Intelligence tests
Intelligence tests are widely used for a range of purposes, but what can they tell us? In 1994, the American Psychological Association Task Force on Intelligence summarized their findings on this subject in a review of research. They found that intelligence test performance correlates with school grades at about 0.50, total years of education about 0.55 and supervisor ratings of job performance between 0.30 and 0.50. This means that intelligence test performance is one of the best predictors we have of academic and work-related performance (Schmidt & Hunter, 1998). On the other hand, the modest magnitude of these correlations suggests that other factors – such as personality and socio-economic status – also significantly contribute to these outcomes. Indeed a correlation was found between IQ and socio-economic status of about 0.33. In addition to educational and vocational uses, intelligence tests form an important part of neuropsychological assessment for people with suspected brain injury, tumours or diseas (such as dementia) (Lezak, 1995). The Wechsler tests are now probably the most widely used individual tests of intelligence and have impressive reliability and validity. They include the WAIS (Wechsler Adult Intelligence Scale), the WISC (Wechsler Intelligence Scale for Children) and the WPPSI (the Wechsler Preschool and Primary Scale of Intelligence) – each devised very much in the spirit of the Binet scales. The tests are divided into performance and verbal subscales, and the resulting scores can be used to compute three intelligence quotients, or IQs – verbal, performance, and full scale (which is derived from combining the verbal and performance scores).
The verbal subscales, as their name suggests, usually require a verbal response and test verbal knowledge. Examples include:
tests of vocabulary;
general information, or common world knowledge (e.g. ‘In which direction does the sun rise?’);
comprehension about problem solving in daily life situations (e.g. ‘What would you do if . . . ?’);
ability to draw out similarities between objects or ideas (e.g. ‘In what way are a dog and a cat alike?’); and
ability to recall strings of numbers and/or letters.
The performance subscales do not require a verbal response. Instead they usually require the testee to manipulate pictures, objects and non-verbal symbols. Non-verbal tests are considered to be more ‘culture-free’ than other intelligence tests as they do not depend on language or culturally embedded knowledge (though it should be noted that no test can be completely ‘culture-free’). The performance tests include:
block design, in which red and white cube-shaped blocks must be arranged to match a pattern shown to the testee;
matrix reasoning, in which a series of figures is presented and a missing figure must be identified from a set of alternatives;
digit symbol, in which a list of arbitrary symbols must be replaced by their corresponding numbers under speeded conditions;
picture completion, in which a missing part of a picture must be indicated;
picture arrangement, in which a series of pictures must be arranged in an order that generates a coherent narrative; and
object assembly – essentially a jigsaw puzzle involving familiar objects.
A major difference between the Wechsler tests and Binet’s intelligence tests is the way in which IQ is calculated. The Binet tests typically calculated an IQ after first calculating a mental age (see page 272). The Wechsler scales, by contrast, calculate a deviation IQ directly from age norms – that is, IQ reflects the position in the distribution of scores obtained from a standardization sample of people of the same age.
Schmidt, F.L., & Hunter, J.E., 1998, ‘The validity and utility of selection methods in personnel psychology: Practical and theoretical implications of 85 years of research findings’, Psychological Bulletin, 124, 262–74.
Explanations for the false correlation
Why is a genetic explanation usually cited by the proponents of racial differences in intelligence, rather than the more obvious environmental explanation (such as the different socio-economic circumstances of the two groups)? The likely reason is the association of group membership with a strong genetic marker – in this case skin colour. To illustrate this point in another way, imagine a fictitious land governed by a dictator who wants a class of workers to perform the more menial roles in society. This dictator introduces the ‘regime of the hammer’, whereby every child with red hair (which, for the sake of this argument, we will presume is a highly heritable trait, just like black versus white skin colour) is tapped on the head with a hammer in a way that lowers their IQ by one standard deviation. This leads to an over-representation of red-haired people with lower IQ.
Intelligence tests are widely used for a range of purposes, but what can they tell us? In 1994, the American Psychological Association Task Force on Intelligence summarized their findings on this subject in a review of research. They found that intelligence test performance correlates with school grades at about 0.50, total years of education about 0.55 and supervisor ratings of job performance between 0.30 and 0.50. This means that intelligence test performance is one of the best predictors we have of academic and work-related performance (Schmidt & Hunter, 1998). On the other hand, the modest magnitude of these correlations suggests that other factors – such as personality and socio-economic status – also significantly contribute to these outcomes. Indeed a correlation was found between IQ and socio-economic status of about 0.33. In addition to educational and vocational uses, intelligence tests form an important part of neuropsychological assessment for people with suspected brain injury, tumours or diseas (such as dementia) (Lezak, 1995). The Wechsler tests are now probably the most widely used individual tests of intelligence and have impressive reliability and validity. They include the WAIS (Wechsler Adult Intelligence Scale), the WISC (Wechsler Intelligence Scale for Children) and the WPPSI (the Wechsler Preschool and Primary Scale of Intelligence) – each devised very much in the spirit of the Binet scales. The tests are divided into performance and verbal subscales, and the resulting scores can be used to compute three intelligence quotients, or IQs – verbal, performance, and full scale (which is derived from combining the verbal and performance scores).
The verbal subscales, as their name suggests, usually require a verbal response and test verbal knowledge. Examples include:
tests of vocabulary;
general information, or common world knowledge (e.g. ‘In which direction does the sun rise?’);
comprehension about problem solving in daily life situations (e.g. ‘What would you do if . . . ?’);
ability to draw out similarities between objects or ideas (e.g. ‘In what way are a dog and a cat alike?’); and
ability to recall strings of numbers and/or letters.
The performance subscales do not require a verbal response. Instead they usually require the testee to manipulate pictures, objects and non-verbal symbols. Non-verbal tests are considered to be more ‘culture-free’ than other intelligence tests as they do not depend on language or culturally embedded knowledge (though it should be noted that no test can be completely ‘culture-free’). The performance tests include:
block design, in which red and white cube-shaped blocks must be arranged to match a pattern shown to the testee;
matrix reasoning, in which a series of figures is presented and a missing figure must be identified from a set of alternatives;
digit symbol, in which a list of arbitrary symbols must be replaced by their corresponding numbers under speeded conditions;
picture completion, in which a missing part of a picture must be indicated;
picture arrangement, in which a series of pictures must be arranged in an order that generates a coherent narrative; and
object assembly – essentially a jigsaw puzzle involving familiar objects.
A major difference between the Wechsler tests and Binet’s intelligence tests is the way in which IQ is calculated. The Binet tests typically calculated an IQ after first calculating a mental age (see page 272). The Wechsler scales, by contrast, calculate a deviation IQ directly from age norms – that is, IQ reflects the position in the distribution of scores obtained from a standardization sample of people of the same age.
Schmidt, F.L., & Hunter, J.E., 1998, ‘The validity and utility of selection methods in personnel psychology: Practical and theoretical implications of 85 years of research findings’, Psychological Bulletin, 124, 262–74.
Explanations for the false correlation
Why is a genetic explanation usually cited by the proponents of racial differences in intelligence, rather than the more obvious environmental explanation (such as the different socio-economic circumstances of the two groups)? The likely reason is the association of group membership with a strong genetic marker – in this case skin colour. To illustrate this point in another way, imagine a fictitious land governed by a dictator who wants a class of workers to perform the more menial roles in society. This dictator introduces the ‘regime of the hammer’, whereby every child with red hair (which, for the sake of this argument, we will presume is a highly heritable trait, just like black versus white skin colour) is tapped on the head with a hammer in a way that lowers their IQ by one standard deviation. This leads to an over-representation of red-haired people with lower IQ.
This highlights two points:
1. The lower IQ of red-haired people is environmentally caused (by a hammer blow), even though it is correlated with a genetic difference. (Genetic differences lead to differences in hair colour, and it is the genes for red hair that lead to the environmental insult.) Because the genetic differences associated with hair colour are so compelling, it is likely that a genetic cause for the differences in intelligence will be seized upon and used to explain the lower IQ of people with red hair.
2. The environmental cause of the lower IQ of red-haired people is consistent with a high heritability for differences in the population as a whole. The analogy with black/ white differences in American data should be obvious. If being black means a lifetime of disadvantage and different treatment, this environmental effect could cause a difference in group means in IQ ( just like being hit on the head with a hammer) that is perfectly consistent with the idea that differences in general have a high heritability.
In case you are still finding it difficult to imagine environmental or cultural differences causing a difference as large as one standard deviation (the reliable race-related IQ difference) in the face of a high heritability for IQ, consider the ‘Flynn effect’. Flynn (1987) has shown that there has been a standard deviation increase, per generation, in the mean level of intelligence test performance for most of this century in Western society, which equates to about three IQ points per decade. Such a difference has to be the result of an environmental change, because gene frequencies in populations could not change so quickly. If there can be shifts of one standard deviation in IQ between generations that are environmental in origin, despite the high heritability of IQ differences, why could there not be similar environmentally mediated differences between populations within the current generation?
1. The lower IQ of red-haired people is environmentally caused (by a hammer blow), even though it is correlated with a genetic difference. (Genetic differences lead to differences in hair colour, and it is the genes for red hair that lead to the environmental insult.) Because the genetic differences associated with hair colour are so compelling, it is likely that a genetic cause for the differences in intelligence will be seized upon and used to explain the lower IQ of people with red hair.
2. The environmental cause of the lower IQ of red-haired people is consistent with a high heritability for differences in the population as a whole. The analogy with black/ white differences in American data should be obvious. If being black means a lifetime of disadvantage and different treatment, this environmental effect could cause a difference in group means in IQ ( just like being hit on the head with a hammer) that is perfectly consistent with the idea that differences in general have a high heritability.
In case you are still finding it difficult to imagine environmental or cultural differences causing a difference as large as one standard deviation (the reliable race-related IQ difference) in the face of a high heritability for IQ, consider the ‘Flynn effect’. Flynn (1987) has shown that there has been a standard deviation increase, per generation, in the mean level of intelligence test performance for most of this century in Western society, which equates to about three IQ points per decade. Such a difference has to be the result of an environmental change, because gene frequencies in populations could not change so quickly. If there can be shifts of one standard deviation in IQ between generations that are environmental in origin, despite the high heritability of IQ differences, why could there not be similar environmentally mediated differences between populations within the current generation?
No comments:
Post a Comment