The Summers Debate: Is Choice the Driving Force Behind Sex Disparities in Science and Engineering?

October 2005

In January 2005, Lawrence H. Summers, the president of Harvard, triggered a storm of controversy when he provocatively raised the issue of why there are so few women in tenured positions in science and engineering. Contrary to the portrayal in many news reports, Summers did not simplistically claim it is a matter of ‘intrinsic aptitude’ differing between the sexes. In fact, he attributed sex disparities in academic departments primarily to what he called the ‘high-powered job hypothesis’: differential willingness of men and women to undertake the high level of commitment and single-minded devotion that high-powered jobs require. He alleged that the phenomenon is apparent in top professional and managerial positions across the board, and is not unique to science and engineering.

Is Summers right? Are women opting out? Or, as some have suggested, are there other reasons for the sex disparities in science and engineering departments, reasons having to do with inequalities in hiring or inequalities in competence?

Summers argued that women are under-represented in faculties of engineering, relative to the availability of qualified women in graduate school a generation ago. Women comprised 10-12% of the graduate pool 20-25 years ago, according to data from the American Association of Engineering Societies and National Science Foundation. So they should make up 10-12% of full professors today if they entered academia and were promoted in the same proportions as men. In fact, they make up only 4% of full professors in Canada (CAUT Almanac of Post-Secondary Education, 2001-02 figures) and 2.7% in the U.S. At present, women do make up 12.6% of new assistant professors. But this is low when you consider that women received 21% of the PhDs in engineering in 2004.

Remember – these are women who had the cognitive credentials required. They applied to and were accepted into undergraduate programs, successfully earned four-year degrees, had the interest and qualifications to pursue graduate work, and competed successfully for entry into graduate programs with competitive admission. Why shouldn’t they be found in representative proportions in the academic workforce? A survey of women engineers by the National Research Council showed that the top-producing doctoral institutions of female engineering faculty were Massachusetts Institute of Technology, Stanford University, University of California – Berkeley, University of Illinois, and Carnegie Mellon – an illustrious group. It seems unlikely, therefore, that women engineers are systematically under-qualified or less desirable than men. Why, then, are they so under-represented in the top ranks of academia?

Attributing the differences to cognitive sex differences is too simplistic. We need only look at other academic fields to see that the situation in engineering is not unique. CAUT data for 2001-02, for example, show that only 4% of full professors of engineering were female, but the figures for many other disciplines are similar: archeology (7%), economics (4%), geography (4%), philosophy (12%), classical studies (8%), psychology (13%). There are exceptions with relatively high percentages of women among full professors, such as English literature (31%). However, if we were to make a prediction based solely on cognitive sex differences, we’d have to predict, if anything, that females would outnumber males among the full professors in some of these fields. Empirically, this is not the case.

CAUT figures show a diminishing proportion of women at each successive stage of the academic pipeline. In chemistry, the proportion of women decreases from bachelor’s (51%), to master’s (41%), PhD (32%), assistant (21%), associate (13%), and full professor (6%) (CAUT, 2001-02 figures). If this were true only in science and engineering, it would be one thing. But the same pattern is seen across the board: e.g. in anthropology: bachelor’s (73%), PhD (62%), assistant (52%), associate (45%), full professor (29%). Cross-sectional data are hard to interpret because historical demographics could complicate the picture. However, in progressing from PhD to assistant professor, there is usually a spread of only a year or two. Similarly, assistants constitute the pool for promotion to associate. Whence the drop-off? A university-wide survey of graduate students at the University of Western Ontario (van Anders, 2004) seems to confirm that some women are actively opting out of the academic stream, as late as the PhD level. In a survey of career goals, a sex difference was identified in the percentage of men and women who aspired to enter academia upon graduation: more men than women responded “definitely yes” or “probably yes” when asked about their intentions to enter academia, and more women than men responded “definitely no” or “probably no”. There were no sex differences in the perceived importance of interest in research, interest in teaching, extended family issues, financial issues, or the appreciation of the academic lifestyle. But issues related to family mobility and plans for parenthood stood out as concerns among women.

In the National Research Council data, 52% of women faculty employed in engineering departments across the U.S. said balancing work and family responsibilities had had a negative impact on their careers (vs 16% who said the impact was positive); 35% said having children had had a negative impact (vs 17% who said the impact was positive). Marriage was viewed more favorably – 38% said the impact was positive, only 15% said it was negative.

The implication is that women are opting not to go into academia in the same proportions as men, or if they do, they are less fully engaged. And they opt out primarily for reasons related to family issues, not for reasons of competency. Summers might be right – this is a problem and not just for universities. A recent newspaper article based on trends at the University of Toronto business school claimed the enrollment of women in the executive MBA program had flat-lined at 25% (National Post, Aug. 20); and that while women were satisfactorily represented in middle management, they are thinly seen in the upper echelons of the corporate world – only 4.5% of FP500 companies’ top earners are women. For women more so than men, it may come down to an issue of balance: between home, family, and the demands of the high-powered workplace.

As for cognitive specializations that differ between the sexes, this is a red herring in the Summers debate. Summers was not musing about why so few women go into engineering in the first place, but why so few highly qualified women – who have already demonstrated cognitive competence by competing successfully for admission to graduate programs – end up in academia and particularly at the rank of full professor. Yes, research has shown cognitive sex differences do exist in certain domains of intellectual functioning. There is no sex difference in intelligence, but there are other cognitive domains – more circumscribed – where sex differences do occur. For instance, women have been found to excel in many components of receptive and expressive language, including things like the ability to quickly and accurately find the appropriate words to express oneself, the ability to comprehend complex verbal text, to write fluently and grammatically, and to capture one’s thoughts in words. And working memory – the basis for a number of mental operations – shows a female advantage. These differences could be part of the reason for better performance by girls than boys on standardized tests of academic achievement (e.g. Program for International Student Assessment) and, possibly, for greater representation of women than men in a number of undergraduate programs where women now comprise ~60% of the undergraduates.

The situation is different in mathematics and engineering. These are fields where verbal competence is not the primary criterion for advancement. To excel in math or science careers, individuals need to have high levels of intellectual ability, but especially quantitative reasoning. There are sex differences in elementary and high-school mathematics performance – some areas of math show better performance by females, others by males. As always, these differences are statistical averages; there are many females who outscore most males, and vice versa. But in his remarks, Summers was careful to emphasize that he was not talking about elementary and high-school math, nor about individuals in the normal range of aptitude. Rather, he was describing a type of high-level conceptualization and creativity of thought, which could be a basis for outstanding performance at the very highest levels in science and engineering. We do not know which sex, if either, has an advantage when it comes to this type of complex innovative thinking. Current psychological tests simply do not assess this.

The Study of Mathematically Precocious Youth (SMPY) has been quoted in the Summers debate. The SMPY is a longitudinal study of over 5,000 intellectually talented kids recruited in the U.S. between 1972 and 1997 through talent- search methods. ‘Mathematically talented’ was defined as scoring in the top 1% in quantitative reasoning at age 13. In this highly elite group of kids, males did outnumber females. This is the usual context for citing the SMPY data. In fact, however, the SMPY study helps to underscore Summers’ essential point about the importance of self-selection. The SMPY data showed that of the gifted girls who scored in the top 1% – who exhibited exceptional mathematical talent above and beyond the typical physical scientist – only 34% went on to earn undergraduate degrees in math or science. And less than 1% of the girls in the top 1% of mathematical ability pursued PhDs in mathematics, engineering, and the physical sciences combined (Lubinski & Benbow, 1992). Eight times as many of the gifted males did so.

Considering all the evidence, I find myself in agreement with Larry Summers. It seems that self-selection goes far toward explaining the sex differential we see in the highest ranks of academia. We need to understand this phenomenon and ask what can be done to involve more women in academic life - so we can capitalize more fully upon the nation’s talent pool.

Useful Reading Websites

  • Statistics Canada (2004). Measuring up: Canadian results of the OECD PISA study – The performance of Canada’s youth in mathematics, reading, science and problem solving – 2003 first findings for Canadians aged 15.
  • Summers, L. H. (2005). Remarks at NBER conference on diversifying the science and engineering workforce.


  • Lubinski, D. & Benbow, C. P. (1992). Gender differences in abilities and preferences among the gifted: Implications for the math-science pipeline. Current Directions in Psychological Science, 1, 61-66.
  • National Research Council Committee on Women in Science and Engineering (2001). Female engineering faculty at U.S. institutions: A data profile. Washington DC: National Academies Press.
  • van Anders, S. M. (2004). Why the academic pipeline leaks: Fewer men than women perceive barriers to becoming professors. Sex Roles, 51, 511-521.