DEFENDING THE SCIENCE OF MENTAL ABILITY AND ITS CENTRAL DOGMA
Book Review of Jensen on Intelligence-g-Factor
Joseph A. Buckhalt
Department of Counseling & Counseling Psychology
Auburn University
Auburn, AL 36849
buckhja who-is-at auburn.edu
ABSTRACT: Reaching back to the original theory of Spearman, Jensen
presents a comprehensive synthesis of the past century's evidence
and concludes that g and its nexus constitute no less than the
central dogma of psychometrics, the science of mental ability.
While his conclusions regarding behavioral and psychometric
correlates of g are largely convincing, those regarding the
presumed biological and evolutionary foundations of g are less
strong. It is a formidable work that enriches our understanding of
individual differences in intelligence and it is sure to stimulate
more scientific research and debate.
1. Jensen (1999) describes this book (1998) as his third major work,
with the first two being the notable article published in Harvard
Educational Review in 1969, and Bias in Mental Testing in 1980 (a,b).
In his introduction, he links the themes of the three works together. I
believe that Jensen views much of the controversial and disputed
aspects of the first work as being due to misconceptions and
ignorance. Among the loudest critics early on were those who claimed
that IQ tests were biased irreparably against ethnic minorities. Bias
in Mental Testing was a methodical march through basic tenets and
procedures of test construction, the determination of reliability and
validity, and primarily, how bias may appropriately be defined. As
Jensen states in the present work, the conclusions of that book are now
widely accepted. And furthermore, although he is too modest to admit
it, many of his critics have come away looking foolish on the topic of
bias.
2. Once Jensen essentially quieted the critics on the subject of bias,
critics took another tack and dismissed the importance of the tests
altogether, calling them trivial, "merely paper and pencil" (actually,
the tests rely very little on paper or pencils), measuring only
"academic intelligence" that is pertinent only in the limited venue of
schools (e.g., Gould 1981/1996). Jensen's task now is no less than to
mount a defense of the construct of mental ability (g) as it has been
classically defined and measured. He does so by reconstructing g from
the ground up, beginning with Spearman's 1904 formulation. He assembles
an enormous number of studies and convincingly argues that g is a
legitimate, robust scientific construct, and that it is a reflection of
a basic biological dimension along which humans vary. It is,
furthermore, a source of individual differences with profound
implications for every aspect of life, including, but not limited to,
school achievement.
3. In the present work (as in Bias in Mental Testing), Jensen seems to
be speaking not only to scientists within his discipline, but also to
scientists across disciplines and to nonscientists. To nonscientists,
Jensen is saying, "My scholarship on this topic is within the domain of
science. If you wish to understand what I have concluded, you must step
inside the scientific realm and abide by the rules, which I have
summarized in brief form. Rhetoric outside of these parameters is quite
another matter." To scientists, he is saying, "Psychometrics is a
legitimate scientific enterprise. It has rules of observation,
quantification, measurement, and logic. What I propose is testable
within the rules of science. Everything I conclude is subject to
empirical disconfirmation, and those who wish to challenge me should
bring data."
4. It is unlikely that nonscientists will readily relinquish their
presumed authority on human intelligence. But Jensen argues for
removing the term from the realm of ordinary discourse where it has
equal status with other abstractions like love, justice, and the like.
Rather, he argues that intelligence, like gravity or radiation, must be
defined precisely for any reasonable dialogue about its properties to
take place. Moreover, since it cannot be observed directly, we must
take time to understand the highly standardized manner in which it may
be indirectly observed and understood. To avoid confusion, Jensen
suggests abandoning the term intelligence altogether, and starting anew
with the g factor. He suggests always using the term in that form to
remind us that g is a factor with a particular mathematical
definition. I'm not altogether convinced that we need to abandon the
term for scientific discourse, but I share his frustration with the
gratuitous stretching of the term by contemporary writers such as
Sternberg, Gardner, and Goleman. While many have eagerly embraced
broadening of the construct to include a wider and wider range of human
characteristics, Jensen successfully distinguishes the classical notion
of intelligence from all of the misuses of the term popularized
recently. He is kind to those who have offered what they believe to be
serious challenges to a g based theory of intelligence, but ultimately
concludes that while current speculations are interesting, they have
diverged much too far to be subsumed under the term intelligence, they
are largely lacking in scientific validity, and they do nothing to deny
the reality of g. Given the misconceptions and disagreements
surrounding the terms intelligence and IQ, it is perhaps justifiable
that Jensen tries to distance himself from them early in the book.
5. Many contemporary psychologists view the field of differential
psychology as moribund, or dormant at best. Differential psychology is
regarded as a purely applied field, with the IQ test being a
successful, if tired, tool for classifying individuals for educational
and clinical purposes. And, most damningly, applied differential
psychology has long been said to have no viable theories of
intelligence. Jensen's claim is that we've had a theory all along, but
we've just insufficiently acknowledged or appreciated it. In some
sense, Jensen's work is a defense of early 20th century psychometrics.
He concentrates primarily on how research of this century has
essentially borne out the primary ideas of Spearman. Further, he
elevates the discovery of g by Spearman in 1904 to the same status as
Thorndike's law of effect. In other words, he regards (correctly in my
view) the discovery of g as one of the very few singular, lasting, and
influential accomplishments of psychology. It is a central dogma
supported by a mountain of evidence and broad consensus among
psychometricians.
6. Jensen's current work is an ambitious, comprehensive, and in my
mind, successful articulation of the central dogma of the psychometrics
of mental ability. As he did in Bias in Mental Testing, he goes to
great length to present, often in primer format, the basic foundations
for his arguments. Among these are the scientific foundations and
methods of classic psychometrics, the rationale and methods of factor
analysis, and principles of behavior genetics and evolution. While I
have no doubt that critics will seize upon particular studies or their
interpretation and conflate them into fatal flaws, my opinion is that
Jensen's scholarship is accurate, his conclusions reasonable where much
empirical data exist, and his speculations prescient. His arguments are
strengthened by his positioning himself firmly within the scientific
tradition of building hypotheses subject to disconfirmation. Within the
rules of this endeavor, anyone wishing to bring down the overall
construction (and many will try), will have a herculean task ahead of
them. Critics of the conclusions reached by Herrnstein & Murray (1994)
in The Bell Curve, had essentially one (albeit very sound) study to
attack. In this book, Jensen has pulled together a nexus of hundreds of
studies, and finding fault with one or another of them will not
threaten the edifice.
7. While the edifice that Jensen has erected is an impressive one, a
few weaknesses are present, some within Jensen's area of undisputed
expertise, psychometrics, and others in his attempt to explore what he
terms in the last chapter, the "vertical" directions of g's
foundations. First the weaknesses within psychometrics. Jensen argues
that experts cannot agree on a definition of intelligence. But to some
degree (admittedly some less degree), the same can be said of
definitions of g. Spearman was elusive when trying to pin down the
essence of g, and so is Jensen. After presenting a definition deriving
from factor analysis, many of the studies he relies upon for his case
have used a multitude of different tests, some of which are
unidimensional such as the Ravens, which Jensen regards as an "almost
pure measure of g". But should not the definitive measure of g be the
hierarchical 3rd order factor? And what of Spearman's own admonition
(cited on p. 31), that with more tests, more g is captured? The problem
is, of course, that a relatively small number of the total number of
studies used by Jensen to marshal his evidence have been done with g
determined according to Spearman's recommendation. Notwithstanding this
point, my guess is that Jensen's argument would likely be strengthened,
rather than diminished, had more of the research he cites used better
measures of g. Another problem, though, is that Jensen regards
psychometric measures of g to be reflections of an underlying set of
biological variables that are now only dimly understood. This dim
understanding is problematic as Jensen reaches "downward" through
physiological mechanisms toward genetics.
8. In building up the case for g, Jensen gives only cursory treatment
to mental ability factors other than g. While he builds a convincing
case that the g factor is ubiquitous, and accounts for much variability
in a wide range of human endeavors, what of the other factors in the
hierarchy? Jensen states that Carroll's three stratum theory is not
inconsistent with his views on g, but he pays scant attention to
Carroll's other strata. Jensen's admiration for Spearman's theory is
obvious, but why is there so little mention of Spearman's specific (s)
factors? It seems that a goal of psychometrics is to account for more
and more variance, and to do that, it must go beyond g. Little
attention is given even to the best-known 2nd order factors, those of
verbal and nonverbal abilities, which as Jensen documents, were even
acknowledged by Spearman to be important. Since Wechsler's bifurcation
in the 1930's, verbal and nonverbal IQ have become integral parts of
intelligence tests, with both practical value in diagnostics and very
good factor validity. But on this matter I cannot really fault Jensen,
because his stated purpose in the book is to restore the centrality and
dominant position of g, not to construct a comprehensive theory of
intelligence.
9. When Jensen departs from pure psychometrics, his attempts to link g
with with variables such as nerve conduction velocity, brain size, and
myopia, seem very crude or at least preliminary. They seem something
akin to showing a relationship between blood pressure and temperature
to general health in an era where much more complex and sophisticated
causal mechanisms are known. Jensen no doubt believes that future
consilience between the neurosciences and psychometrics will only
strengthen his case. It is probably best that Jensen is intentionally
conservative and vague in his comments about what illumination
neuroscience has provided thus far and is likely to provide in the
future. He no doubt realizes that any pronouncements at this point are
at best premature, and at worst, dead wrong. Here I am reminded of
attempts to reconcile intelligence theory (and construct intelligence
tests) with the discovery of hemispheric specialization. One gets the
sense that Jensen yearns to build stronger bridges to the neurosciences
than he does, but realizes we must wait for further research in other
disciplines to be done.
10. In what will unquestionably be seen as his most controversial
chapter, Chapter 12, Population Differences in g: Causal Hypotheses,
Jensen turns to his hypotheses about race differences in g. Beginning
with what he calls "Spearman's hypothesis", namely, that the degree of
race differences on tests is proportional to the tests' g loadings, he
casts a very broad net, extending from psychometrics to population
genetics and evolutionary theory. Here he is certain to be criticized
for not being an expert in those fields. Others may reach different
conclusions about the data he reviews, and future research may prove
him wrong on some counts, but he is to be admired for continuing to put
the questions before us and place their investigation in the realm of
science as opposed to rhetoric. One falsifiable hypothesis is worth
more than thousands of pages of prose by self-proclaimed experts.
11. While much of the book is a reiteration (and a very good one) of
the research of the 20th century related to g, in the last chapter, The
g Nexus, Jensen offers numerous directions for research in the next
century. He suggests that the prediction of behavior would be
considerably strengthened by supplanting measures of g with measures of
a variety of nonintellective (or personality) factors such as
motivation, perseverance, and conscientiousness. He presents these in a
section titled "The Limitations of g", and reminds us that while g is
the best single predictor of many types of performance, future research
should explore the interactions of other variables and factors with g.
I have long been puzzled by the failure of psychology to seek and
achieve a synthesis between so-called intellective and nonintellective
components of human performance. The interaction between intelligence
and personality is still for the most part uncharted territory.
Consider the work of Eysenck, for example, who made major contributions
to both intelligence and personality theory but offered very little
integration to synthesize the two areas. The linking of g to other
variables is one of Jensen's recommendations for enlarging the g nexus
in a "horizontal" direction.
12. In the book's very last paragraph, Jensen proposes a possible
heuristic for future research in the "vertical" direction, the
neurophysiological basis of g. While I have commented above that
Jensen's attempts to link g to biology are necessarily weak, given our
present knowledge, he nevertheless sees further understanding of g to
be in the realm of neuroscience, as apparently did Spearman. In fact,
while Jensen's view seems to be that very little advancement (and
considerable confusion) in intelligence theory has taken place since
Spearman published The Abilities of Man in 1927, he does seem hopeful
that the next century holds promise for greater understanding. Arthur
Jensen has enriched our understanding of intelligence (no, I have not
abandoned the word) for 30 years through his systematic and exhaustive
scholarship. Count me among those who agree with his major conclusions
about the past century's psychometrics. I would not bet against his
intuitions concerning what the next century will bring.
REFERENCES
Gould, S. J. (1981/1996) The Mismeasure of Man. New York: Norton.
Herrnstein, R. J. & Murray, C. (1994) The Bell Curve: Intelligence and
Class Structure in American Life. New York: Free Press.
Jensen, A. R. (1969) How much can we boost IQ and scholastic
achievement? Harvard Educational Review, 39:1-123.
Jensen, A.R. (1980a). Bias in mental testing. New York: Free Press.
Jensen, A. (1980b) Precis of "Bias in Mental Testing." Behavioral and
Brain Sciences 3: 325-372.
Jensen, A. (1985) The Nature of the Black-White Difference on Various
Psychometric Tests: Spearman's Hypothesis. Behavioral and Brain
Sciences 8:193-219
Jensen, A. (1998) The G Factor: The Science of Mental Ability.
Praeger.
Jensen, A. (1999) Precis of: "The G Factor: The Science of Mental
Ability" PSYCOLOQUY 10(23).
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?10.23
Spearman, C. (1904). 'General intelligence, objectively determined and
measured.' American Journal of Psychology, 15, 201-293.
Spearman, C. (1927). The abilities of man: Their nature and
measurement. New York: Macmillan.