“After years of trying to explain comparative female academic success, a leading education expert has concluded that girls are simply cleverer than boys”, so reads an article in the Times, 12 Aug’21.
“When girls leapt ahead at GCSE”, he said, “this was attributed to the modular structure of the course, which favoured their conscientious approach, and when GCSEs were reformed to be more like O-levels, the girls’ lead was dented only slightly”.
The “leap ahead” experienced by girls when O Levels were replaced by GCSEs is shown in Figure 1, below. This “leap ahead” is either a result of the change in the nature of the award or a remarkable coincidence that it happened at the same time, 1987/88. To fail to acknowledge that this is evidence that the origin of the sex difference in attainment lies in the nature of the award appears to be wilful denial.
As for the reform of GCSEs making them “more like O Levels” I shall post some old O Level papers in comparison with recent GCSE papers so readers can make up their own mind just how comparable they may or may not be. (I’ll not do so with this post to avoid delay as the scanning will take some time).Figure 1: Percentage of School Leavers Achieving 5+ A-C (or Pass) O levels or A*-C GCSEs, by Sex (1962 – 2006) from Gender and education: the evidence on pupils in England: Department for Education and Skills, 2007 (Figure 3-7).
Professor Smithers continues,
“Now with teacher assessment girls have gone further ahead and it has been suggested that this is because teachers favour them. On the strength of their superior performance in schools, girls are more likely to go to university. In 1980, they were outnumbered by three to two but 40 years on that has been a complete reversal.”
It is not a “suggestion” that teachers’ assessments favour girls, but clear and unambiguous evidence from the data. I have been pointing this out in the context of Key Stage 2 SATS (age 10/11) for 7 years, and 2020 provided clear evidence that this bias also exists in A Levels when based on teachers’ assessments. (I shall update this analysis for the 2021 results shortly, and include GCSEs in that analysis). My latest update of the KS2 SATS analyses are given in Figures 2 and 3 (positive bias means in favour of girls).
Figure 2: Teacher Bias in KS2 SATS (reading & writing)
Figure 3: Teacher Bias in KS2 SATS (maths)
The good professor asks,
“Why then does it seem so difficult to accept that females are cleverer?”
One might have thought that a man with an MSC and a PhD in the Psychology and Sociology of Education might have had a suspicion that psychology and sociology could have an impact on education. But it seems not. Perhaps it’s with his being male – he’s just not very bright. Oh dear, what a pity, never mind, allow me to assist. Just why is it difficult to accept that females are cleverer?
Well, what on earth does the Professor mean by “cleverer”. With a PhD in a psychological subject he should know that such terms are undefined until a measurement procedure is specified. If the measurement procedure is defined by the attainment of education awards sanctioned by the JCQ then I grant you, it’s a slam-dunk. (Do note I write “awards”, not “exams”). Girls are doing very well in attaining these awards, in comparison with boys, so if that’s the definition of “cleverer” that’s the end of the matter.
But hold on – the good Professor is claiming that the reason why girls are doing better in these awards is because they are cleverer. I declare a foul. The explanation is void if you use circular logic. In order for the Professor’s claim that girls are cleverer to provide an explanation of their superior attainment of awards, there must be some independent measure of “cleverness” which makes the assertion meaningful. What the Dickens could it be? Hmm…what might a person versed in psychology and educational matters think of as a measure of “cleverness”. Tricky, eh?
Not really, no. IQ, of course (or, better, “g” – General Intelligence).
There is a vast literature on IQ, much of it on sex differences. And whilst there is some contention around sex differences in IQ, these pale into insignificance compared to the effects of ethnicity and country-of-origin on IQ. Thankfully we don’t need to go there. And in respect of sex, the controversies are over the minutiae, not the broad perspective – at least as regards the vast bulk of people. Here’s a short lecture on sex differences in IQ…
Firstly there is the thorny issue of whether IQ is a well defined quantity. There are many different variants of test, and differences in results between test types are inevitable but notable, in general, for being small, the test results being very strongly correlated. Nevertheless, there will remain a concern that, where only very small sex differences are apparent, that these differences are an artefact of the particular tests used.
This situation is exacerbated by the one finding that appears consistent over time, consistent across cultures, and broadly (if not invariably) consistent between test variants: namely that sub-categories of test question show sex differences. This is illustrated by Table 1, taken from the meta-analysis by Hedges and Nowell, (1995) Sex differences in mental test scores, variability, and numbers of high-scoring individuals, Science, 269, 41-45. The magnitudes of the sex differences illustrated by Table 1 will vary between studies, but the signs of the sex differences are pretty robust. Some aspects of verbal skills/comprehension, perceptual speed and associative memory tend to favour women and girls, whereas mathematics and visuospatial skills favour men and boys.
Table 1: from Hedges and Nowell, (1995), Science, 269, 41-45
Sex differences in cognitive ability are easily greatest in spatial ability, and especially in mental rotation, where males outperform females by almost a standard deviation. To quote N.J.Macintosh, Sex Differences and IQ, J.biosoc.Sci. (1996) 28, 559-571: “The largest, most reliable and most persistent difference between the sexes is that observed in tests of spatial ability. As already noted, tests of three-dimensional mental rotation yield a difference of some 13-14 IQ points – a difference which, unlike many others, has shown no signs of decreasing in the past 20 years…. and while the determined environmentalist will obviously be able to dismiss this, it does suggest that the difference is not simply a consequence of cultural attitudes.”
Standard IQ tests contain a mix of question types, with the overall IQ score being determined by the aggregate result. Obviously, then, whether a sex difference is apparent in the resulting IQ score will depend upon the particular mix of question types adopted. Quoting Macintosh, Sex Differences and IQ, again,
“Although the overall difference between the sexes was trivial, there were some items or sub-tests on which females consistently did better than males, and others on which males consistently obtained higher scores than females. The two happened more or less to cancel each other out to yield approximate overall equality. But it now seems obvious that a judicious choice of sub-tests for inclusion in one’s test battery could yield any outcome one wanted. Indeed, feminists inclined to see male conspiracies lurking everywhere may think that this is just what has happened with the revisions of the Wechsler tests, for both the WISC-R and WAIS-R now yield a significant overall male superiority.”
Those last-mentioned test types feature large in the analyses which I quote below. However, this ambiguity due to the arbitrariness of weighting between question types is minimised in what is referred to as General Intelligence, denoted “g” rather than IQ. This is measured by essentially the same types of test but analysed differently. The general ability factor, g, is technically defined as the unrotated first factor or principal component of a factor analysis or principal components analysis. However, whilst g minimises dependence on the arbitrary mix of questions it does not eliminate it entirely. Some authors claim that g is “very close to the IQ calculated from a battery of diverse tests”, for example Richard Lynn and Gerhard Meisenberg, Sex Differences in Intelligence (2016), Mankind Quarterly 57(1):5-8.
For the consensus view of sex differences in intelligence as it stood in the mid-90s we can again quote Macintosh’s review, Sex Differences and IQ,
“The differences in the original standardisation samples for the two tests were small, 1.7 points for the WISC-R and 2.2 for the WAIS-R. But Lynn (1994) reviewed a number of other large scale studies of the Wechsler tests which consistently found a significant male superiority, averaging 2.35 points on the WISC and 3.08 points on the WAIS. There can be little doubt that the sex difference on these tests is reliable – and slightly larger for adults on the WAIS than for children on the WISC. But is it real, or at least, typical? One might suppose that the best way to answer the question is to look at the results obtained with other general IQ test batteries on large, representative samples of the population. The answer turns out to be somewhat equivocal. Thus Herrnstein & Murray (1994) obtained the test scores of some 12,000 teenagers and young adults on the AFQT test and found a difference of 0.9 IQ points in favour of men. But Lubinski & Humphreys (1990) analysed the test scores of some 100,000 16-year-old American schoolchildren, and found a difference of 0.3 IQ points in favour of girls; while the 1980 standardisation of the Differential Aptitude Tests, on a representative sample of American 14-18-year-olds, yielded an overall difference of 0.8 IQ points in favour of females (Feingold, 1988).”
The latter, apparently conflicting results, would seem to be an age effect. Introducing the special edition of Mankind Quarterly, September 2016, 57(1), Richard Lynn and Gerhard Meisenberg in Sex Differences in Intelligence note that the consensus position is for negligible sex difference in IQ (or g), but state that, “This special issue presents a number of papers challenging the widespread consensus that there is no sex difference in general intelligence”. They opine that, “while there is virtually no sex difference up to the age of 16 years, from this age onwards males develop an advantage that increases with age reaching approximately 4 IQ points among adults (Lynn, 1994). Further data documenting this male advantage was given in Lynn (1999) and in a meta-analysis of sex differences on the Progressive Matrices by Lynn and Irwing (2004) concluding that among adults males obtain a 5 points higher IQ than females.”
Richard Lynn has been a bullish proponent of adult male advantage in general intelligence for the last 27 years or so. He has a particular theory of cognitive development. The theory states that boys and girls mature at different rates such that the growth of girls accelerates at the age of about 9 years and remains in advance of boys until 14–15 years. At 15–16 years the growth of girls decelerates relative to boys. As boys continue to grow from this age their height and their mean IQs increase relative to those of girls.
If correct, this theory would caution against drawing any conclusions for adults from tests on school-age children. This is important because some papers on sex differences are indeed based on school age children and so potentially misleading. The theory seems to have some merit in respect of the commonplace observation that girls at primary school have always outperformed boys, but that as adults men catch up and potentially exceed women in ability (see Figures 5 and 6 of this post).
In support of this theory, Roberto Colom and Richard Lynn, in Testing the developmental theory of sex differences in intelligence on 12–18 year olds (Spanish data), Personality and Individual Differences, 36(1), 75-82, Jan 2004, conclude,
(Abstract extract) “This paper presents new evidence for the theory from the Spanish standardization sample of the fifth edition of the DAT. 1027 boys and 924 girls between 12 and 18 years were tested. The general trend shows that girls do better at the younger ages and their performance declines relative to boys among older age groups, which supports the developmental theory. The sex difference for the DAT as a whole for 18 year olds is a 4.3 IQ advantage for boys, very close to the advantage that can be predicted from their larger brain size (4.4 IQ points). The profile of sex differences in abilities among the Spanish sample is closely similar to that in the United States and Britain, which is testimony to the robustness of the difference in these different cultures.”
The Abstract of the influential, if now rather old, meta-analysis by Hedges and Nowell, (1995) Sex differences in mental test scores, variability, and numbers of high-scoring individuals, Science, 269, 41-45, appears supportive. It reads,
“Sex differences in central tendency, variability, and numbers of high scores on mental tests have been extensively studied. Research has not always seemed to yield consistent results, partly because most studies have not used representative samples of national populations. An analysis of mental test scores from six studies that used national probability samples provided evidence that although average sex differences have been generally small and stable over time, the test scores of males consistently have larger variance. Except in tests of reading comprehension, perceptual speed, and associative memory, males typically outnumber females substantially among high-scoring individuals.”
I now present the results from the special issue of Mankind Quarterly, September 2016, 57(1) but with a warning that many of the papers include Richard Lynn as a co-author. I do not suggest that this impacts on their reliability but it may well have a bearing on the type of testing and analysis performed.
Davide Piffer, Sex Differences in Intelligence on the American WAIS-IV,
Abstract: “Data are reported for the scores of men and women in the standardization of the American WAIS-IV. Men obtained a significantly higher Full Scale IQ than women by 2.25 IQ points and on the General Ability Index by 4.05 IQ points. Men obtained significantly higher scores on the index IQs of Verbal Comprehension, Perceptual Reasoning and Working Memory, and women obtained a significantly higher score on the Processing Speed index IQ. Men showed greater variability than women on the Full Scale IQ, the General Ability Index and on twelve of the fifteen subtests.”
Salaheldin Farah Attallah Bakhiet et al, Sex Differences in the Intelligence of University Engineering Students in Sudan,
Abstract: “The intelligence of 1936 engineering students in three universities in Sudan was tested with the Advanced Progressive Matrices. The sample obtained an average British IQ of 93. Males obtained marginally higher average scores than females, equivalent to approximately 1.2 IQ points.”
George Spanoudis, et al, Sex Differences for 10 to 17 Year Olds on the Standard Progressive Matrices in Cyprus,
Abstract: “Sex differences on the Standard Progressive Matrices are reported for 10 to 17 year olds in Cyprus. There were no significant differences among 10 to 16 year olds but among 17 year olds males obtained a mean IQ 4.4 points higher than females.”
Yoon-Mi Hur, Jan te Nijenhuis and Hoe-UK Jeon, Testing Lynn’s Theory of Sex Differences in Intelligence in a Large Sample of Nigerian School-Aged Children and Adolescents (N >11,000) using Raven’s Standard Progressive Matrices Plus,
Abstract: “Sex differences in intelligence have been much disputed for many decades. The present study examined the issues of whether sex differences in intelligence change during development. In total, 11,164 children (mean age = 13.5 years; SD = 2.6 years) completed the Standard Progressive Matrices Plus (SPM+). From age 8 to 19 years, sex differences in the total score of the SPM+ increased from -0.06d (favoring females) to 0.46d (favoring males), with an average of 0.23d. Our findings support Lynn’s developmental theory of sex differences in cognitive abilities.” (NB: 0.23d is about 3.5 IQ points).
Hsin-Yi Chen, Richard Lynn and Helen Cheng, Sex Differences on the WISC-III in Taiwan and the United States,
Abstract: “Sex differences on the WISC-III are reported for the 13 subtests, the Verbal and Performance IQs, the four Index IQs and the Full Scale IQs in Taiwan and the United States. The sex differences are closely similar in the two samples with a correlation of .87 (p<.001) in the 13 subtests. Males obtained significantly higher Full Scale IQs in the two samples of .21d and .11d, respectively.” (NB: 0.21d corresponds to about 3 IQ points).
Flores-Mendoza, Darley and Fernandes, Cognitive Sex Differences in Brazil,
Abstract (extract): “… we reviewed and compiled intelligence data from Brazil, and found evidence for a male advantage in adulthood in most dimensions of intelligence. We then analyzed new adult data on three unidimensional IQ tests for the measurement of general intelligence (the g factor), and found evidence of a male advantage in two, but a female advantage in the third. However, scores on two tests appeared to be highly confounded with education level, and once this variable was controlled, the female advantage in one test and the male advantage in another were not noticeable.. In general, our results were mostly in line with the male advantage hypothesis, although this did not appear to be uniformly consistent or of high magnitude in Brazil.”
I am not attempting to present a definitive case for male advantage in intelligence, despite the clear direction of the above studies. My aim is the more modest one of definitively answering Professor Smithers’ question “Why does it seem so difficult to accept that females are cleverer?” To put an alternative to the Richard Lynn point of view I refer you to the above quote by Macintosh, and some further observations of his which are probably still apposite despite their vintage…
“…it will be argued that performance on any one of these tests is correlated with performance on any other: hence, general intelligence, (may be) defined as g, or the first principal component of any suitably diverse battery of mental tests. Is there not a sex difference in g? The answer is still: it depends. The general factor extracted from the Wechsler tests yields a difference between the sexes, in favour of males, that is actually some 40% larger than their difference in average scores (Jensen & Reynolds, 1983). Thus the 2.3-point advantage for males in overall IQ on the WAIS would again, to Lynn’s satisfaction, translate into a 4-point male advantage on the general factor extracted from the test. The reason for this is that the various sub-tests of the WAIS differ rather widely in their loadings on this general factor and there is some tendency for males to do better on sub-tests with higher g-loadings.”
Macintosh also takes a dim view of the claims for a larger male variance,
“…there would then be equally little reason to believe in that other hoary explanation of greater male eminence – greater variability in general intelligence. Neither Raven’s matrices nor the DAT reasoning tests reveal reliable evidence that the variance of male scores is greater than that of females (Court, 1983; Feingold, 1992).”
All this might seem like much ado about nothing as the differences in question are so small, and this is indeed the case as regards the vast bulk of the populace. Many of the above sex differences are fractions of an IQ point, and even the larger ones are just a few IQ points. To the general population this is hardly going to be perceptible. However, where these small differences become amplified into a major effect is in the tail of the distribution. The IQ Comparison Site illustrates this. I illustrate it in Figure 4 in terms of the ratio of men to women at high IQ for the cases,
- Sex difference in mean 5 IQ points (same standard deviation)
- Sex difference in mean 1.7 points plus difference in standard deviation of 1 point.
Only 1% of the public has an IQ above 135, but, based on results like that of Richard Lynn and others quoted above, there would be more than twice as many men as women at such high IQ levels.
Figure 4: The ratio of men to women at high IQ given illustrative small differences in population means and/or standard deviations
The small sex difference in IQ which Lynn and others assert might, via its amplified effect on the tail of the distribution, provide an explanation for the remarkable observation by Olson (2014). In the USA there is a clear correlation between the average IQ of students studying a given subject and the gender ratio of that subject. Subjects dominated by men have high IQ averaged across all people studying the subject, whilst subjects dominated by women have relatively lower IQ averaged across all those studying the subject, see Figure 5. (Note that “relatively lower” means closer to the national population average, but still above average (>100) because we are dealing with college students in all cases).
Figure 5: Average IQ of USA Students by College Major Gender Ratio
So, finally, to answer Professor Smithers, all that is why I find it so difficult to accept that females are cleverer. Not only does the best available (and extensively researched) measure of “cleverness” not support it, but there is ample evidence that the nature of the awards upon which the claim is based are the source of the skew in performance. Moreover, there is clear evidence of sex bias by those involved in the process. And – what I have not even begun to address – a host of psychological and sociological issues which will impact on school attainment irrespective of innate ability.
But worse than all this is that the education machine is stuffed full of people like Professor Smithers who are entirely comfortable with making public statements along the lines that “girls are cleverer” whilst also waxing lyrical about the dangers of “stereotype threat” to educational attainment, and yet failing to realise that they themselves constitute such a threat.
I look forward to the Professor’s next article about the educational underachievement of black Caribbeans and Pakistanis, and those from poor backgrounds, exhorting us to embrace the obvious truth that “they are just less clever”. But there will be no such article because, in those cases, the Professor will not be doing as he advises in the context of sex – taking the data at face value rather than explaining it away – instead he will suddenly remember his background in sociology and have an urgent need to explain away.
There is nothing progressive about views such as those expressed by Smithers; actually they are embarrassingly atavistic. It is bad enough that power hungry and male-hating female feminists have become so common, but let’s not forget that this feminist power has been granted by men. A cheap trick that advances a man, unjustified, on the male hierarchy is to make himself look good by trashing other males. Badmouthing males is always safe and can only assist the ambitious, of either sex, regardless of ideology. We deplorables are no longer fooled. Feminism is the servant of the elites, and we know it now.
No comments:
Post a Comment