Wtf is reification?

As the reactionary scientists failed to establish valid biophysical explanations for social inequality by measuring skulls both on the outside (using calipers and rulers) and the inside (using mustard seed and lead shot), they moved into the realm of measuring “the content of brains by intelligence testing” (23, revised edition). It is here that the error that perhaps receives the bulk of Gould’s attention is to be found: reification—the process of treating as a real entity something that is in fact an abstract concept. Cyril Burt and current advocates of the notion that intelligence is literally a one-dimensional feature of the brain that is measurable by psychometric tests are guilty of reifying their own constructions. Those guilty of reifying IQ argue that there is a general underlying intellectual ability in each of us, g, that is measured reasonably well by IQ tests, in spite of the evidence suggesting that g is a product of the tests themselves, a statistical creation, not a genuine mental attribute.

In an unflinchingly rational manner, Gould devastates this “IQ as indicator of general intelligence” interpretation by showing it to be a creation of the statistical procedures used and the a priori convictions of the researchers. The general intelligence factor emerges from factor analysis of a variety of mental tests. Factor analysis is a statistical procedure that attempts to explain the covariance among variables (various mental tests in this case) by extracting one or a few factors that can account for the observed inter-correlations (individuals’ scores on different tests tend to be positively correlated with one another—i.e., people who do well on one type of test tend to do well on other types). IQ proponents have long argued that only one factor is necessary to explain observed correlations among a variety of mental tests, which they take to indicate the existence of a general intelligence that is an actual characteristic of the brain. However, as Gould explains, factor analysis does not work magic; it is entirely based on the observed correlations among tests. The belief that a factor extracted via factor analysis is a real entity is based on the heroic assumption that the variables under analysis (performance on various mental tests in this case) are connected by an underlying causal regime (stemming from a feature of the brain). This assumption is not and cannot be established by statistical methods alone and is only valid to the extent that correlation is indicative of causation. Although determining correlation is necessary for establishing a causal relationship among variables, it is not sufficient. Factor analysis alone simply cannot adjudicate the matter of causality, nor establish whether a factor corresponds with a real entity.