Intelligence and IQ: What teachers should know

Abstract

We review past and current psychometric theories about intelligence and critically evaluate the usefulness of modern IQ tests in guiding decisions within an educational context. To accomplish this we consider whether knowledge about intelligence extends beyond mere description to provide a scientific framework for further advancing our understanding. We conclude that it does. We also conclude that current evidence supports the importance of general ability, as well as several different specific abilities, although whether emotional intelligence can yet be affirmed is not clear. Additionally, we conclude that creativity is something separate from intelligence. Despite strong evidence that intelligence and IQ must be different constructs, we conclude that the latter provides the best available means for investigating and making decisions about the former, with higher validity for this purpose than has frequently been realised. We therefore recommend aptitude and achievement testing as useful tools for educational settings, provided they sample a broad range of different intellectual domains in addition to general ability. We also emphasise the importance of such tests being culturally compatible with the child’s background.

Notes

1. Over the past 20 years there has been a growing trend to regard all forms of validity as subsumed under “construct validity”, which is concerned with confirming the theory that IQ measures intelligence (Messick, 1980). Ultimately, validity can only be demonstrated in concrete terms by a correlation coefficient, and it makes little sense to rely on just one coefficient as the validity for a test. Thus, construct validity embraces all other types, principally predictive (correlates with future performance) and concurrent (simultaneous measurement). The American Psychological Association’s standards manual combines these as “criterion validity”. Various forms of validity must first be determined in order to establish construct validity. It is important, too, to appreciate that even a low validity coefficient has potential benefit. In the general sense that a test measures to some extent what it has been designed to measure, the test will improve on random, chance procedures. As discussed throughout this article, omnibus IQ tests have long developmental histories and they have convincing construct validity because IQ tests intercorrelate with each other but not with tests that are not IQ tests; they predict academic and other life achievements after partialling out social class; they discriminate occupational groups defined in terms of intelligence requirements; factor analysis establishes a strong general factor, correlated with IQ, on which all subtests load; IQ is stable across the lifetime; and IQ has high heritability (see Note 5), particularly beyond childhood.

2. Although theoretically distinguishable, the practical differences between so‐called “aptitude” and “achievement” tests are frequently confused. Educational selection tests like the Scholastic Aptitude Test in the US have frequently been described as achievement tests. Similarly, the Woodcock‐Johnson battery, the content of which is consistent with other IQ tests, has been so described.

3. This result has not been corrected upwards to take account of either test unreliability or restricted range of scores. The correlation therefore means that, at minimum, about 50% of variability in later scores was accounted for across most of the lifespan by individual differences in the initial measure.

4. Gf stands for “general fluid ability”. Similarly Gc stands for “general crystallised ability”. Cattell initially proposed that these two broad group factors should substitute for Spearman’s g. In later versions of the theory, expanded primarily by J. L. Horn, these two factors have remained paramount.

5. Inspection time (IT) is a measure of processing speed unconfounded by motor speed because it is a threshold estimate of the time required by an individual to make a simple judgement with specified high accuracy. Thus, speed of reaction is irrelevant. Individual differences in IT are highly reliable, moderately heritable, and share genetic influences on IQ. Teachers reliably identify individual differences in IT. Unlike IQ, childhood estimates are also stable across generations. IT and IQ among both children and adults correlate about −0.5 (slower times with lower IQs). For further details see contributors to Intelligence (2001, Vol. 29, Special Issue: Inspection time).

6. Broad heritability is the proportion of total variation in IQ in a population that is explained by genetic variation. Heritability will vary across time because of changed circumstances, diminishing where environmental influences rise and increasing where environmental influences fall. It is possible to partition broad heritability, which includes all sources of genetic variation, into narrow (“additive”) influences and “nonadditive” sources. Additive influences are genes critical to the expression of the parental trait in the offspring. Nonadditive sources include genetic–environmental confounding, indirect genetic influences on IQ from personality variables, and differences in mood and motivation. Including such nonadditive sources therefore inflates estimates of broad heritability.