|
Article Excerpt
It has frequently been asserted that there is no sex difference in average general intelligence but that the variance is greater in males. In this paper we examine these two propositions by a meta-analysis of studies of sex differences on the Progressive Matrices among university students. We find that both are incorrect.
The assertion that there is no sex difference in average general intelligence has been made repeatedly since the early decades of the twentieth century. One of the first to adopt this position was Terman (1916, pp.69-70) who wrote of the American standardization sample of the Stanford-Binet test on approximately 1,000 4- to 16-year-olds that girls obtained a slightly higher average IQ than boys but 'the superiority of girls over boys is so slight ... that for practical purposes it would seem negligible'. In the next decade Spearman (1923) asserted that there is no sex difference in g. Cattell (1971, p. 131) concluded that, 'it is now demonstrated by countless and large samples that on the two main general cognitive abilities--fluid and crystallized intelligence--men and women, boys and girls, show no significant differences'. Brody (1992, p. 323) contended that, 'gender differences in general intelligence are small and virtually non-existent'. Jensen (1998, p. 531) calculated sex differences in g on five samples and concluded that, 'no evidence was found for sex differences in the mean level of g'. Similarly: 'there is no sex difference in general intelligence worth speaking of' (Mackintosh, 1996, p. 567); 'the overall pattern suggests that there are no sex differences, or only a very small advantage of boys and men, in average IQ scores' (Geary, 1998, p. 310); 'most investigators concur on the conclusion that the sexes manifest comparable means on general intelligence' (Lubinski, 2000, p. 416); 'sex differences have not been found in general intelligence' (Halpern, 2000, p. 218); 'we can conclude that there is no sex difference in general intelligence' (Colom, Juan-Espinosa, Abad, & Garcia, 2000, p. 66); 'there are no meaningful sex differences in general intelligence' (Lippa, 2002); 'there are negligible differences in general intelligence' (Jorm, Anstey, Christensen, & Rodgers, 2004, p. 7); and 'the evidence that there is no sex difference in general ability is overwhelming' (Anderson, 2004, p. 829).
The question of whether there is a sex difference in average general intelligence raises the problem of how general intelligence should be defined. There have been three principal answers to this question. First, general intelligence can be defined as the IQ obtained on omnibus intelligence tests such as the Wechslers. The IQ obtained from these is the average of the scores on a number of different abilities including verbal comprehension and reasoning, immediate memory, visualization, and spatial and perceptual abilities. This is the definition normally used by educational, clinical, and occupational psychologists. When this definition is adopted, it has been asserted by Halpern (2000) and reaffirmed by Anderson (2004, p. 829) that 'the overall score does not show a sex difference' Halpern (2000, p. 90). Second, general intelligence can be defined as reasoning ability or fluid intelligence. This definition has been adopted by Mackintosh (1996, p. 564; Mackintosh, 1998a) who concludes that there is no sex difference in reasoning ability. Third, general intelligence can be defined as the g obtained as the general factor derived by factor analysis from a number of tests. This definition was initially proposed by Spearman (1923, 1946) and was first adopted to analyse whether there is a sex difference in g by Jensen and Reynolds (1983). They analysed the American standardization of the WISC-R on 6- to 16-year-olds and found that this showed boys to have a higher g by d = .16 (standard deviation units), equivalent to 2.4 IQ points (this advantage is highly statistically significant). In a second study of this issue using a different method for measuring g, Jensen (1998, p. 539) analysed five data sets and obtained rather varied results, in three of which males obtained a higher g than females by 2.83, 0.18, and 5.49 IQ points, while in two of which females obtained a higher g than males by 7.91 and 0.03 IQ points. Jensen handled these discrepancies by averaging the five results to give a negligible male advantage of .11 IQ points, from which he concluded that there is no sex difference in g. This conclusion has been endorsed by Colom and his colleagues in Spain (Colom, Garcia, Juan-Espinoza, & Abad, 2002, Colom et al., 2000).
Thus there has evolved a widely held consensus that there is no sex difference in general intelligence, whether this is defined as the IQ from an omnibus intelligence test, as reasoning ability, or as Spearman's g. However, this consensus has not been wholly unanimous. Dissent from the consensus that there is no sex difference in general intelligence has come from Lynn (1994, 1998, 1999). The starting-point of his dissenting position was the discovery by Ankney (1992) and Rushton (1992), using measures of external cranial capacity, that men have larger average brain tissue volume than women, even when allowance is made for the larger male body size (see also, Gur et al., 1999). That men have a larger cerebrum than women by about 8-10% is now well established by studies using the more precise procedure of magnetic resonance imaging (Filipek, Richelme, Kennedy, & Caviness, 1994; Nopoulos, Flaum, O'Leary, & Andreasen, 2000; Passe et al., 1997; Rabinowicz, Dean, Petetot, & de Courtney-Myers, 1999; Witelson, Glezer, & Kigar, 1995). The further link in the argument is that brain volume is positively associated with intelligence at a correlation of .40. This has been shown by Vernon, Wickett, Bazana, and Stelmack (2000, p. 248) in a summary of 14 studies in which magnetic resonance imaging was used to estimate brain size, as compared with previous studies based on measures of external cranial capacity (Jensen & Sinha, 1993). It appears to follow logically that males should have higher average intelligence than females attributable to their greater average brain tissue volume. It has been argued by Lynn (1994, 1998, 1999) that this is the case from the age of 16 onwards and that the higher male IQ is present whether general intelligence is defined as reasoning ability or the full scale IQ of the Wechsler tests, and that the male advantage reaches between 3.8 and 5 IQ points among adults. Lynn maintains that before the age of 16, males and females have approximately the same IQs because of the earlier maturation of girls. Lynn attributes the consensus that there is no sex difference in general intelligence to a failure to note this age effect. Further evidence supporting the theory that males obtain higher IQs from the age of 16 years has been provided by Lynn, Allik, and Must (2000), Lynn, Allik, Pullmann, and Laidra (2002), Lynn, Allik, and Irwing (2004), Lynn and Irwing (2004) and Colom and Lynn (2004). The only independent support for Lynn's thesis has come from Nyborg (2003, p. 209), who has reported on the basis of research carried out in Denmark that there is no significant sex difference in g in children, but among adults, men have a significant advantage of 5.55 IQ points. Most authorities, however, have rejected this thesis, including Mackintosh (1996, p. 564, Mackintosh (1998a, 1998b), Jensen (1998, p. 539), Halpern (2000), Anderson (2004, p. 829) and Colom and his colleagues (Colom et al., 2002, 2000).
There is a large amount of research literature on sex differences in intelligence including books by Caplan, Crawford, Hyde, and Richardson (1997), Kimura (1999) and Halpern (2000). This makes integrating all the studies and attempting to find a solution to the problem of sex differences an immense task. We believe that the way to handle this is to examine systematically the studies of sex differences in intelligence in each of the three ways defined above (i.e. as the IQ in omnibus tests, in reasoning, and in g). To make a start on this research programme we have undertaken the task of examining whether there are sex differences in non-verbal reasoning assessed by Raven's Progressive Matrices. The Progressive Matrices is a particularly useful test on which to examine sex differences in intelligence defined as reasoning ability, because it is one of the leading and most frequently used tests of this ability. The test has been described as 'the paradigm test of non-verbal, abstract reasoning ability' (Mackintosh, 1996 p. 564). It is also widely regarded as the best or one of the best tests of Spearman's g, the general factor underlying all cognitive abilities. This was asserted by Spearman himself (1946) and confirmed in an early study by Rimoldi (1948).
By the early 1980s Court (1983, p. 54) was able to write that it is 'recognised as perhaps the best measure of g'; and some years later Jensen (1998, p. 541) wrote, that 'the Raven tests, compared with many others, have the highest g loading'. In a recent factor analysis of the Standard Progressive Matrices administered to 2,735 12- to 18-year-olds in Estonia, Lynn et al. (2004) showed the presence of three primary factors, and a higher order factor, which was interpreted as g, and correlated at .99 with total scores, providing further support that scores on the Progressive matrices can be identified with g. Sex differences on the Progressive Matrices should, therefore, reveal whether there is a difference in g as well as whether there is a sex difference in reasoning ability.
The Progressive Matrices test consists of a series of designs that form progressions. The problem is to understand the principle governing the progression and then to extrapolate this to identify the next design from a choice of six or eight alternatives. The first version of the test was the Standard Progressive Matrices constructed in the late 1930s as a test of non-verbal reasoning ability and of Spearman's g (Raven, 1939) for the ages of 6 years to adulthood. Subsequently, the Coloured Progressive Matrices was constructed as an easier version of the test designed for children aged 5 through 12; and later the Advanced Progressive Matrices was constructed as a harder version of the test designed for older adolescents and adults with higher ability.
The issue of whether there are any sex differences on the Progressive Matrices has frequently been discussed and it has been virtually universally concluded that...
|
