Wednesday 27 May 2009

Do elite US colleges choose personality over IQ?

Sex ratios in the most-selective elite US undergraduate colleges and universities are consistent with the hypothesis that modern educational systems increasingly select for conscientious personality compared with intelligence

Medical Hypotheses. 2009; 73: 127-129

Bruce G. Charlton, , Editor-in-Chief, Medical Hypotheses

Professor of Theoretical Medicine University of Buckingham, UK


Summary

The main predictors of examination results and educational achievement in modern societies are intelligence (IQ – or general factor ‘g’ intelligence) and the personality trait termed ‘Conscientiousness’ (C). I have previously argued that increased use of continuous assessment (e.g. course work rather than timed and supervised examinations) and increased duration of the educational process implies that modern educational systems have become increasingly selective for the personality trait of Conscientiousness and consequently less selective for IQ. I have tested this prediction (in a preliminary fashion) by looking at the sex ratios in the most selective elite US universities. My two main assumptions are: (1) that a greater proportion of individuals with very high intelligence are men than women, and (2) that women are more conscientious than men. To estimate the proportion of men and women expected at highly-selective schools, I performed demonstration calculations based on three plausible estimates of male and female IQ averages and standard deviations. The expected percentage of men at elite undergraduate colleges (selecting students with IQ above 130 – i.e. in the top 2% of the population) were 66%, 61% and 74%. When these estimates were compared with the sex ratios at 33 elite colleges and universities, only two technical institutes had more than 60% men. Elite US colleges and universities therefore seem to be selecting primarily on the basis of something other than IQ – probably conscientiousness. There is a ‘missing population’ of very high IQ men who are not being admitted to the most selective and prestigious undergraduate schools, probably because their high school educational qualifications and evaluations are too low. This analysis is therefore consistent with the hypothesis that modern educational systems tend to select more strongly for Conscientiousness than for IQ. The implication is that modern undergraduates at the most-selective US schools are not primarily an intelligence elite, as commonly assumed, but instead an elite for Conscientious personality.

***

IQ and C predict educational attainment

Evidence from a range of studies suggests that the main determinants of examination results and educational achievement in modern societies are intelligence (IQ – or general factor ‘g’ intelligence) and the personality trait variously described as ‘Conscientiousness’, self-discipline, perseverance or something similar (see Ref. [1] for review). IQ is (roughly speaking) that cognitive ability which enables people to think abstractly and learn quickly; Conscientiousness (broadly synonymous with perseverance or self-discipline) is the personality trait that enables people to work hard for long periods at dull tasks, to think before acting and to take a long term view.

I have previously argued that a combination of the increased use of continuous assessment (e.g. course work rather than timed and supervised examinations) and the increased duration of the educational process implies that modern educational systems have become increasingly selective for Conscientiousness (C) [1]. My argument is that, because C is not closely correlated with intelligence, then demand for increasing levels of C will inevitably lead to reduced selectivity for intelligence. Ever-higher levels of C will usually only be attainable by progressively relaxing standards for IQ.

If this reasoning is correct, it would be predicted that the most highly-educated and most educationally-selected people would be characterized more by their extremely-high Conscientiousness than by their extremely-high intelligence. More precisely, there would be a trend for educational selectivity to increase average ranking for C more than the average ranking for IQ.

Sex ratios at the most-selective US colleges and universities

I have tested this prediction (in a preliminary fashion) by looking at the sex ratios in the most selective elite US universities. My two main assumptions are: (1) that a greater proportion of individuals with very high intelligence are men than women, and (2) that women are more Conscientious than men.

Most IQ studies find a greater proportion of men than women among very high IQ adults. For example, the US national 2008 SAT results show a higher proportion of men than women scoring in the highest band for the most g-loaded sections (Critical Reading and Mathematics) [2]. This male domination of the highest scorers in IQ testing is consistent with IQ surveys going back over many decades [3] and studies of creative and intellectual genius [4] and [5]. I will therefore assume a higher proportion of men than women at levels of very high IQ. In contrast, current evidence suggests that Conscientiousness, especially the academically-relevant sub-trait of self-discipline, is higher in women than men [1] and [6].

On this basis alone, without any calculations, and if we assume equal proportions of men and women in the US population, no important differences in sex applications to college and a sex-blind policy of selectivity; it would be expected that there was a greater proportion of men than women at highly-selective elite colleges, and that the more selective the colleges the greater would be the expected proportion of men. By contrast, if there was an equal or greater-proportion of women at elite colleges then this would be consistent with C being more rigorously selected-for than IQ.

Predicted proportion of men at elite schools – on the basis of IQ
To make this exercise more precise, it is helpful to estimate the proportion of men and women which would be expected at highly-selective schools.

I have focused on predictions related to IQ because much more is known about IQ than C, and C cannot yet be quantified as precisely as IQ. An IQ of 130 is used as a plausible threshold for selectivity at elite universities: this is approximately two standard deviations above the average IQ and includes the top 2% of the population.

However, the magnitude of the expected sex differential is relevant, since if the expected sex differential was small it could easily be swamped by statistical noise, or by other relevant variables. The magnitude of the predicted sex differential depends on the assumptions of male and female IQ average differences and distributions.

There are three mainstream explanations of why there are more men than women among the population of very high IQ people.

1. Men have a higher average IQ than women, but the same variance. For instance, Lynn suggests that men have an average IQ about 4–5 points higher than women with the sexes having the same standard deviation (conventionally 15 IQ points) [7].

2. Men and women have a near-identical average IQ, but men have a greater variance in IQ than women (higher standard deviation). For instance Hedges and Nowell present Project Talent data that suggest men and women have the same IQ, but men have a standard deviation around 10% greater than women [8].

3. Men have both a higher average IQ and larger standard deviation of IQ than women. For instance, Hans Eysenck accepted Lynn’s estimate of about 4 IQ points difference in average IQ and also assumed that women had a standard deviation of 14 compared with the male standard deviation of 15 [4].

We can use these ball-park estimates as the basis for calculating approximate expected sex ratios at elite US undergraduate schools.

Therefore on the basis of IQ considered alone (Table 1), it would be expected to find a considerably greater proportion of men than women at elite undergraduate colleges. The prediction is that the most selective institutions would admit at least 60% men (and probably a higher proportion).

Table 1.

Demonstration calculations of the effect of plausible male versus female IQ averages and standard deviations on the proportion of men and women at elite colleges with threshold IQ of 130 for a US population with average mean IQ 100 (SD 15). Key: SD = standard deviation; M = men; W = women; av. = average.

Assumption Mean IQ (SD) Percentage IQ > 130 Predicted % men at elite college

M higher av. IQ than W: M 102 (15) 3.1% c. 66% W 98 (15) 1.6%

M > SD than W: M 100 (15.75) 2.8% c. 61% W 100 (14.25) 1.8%

Men higher av. IQ: M 102 (15) 3.1% c. 74% & >SD than W W 98 (14) 1.1%





So, any sex ratio less than this would imply that other qualities than IQ are actually determining selection; or alternatively that one or more of the assumptions are incorrect – for example there might be sexually differential patterns of application or selection.

Using the About.com: College Admissions web pages (http://collegeapps.about.com/ – up to March 2009) I generated a list of sex ratios (the percentage of men) at three categories of elite US colleges and universities: (1) Ivy League plus several comparably-selective private research universities; (2) The top 10 public universities; (3) The top 10 liberal arts colleges.

From Table 2 it is clear that almost all these 33 elite US undergraduate schools select approximately equal proportions of men and women with only two technical universities (Caltech and Georgia Tech) having a male sex ratio greater than 60%. If the assumptions hold, then the implication is that elite colleges seem to be selecting mainly on the basis of something other than IQ – probably Conscientiousness.


--------------------------------------------------------------------------------


Table 2.

Sex ratios at undergraduate level – percentage of men. Colleges with more than 60% men are marked with an asterisk.

Ivy League and similar private research universities

Brown 48%
Columbia 51%
Cornell 51%
Dartmouth 50%
Harvard 47%
Pennsylvania 49%
Princeton 53%
Yale 51%
Stanford 52%
Duke 51%
Chicago 50%
MIT 56%
*Caltech 71%


Top 10 public universities

Berkeley 46%
*Georgia Tech 71%
UCLA 45%
UCSD 48%
U Michigan 50%
UNC Chapel Hill 41%
Urbana Champaign 53%
U Virginia 45%
William and Mary 46%


Top 10 liberal arts colleges: (NB: Wellesley is essentially a women’s college)

Amherst 50%
Carleton 47%
Claremont–McKenna 54%
Grinnell 46%
Haverford 46%
Middlebury 48%
Pomona 51%
Reed 45%
Swarthmore 48%
Wellesley 2%
Williams 49%





It seems that there is a ‘missing population’ of very high IQ men who are not getting admitted to the most selective and prestigious undergraduate schools. The likely reason is that their high school educational qualifications and evaluations are too low, since these men probably lack the very high levels of C required to negotiate modern educational systems and achieve the very highest level of success (in the top 2% of attainment). These men with very high IQ but only moderate C are presumably attending a wide spectrum of less-selective and lower-ranked undergraduate schools, or (less plausibly) dropping-out of the educational system altogether.

A further factor may be that colleges are also selecting on the basis of high sociability, which can be measured as the personality trait of Agreeableness [1]. Agreeableness is higher in women. High Agreeableness would not be expected to lead to better educational performance, but instead would be likely to enhance an applicant’s resume with a record of participation in societies, charities and sports together with general friendliness and club-ability – these factors may well be counted in favour of a student and would also tend differentially to favour the admission of women.

My hypothesis [1] that Conscientiousness (and perhaps Agreeableness) count for more than IQ at the level of elite college admissions receives some support from this data set, and could be tested further by longitudinal studies which measured IQ and personality during childhood (rankings of IQ and personality tend to be stable throughout life), and followed-up students through the school and college examination and selection process to observe the interaction between these variables.

The implication is that modern undergraduates at the most selective US universities are not so much an elite for intelligence, as is commonly assumed, but more of an elite in terms of traits such as perseverance and self-discipline.

Acknowledgement

My thanks to Richard Lynn for his help and advice in preparing this editorial.

References

[1] B.G. Charlton, Why are modern scientists so dull? How science selects for perseverance and sociability at the expense of intelligence and creativity, Med Hypotheses 72 (2009), pp. 237–243.


[2] College Board SAT, 2008 College Bound Seniors. ; [accessed 19.03.09].


[3] B.G. Charlton, Pioneering studies of IQ by G.H. Thomson and J.F. Duff – an example of established knowledge subsequently ‘hidden in plain sight’, Med Hypotheses 71 (2008), pp. 625–628.


[4] H.J. Eysenck, Genius: the natural history of creativity, Cambridge University Press, Cambridge, UK (1995).


[5] C. Murray, Human accomplishment. The pursuit of excellence in the arts and sciences 800 BC to 1950, HarperCollins, New York (2003).


[6] A.L. Duckworth and M.E.P. Seligman, Self-discipline gives girls the edge, J Educ Psychol 98 (2006), pp. 198–208.


[7] R. Lynn and P. Irwing, Sex differences on the progressive matrices: a meta analysis, Intelligence 32 (2004), pp. 481–498.


[8] L.V. Hedges and A. Nowell, Sex differences in mental test scores, variability, and numbers of high-scoring individuals, Science 269 (1995), pp. 41–45.