Elizabeth Johnston, Investigating Minds, 11/18/97

Lecture 18: Current Views of Intelligence Testing

Correlation Coefficents

The idea of a correlation coefficent was devised by Francis Galton; the notion is to calculate the degree of relationship between two variables, regardless of the units of independent units of measurement used for the two variables.

Spearman's Concept of 'g'

Charles Spearman was an English psychologist of a hereditarian persuasion. His views were pro-Galton and anti-associationist. He examined correlations between scores on different school subjects and found high positive correlations.

Classics French English Math Pitch Music
Classics - .83 .78 .70 .66 .63
French .83 - .67 .67 .65 .57
English .78 .67 - .64 .54 .51
Math .70 .67 .64 - .45 .51
Pitch .66 .65 .54 .45 - .40
Music .63 .57 .51 .51 .40 -


From the universal positiveness of the correlations, Spearman inferrred that there must be a single factor common to them all. He called this single factor 'General Intelligence', shortened to the handy 'g'. Spearman conceived of this 'g' as a measure of mental energy or cerebral power, akin to the speed of the nervous system factor Galton tried to measure. He considered some tests highly saturated with 'g' or 'g-loaded' and others more dependent on the component of intelligence specific to that task, s. Spearman discussed the distinction between 'g' and 's' in terms of 'energy' and 'engines':
"The [g] factor was taken, pending further information, to consist in something of the nature of an "energy" or "power" which serves in common the whole cortex (or possibly, even, the whole nervous system).

But if, thus, the totality of cognitive operations is served by some general factor in common, then each different operation must necessarily be further served by some specific factor peculiar to it. For this factor also, a physiological substrate has been suggested, namely, the particular group of neurons specially serving the particular kind of operation. These neural groups would thus function as alternative "engines" into which the common supply of "energy" could be alternatively distributed."

Note that the neurospeculation that Spearman indulges in here is linked to a strongly hereditarian view of the nature of intelligence; Spearman considered the g considered essential to powering all of the specific mental engines to be an innate, inherited factor. Spearman held views consonant with those of Francis Galton and he joined Galton's Eugenics Society. He expressed the view that tests of g might one day serve Galton's original purposes for the mental tests:
"One can...conceive the establishment of a minimum index [of g] to qualify for parliamentary vote, and above all, for the right to have offspring."
Later studies do reveal rough hierarchies in the intercorrelations among subtests, but with less perfect structures than the one reported by Spearman in 1904. Further, when the statistics are recalculated from the raw data available in Spearman's 1904 paper the results are not as clear as Spearman claimed: the hierarchy of correlations is more irregular and headed by French rather than Classics. Fancher comments thus on Spearman getting his sums wrong:
"While we cannot explain the reason for these strange mistakes in Spearman's original calculations, they seem to suggest that he had a tendency to see what he wanted to in his data, sometimes at the expense of what was really there.

Stephen Jay Gould's critique of Spearman's interpretation of his results goes beyond concern with getting the numbers right. Gould's dismantling of the hereditarian view of intelligence centers on the concept of reification. Gould defines reification in the context of intelligence thus:

"the notion that such a nebulous, socially defined concept as intelligence might be identified as a "thing" with a locus in the brain and a definite degree of heritability - and that it might be measured as a single number, thus permitting a unilinear ranking of people according to the amount they possess."
Clearly, Gould is referring to Spearmans's view. Indeed, he views subsequent restatements of the hereditarian view as echoes of Spearman's delineation of g.

Jensenism

There were at least two waves of intense interest in and restatement of the hereditarian view in this century of American history prior its current resurgence in the form of Herrnstein & Murray's incendiary tract 'The Bell Curve'. The first coincided with the wave of immigration to the States subsequent to World War I. The 'psychological' side of this argument was fueled by the results of Yerkes extensive army tests. Yerkes tested 1.75 million army recruits on either a written (A) or a pictorial (B) version of intelligence tests. He announced that the average mental age of white American males was merely thirteen, just over Henry Goddard's standard for 'morons', while black American males scored on average just 10.41 years of age. He emphasized that of European immigrants, the fair skinned Western and Northern Europeans had better scores than the dark skinned Eastern and Southern Europeans. This fueled political concern about the enormous wave of immigrants flooding into the States from the war-torn countries of Europe. In 1921 Congress passed an emergency restriction act that limited immigration from any European country annually to 3% of the foreign-born of that nationality in the 1910 U.S. census. A more permanent version was passed by the House and Senate and signed into law by President Calvin Coolidge in 1924; this new legislation restricted immigration even further, setting the limits to only 2% of the foreign-born of each European country present in the U.S. census of 1890. In effect this minimized immigration from Southern and Eastern Europe because the majority of immigrants prior to 1900 originated from Northern and Western Europe.

The influence of eugenics and other "sciences" like psychometrics declined greatly after their prominent and publicized use by the Nazis in World War II. Discourse surrounding IQ, heridity, and immutable intellectual abilties fell off sharply. In the 1950s and 1960s an increased amount of attention was paid to environmental aspects of learning, intellectual growth, and opportunity. This period was the heyday of B.F. Skinner's radical behaviorism. Skinner believed that all behavior is determined by the history of its reinforcement, either positive or negative.

The second strong wave of hereditarianism linked to intelligence testing was led by the Berkeley psychologist Arthur Jensen. In 1969 Jensen published an article entitled 'How much can we boost IQ and scholastic achievement?' in the Harvard Educational Review. In this article Jensen argued that intelligence is largely heritable and immutable, thus programs like HeadStart that attempted to improve children's scholastic performance were doomed to failure. He claimed that blacks scored more poorly than whites on intelligence tests, particulary of 'g-loaded' tests. The political climate was not in tune with Jensen's views and his attempts to establish a scientific rationale for the dismantling of educational programs were roundly criticized.

The Bell Curve

In 1995 Richard Herrnstein, a Harvard psychometrician, and Charles Murray, a social science researcher at a conservative research institute published an incendiary text entitled 'The Bell Curve'. In this lengthy volume they set out the case for their social vision: America as a meritocracy with the ranking based on IQ scores. As Gould points out in his revised edition of 'The Mismeasure of Man' this is a restatement of the hereditarian argument of Galton, Spearman and Jensen, and therefore subject to the same criticisms.

Bad science

Low Correlation Coefficents
Herrnstein and Murray generally deal with correlation coeffiecients of 0.2 to 0.4. These are small effects, and they treat them as immutable. This is by no means an isolated finding; Herrnstein and Murray found the most correlated data that they could, but such measures are typically very low. For example in a recent study (Bridgeman,-Brent; Wendler,-Cathy Prediction of Grades in College Mathematics Courses as a Component of the Placement Validity of SAT-Mathematics Scores. College Board,Report No. 89-9.College Board Publications) the predictive validity of the SAT-M was studied by collecting grades from freshman mathematics courses at 10 colleges (3,499 students). The SAT-M score was a relatively poor predictor of grades in most courses. Even after correcting for the considerable range restriction that may occur when within-course scores are analyzed, coefficients were typically only in the mid 0.30s.
Scatterplots Without the Data to Mask Low Correlation Coefficients
Herrnstein and Murray masked this troublesome data by presenting their regression lines without the accompanying data that the lines were fit to, thus preventing the reader from assessing the dubious nature of the relationshipd they emphasized as fact.

Reification

This flagrant discarding of the tenets of scientific prodedure is disturbing, but even if Herrnstein and Murray had been more honest and rigorous in their procedures they would still be open to the more devastating critique of 'refication'. This is based on four shaky and interrrelated assumptions:
  1. There is a single factor of intelligence. g
  2. There is a single number, IQ, that measures g.
  3. This is inborn and heritable.
  4. IQ score is stable and permanent.

Sources: