor theory to a given individual, however, is not in predicting, but in the structural model of mental abilities it implies. Factor analysis is also a statistical method; hence although when individual data are aggregated they show a high level of correlation, this does not determine a given individual's mental abilities.
5. Criticism based on differential abilities within individuals is not new. Here are Jensen's counter-arguments against this criticism:
"It is sometimes claimed that any given person shows such large differences in various abilities... One student does very well in math, yet has difficulty with English composition; another is just the opposite, a third displays a marked talent for music but is a mediocre in English and math. Is this a valid argument against g? It turns out, that it is not valid, for if it were true, it would not be possible to demonstrate repeatedly the existence of all-positive correlations among scores on diverse tests' abilities, or to obtain a g factor in a hierarchical factor analysis" (p. 93).
This argument is not valid as it is based on statistical methods that (as Jensen also admits, see above) have little to do with individuals. It is an empirical fact that diverse tests of mental abilities correlate positively and that a positive common factor can be extracted, but this does not at all exclude the possibility of someone's doing very well in math for example, but having difficulties in other mental tasks. The higher the level of g, the greater the probability of finding large differences in abilities within persons, because the correlation between mental tests decreases at higher levels of g.
6. This argument is also valid outside the field of mental abilities. The correlation table described by Jensen on page 300 shows clearly how little validity these correlation data have for a given individual. We argued above that it seems improbable that the correlation between abilities should rule out the possibility of individuals doing well in one cognitive domain and much worse in another. Taking a closer look at this correlation table, we even doubt that a single person doing well in all of these domains could even exist.
7. It is clear that Jensen tries to reduce psychological variability to one underlying factor. The same correlation table also shows that g is closely related to certain values -- mainly those of the American middle class. The positive and negative correlations of g suggest a picture of what someone with a high g might be statistically expected to be like: altruistic, healthy, field-independent, successful, emotionally sensitivity, with certain dietary preferences (low-sugar, low-fat) and a good sense of humor, good at shopping in a supermarket, but free of racial prejudice, a nonsmoker, and so on (p. 300). People with high g probably have middle-class values, which means that they are likely to have a high school degree (or higher), were gainfully employed throughout the year, have never been interviewed in jail, and are still married to their first spouse (p. 296). According to this table, g is nothing but a predictor of social adaptation to middle-class values and behavioral patterns.
8. Jensen's research, based on Spearman's theory of mental abilities tries to show that different mental abilities are based on the same entity (although he admits that this entity might be a group of factors, not a single property of the brain), and hence that g is the best performance predictor in diverse cognitive tasks. This seems to contrast with findings from applied psychology. For example, personnel selection is nowadays based on measuring highly specific abilities. It has not proven sufficient even to measure specific abilities (or group factors); these had to be measured in the specific tasks to which they were to be applied, simulating the everyday challenges of the specific occupation in question (Woodruffe, 1993).
9. In most cases, applied educational psychology concentrates on individuals rather than groups. In a school setting it is much more useful to measure a students individual constellation of mental abilities than to classify him by a single number. If someone is doing badly in most subjects they might still have an extraordinary talent in a specific field. Alhough according to statistics this has a low probability, there is always some chance, and teachers must foster the domains in which individuals show any sign of excelling. If someone is a mediocre in math, they may have good language skills, and so on. For example, the correlation of skill in second-language acquisition with g is low, but with other psychological variables it is high (Bass & Selinker, 1994).
10. The first part of Jensen's book deals with the theories, models and characteristics of mental abilities. He uses experimental results and statistical methods to show that there is an underlying factor with which a wide range of mental abilities can be described. Our opinion is that he is absolutely right, but the results of such scientific programs do not strictly determine their application. The case is very similar to personality research. If we wish to measure someone's personality as thoroughly as we can, tests according to the "Big Five" theories are perfect for this purpose. But if we have to measure a single trait (e.g., impulsiveness) for a specific purpose, then it is unnecessary to define the person's whole personality. This is also the case with mental abilities: in everyday situations, such as personnel appraisal or scholastic achievement, we usually have to measure specific aptitudes. Statistically, g might be a good and valid overall predictor, but for a single person it is much more reasonable to measure specific abilities (e.g., spatial abilities for engineers) and their unique pattern within that individual.
11. The same argument appies if we look again at the correlation table discussed earlier (p. 300). It is not emphasized enough in this part of the book that these data are strictly statistical. The g factor may correlate highly and significantly with not smoking. But if we wish to know whether someone smokes, the easiest way is to offer him a cigarette.
12. The opportunities for applying g factor theory seem overestimated. The overall positive correlation among diverse kinds of mental abilities is an empirical fact. However, as this correlation is not perfect (e.g., r << 1) we always lose information when we use measurements of g instead of group factors or specific abilities. Hence the specific objectives of the testing itself should decide whether we should just use g or something else. This is quite similar to the case of medical testing. If we tested for different illnesses and injuries, in different parts and organs of the body, we would almost certainly extract a general health factor, which we would call h for health. This h factor could surely be extracted from a factor analysis; on the one hand, many illnesses (for example cancer) attack different organs of the body, and on the other there is an underlying factor (the immune-system) responsible for our overall protection against illnesses. This common factor would emerge not only between individuals, but within individuals as well. We can also be certain that this h factor would show high heritability, and would correlate highly with different illnesses within as well as between families.
13. If we extracted this h factor, it would obviously have interesting applications. We could use it in an interesting cumulative scientific research program, and search for correlations between h and different phenomena such as life span, occupational success or g. It would also be a perfect entity to measure if we wished to select members for an exclusive International Health Club, where we would accept only the healthiest 2% of the entire population. But it would be highly misleading to use it in everyday medical practice, rather than concentrating on the specific problem of a patient.
14. Applied and academic psychology should be distinguished in a further respect: the criterion level for statistical significance. Tests and theories in scientific research can have a much wider margin of error than practical psychological tools, with the great influence they exert on individual lives (e.g., scholastic aptitude tests and personnel selection tests). The chances of making a mistake should be minimized as much as possible. As Stephen Jay Gould (1996) writes in his popular book:
"We pass through this world but once. Few tragedies can be more extensive than the stunting of life, few injustices deeper than the denial of an opportunity to strive or even to hope, by a limit imposed from without, but falsely identified as lying within" (p. 61.).
15. Although Gould's words are somewhat melodramatic and his views about the uselessness of mental tests are hard to defend, the importance of the fight against any sort of false limitations on human values and talents cannot be overemphasized. Gould is probably wrong when he fights against mental testing itself; we think mental tests are much more useful than he admits. But the idea of mixing scientific research and applied psychology is dangerous, and the critical value for significance should be different in the two domains. The criterion for significance in psychology is p<0.05, which means that our results can be based on pure chance once in every twenty times. This value is unacceptable iwhere decisions based on these methods influence people's lives, as in those situations where the chance of error could amount to wronging thousands of people. This is as unacceptable as it would be in the medical sciences (e.g., pharmacology) or in any science which deals with human life and is responsible for irremediable decisions about people's individual fates and well-being. In pharmacology a p value of less than or equal to 0.001 would still be unacceptably high.
16. This means not only that mental tests used for such decisions should be much more reliable than other experimental tools in psychology; in most cases, they already are. But this criterion should also apply to theories of mental abilities. If we find a high overall correlation between mental abilities, then that can become a good basis for further investigation, and for measuring external correlations with mental abilities as a whole. But even if this overall correlation indicates that the possibility of great differences among someone's mental abilities is small, that possibility must not be excluded as long as it influences an individual's life. The role of error must be minimized, not only in tests, but also in theories.
17. Our main conclusion concerning what Jensen calls the practical validity of g and the g nexus is that two things seem to be mixed here. The first is a scientific paradigm based on Spearman's theory of intelligence and the hypothesis that there is a single factor underlying the diverse mental abilities. This view determines a methodology and a so-called scientific program, which is cumulative, because it leads to successful new investigations in the field of mental abilities. The second is the social application of this theory, which seems to underestimate the possible variability of mental abilities, not merely in a quantitative, but also in a qualitative way. Humans may differ in the structure of their abilities: one might have better grades in math than in literature and so on, and neither students nor recruits should be measured and classified by a single number. It is mainly a question of attitude whether a scientist thinks of human variability as a positive result of evolution, or of people as good or bad according to a single quantity.
Gass, S. M. Selinker, L. (1994). Second Langauge Acquisition: An Introductory Course. Hillsdale, NJ: Lawrence Erlbaum Associates
Gould, S. J. (1996). The mismeasure of man. New York: Norton.
Jensen, A. R. (1998). The g factor: The science of mental ability. Westport, CT: Praeger.
Jensen, A.R. (1999). Precis of. "The g Factor: The Science of Mental Ability" PSYCOLOQUY 10(023). ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.10/ psyc.99.10.023.intelligence-g-factor.1.jensen http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?10.023
Woodruffe, C. (1993). Assessment centres: Identifying and developing competence. London: Institute of Personnel and Development