“What are the benefits and drawbacks of psychometric intelligence measurements on the theoretical development of an understanding of ‘intelligence’?”
“ALL ATTEMPTS TO DEVELOP AMBITIOUS THEORIES OF INTELLIGENCE HAVE FAILED.” Laquer (1985, p.8)
Researchers widely disagree on a common definition of intelligence (Sternberg & Berg, 1986), conceptually describing it as “a general capacity for inferring and applying relationships drawn from experience” (Herrnstein & Murray, 1994), emphasizing that “it is not merely book learning [ ]; [it is] a broader and deeper capability for comprehending our surroundings [ ]” (Gottfredson, 1997). In stark contrast are more operational definitions based on psychometric intelligence measures, such as “Intelligence is what the tests test” (Boring, 1923), or “a person’s score on a statistically determined set of questions” (Herrnstein & Murray, 1994). Although having supported the theoretical development of intelligence, psychometric tests have been criticized for their over-simplification, inaccuracies and potential misuse (Nisbett et al, 2012). This essay will focus on evaluating potential benefits and drawbacks of using psychometric intelligence tests to help define intelligence, concentrating on the most dominant ‘IQ’ test, for which the bulk of evidence exists.
Attempts at defining human intelligence are dependent on the assessment of adequate concepts, investigative tools and statistical techniques used. Hettema and Deary (1993) suggest that it is possible to distinguish between six different approaches, which yield different perspectives on intelligence (psychometric, experimental-cognitive, psychophysics, psychophysiology, physiology and genetic research), each of which unsuccessfully attempts to devise a common definition of intelligence. Investigations of individual differences may aid in theoretical development, but require a meaningful quantification of intelligence. This is only achieved by the psychometric approach, which assumes that intelligence in the population is distributed in a bell curve, ranking people from lowest to highest, as well as presuming a single, underlying intelligence factor (Gardner, 1995). These assumptions allow the allocation of an intelligence score to each person by converting her/his rank into a number (Gould, 1981). The strength of quantifying intelligence is demonstrated when administering a range of psychometric tests to subjects, as performance differences on an individual level can be measured, with the first principal component of the correlation matrix of these tests resolving around 50-60% of all information (Gould, 1981). This finding led to the theoretical development of Spearman’s (1904) ‘g’ (subsequently developed to his ‘two-factor theory’), which emphasized that each test reflects the operation of a single underlying factor ‘g’ (the first principal component), as well as some specific information (‘s’). Towler et al (1993) argue that although psychometric factors must not necessarily be represented analogously within the brain, they can sometimes suggest a biological structure or functional arrangement. Therefore, although Spearman’s ‘g’ may have little substance, it is perfectly sensible to investigate whether it may arise from some biological variance in the nervous system (Deary, 2000). In fact, evidence from physiological studies from speech and language research indicates a certain degree of specificity and modularity in the brain (Deary, 2000), indicating intelligence measurements’ position as a viable aid in the development of a further theoretical understanding of intellect. Moreover, due to the considerable stability of test scores across a person’s lifespan, psychometric intelligence tests have also been used as a measure of state-like indicators of cognitive performance when brain function is impaired or changing, for example in cases of dementia (Deary, 2000).
Psychometric tests were originally devised for the practical purpose of identifying pupils who did not benefit from standard schooling and required special education (Gould, 1981). Appointed by the French ministry of public education, Binet (1903) developed a psychometric test named ‘Binet’s Scale’. It was Binet’s firm belief that intelligence is too complex to capture with a single number, with his scale not designed to support any theory of intellect. Pupils who scored poorly on Binet’s Scale were identified as requiring additional support and consequently received special education. Over time, these pupils were able to augment their test scores, emphasizing Binet’s belief that low-scoring children were not innately incapable. Subsequently, Binet’s scale was further refined into the more well-known intelligence quotient (IQ) test (Stern, 1912). Inherent in Stern’s (1912) scale retitling is the belief that IQ tests measure Spearman’s ‘g’ with fair accuracy (Gould, 1981), thus introducing the idea that psychometric tests measure intelligence. Other researchers have continuously developed this notion, starkly opposing Binet’s belief of tests as an empirical device with a limited practical purpose. Besides findings which have shown that IQ tests, as in their original purpose, are a viable predictor of scholastic performance (Gottfredson, 2004), further research has focussed on correlational studies of IQ test scores and other social outcomes, such as occupational performance, income, jail time and work disability (Herrnstein & Murray, 1994; Neisser et al, 1996). Most correlation scores (r) fall within the range of 0.4 - 0.6 (Jensen, 1975; Brody, 1985), and even when controlling for parental social class, IQ measurements show predictive validity. As argued by Deary (2000), despite tests not accounting for the majority of the variance in the variables they predict, social sciences in general do not usually have a single variable which fully or even largely determines others, with IQ-type tests being the single best predictive variable available. These findings have been taken as support for the idea that IQ tests measure intelligence, with the implicit assumption that those with higher intelligence will also have more favourable social outcomes. Through the diagnosis of individual differences in intelligences, as assessed via IQ tests, it may be possible to broaden theoretical developments of intelligence by analysing differences and similarities between individuals who score similarly.
One main criticism directed at psychometric measurements of intelligence focuses on whether intelligence can be measured and quantified at all. Unless intelligence has been defined, thus determining what tests should measure, there are great difficulties in evaluating whether or not tests are accurate, i.e. measuring what they are supposed to assess. Even Burt (1921), an advocate of psychometric testing, declared the inability of a mere number, calculated during a short series of tests, to capture the complex notion of intelligence.