Below is the short version of my full research essay – Inconvenient truths on gender inequality in STEM – How differences in choices explain disparities better than the discrimination narrative

In the first part of the essay, Inconvenient truths on gender inequality in STEM – the failure of the discrimination narrative, I argued that discrimination cannot be held responsible for disparities in STEM. We come now to the more difficult and ambitious second part of the essay. After demonstrating that the discrimination and bias narrative does not explain STEM (GEEMP; geo-sciences, economics, engineering, math/computer science, physics) inequality, other more compelling and persuasive factors are needed to account for the disparity. I turn now to what I call the choice narrative – the idea that gender disparities in GEEMP are better explained as being the result of choices influenced by a complex set of gender differences in abilities, interests and lifestyle preferences.

The first fact to note is that there are no differences in general intelligence between men and women. We also know that on average females do better than males in overall school achievements. The Programme for International Student Assessment (PISA) is an international survey taken by 15 year old students in 72 countries every 3 years to assess science, mathematics, reading, collaborative problem solving and financial literacy (OECD, 2018). PISA results show a female advantage in overall scores. However, males scores are more variable then females which means they are over-represented at the top percentile and bottom percentile as well. It is important to note that a lack of differences in general abilities, does not mean there are no differences in specific cognitive abilities. There are three broad cognitive abilities where gender differences have been found: verbal, quantitative and spatial

Females on average do better than males at most verbal abilities. Verbal skills includes a host of different tasks such as reading, writing, comprehension, language usage. The average female advantage in particular verbal abilities like writing and reading is robust, large and found to be consistent internationally. However, the type of test used to measure verbal abilities can yield differences. Studies finds that there are no average sex differences in verbal reasoning when it is measured by the Cognitive Abilities Test (CogAT). In general research looking at different standardized tests for verbal abilities found a female advantage and one that increases the higher the achievement percentile one looks at.

Quantitative and mathematical abilities will have direct relevance on STEM outcomes. Although quantitative abilities consists of a wide range of tasks and can be measured in different ways and at different stages; the following general conclusions can be drawn from several meta-analytic studies. Hyde in her research found that there are negligible differences in average math abilities between men and women from childhood until adulthood. However, there are small differences that emerge depending on the types of test used to measure abilities. The average differences are too small and on most measures are non-existent and therefore cannot account for gender disparities in STEM. But a further reason why average differences will not explain STEM disparities is because STEM individuals will more likely be individuals with high quantitative abilities rather than average and it is at the tails of cognitive abilities where important sex differences emerge.

A large body of evidence has demonstrated that males have greater variability in mathematical and quantitative abilities than females. A number of studies that measured math ability by grade scores or standardized tests such as the SAT, Cog AT and PISA to mention a few have found that there is a consistent and significant male advantage amongst high performing individuals but also that there are more males than females amongst the lowest performing group. The greater variability of males has been stable for the last 20 years on a number of tests used to measure math abilities.

What this means then is that if GEEMP fields draw from high performing math individuals then it will find more males than females in this pool. However, it offers only a partial explanation because the 2:1 ratio amongst the top math scorers (top 1%) is less than the actual male-female ratio (roughly 3:1) seen in bachelor graduations in GEEMP. Also researchers have found that STEM draws from the top 25% rather than top 1% and at that threshold the ratio of male to females is much lower than 2:1. But even amongst the top 1% of math scorers there are differences in gender educational outcomes: women more likely to study organic sciences while men are more likely to choose inorganic sciences. These differences cannot be explained by differences in quantitative ability.

The third cognitive ability where there are sex differences is visuospatial ability. Visuospatial ability refers to abilities such as: visualizing 2D objects after they have been folded into 3D objects; visualizing 2D objects when they are rotated or flipped in a plane; mechanical reasoning where relationships between mechanical systems like gears, pulley and springs are deduced; abstract reasoning which involves the finding logical relationships between figures.   A male advantage particularly in mechanical reasoning has been found and this specific cognitive skill has clear implications for engineering and physical sciences. Wai and researchers demonstrated in their longitudinal study the importance of spatial abilities in determining STEM outcomes particularly in physical sciences and engineering. Their study tracked a random sample of high school students over a period of 11 years after their graduation. They found that graduates in engineering, physics, maths, and computer science at the bachelors, masters and doctorate level had higher spatial ability than other graduates in LPS fields and non-STEM fields.

Figure 1 – Spatial, verbal and mathematical ability of students in various fields as well as at different university achievement levels. The vertical axis shows the ability levels for the three different cognitive abilities, the horizontal axis shows the average ability of students in various fields.

It is not just individual ability in these cognitive abilities that matters but also how they combine and interact within an individual. Relative strength is far more important than just cognitive ability in a specific domain.  If I score 70% on verbal tests and 75% on math tests, and you score 85% on verbal and 80% on math; then my intra-individual academic strength is math and yours is verbal. This implies that I am more likely to choose fields and occupations math-related and you are likely to choose verbal related fields even though your math score is higher than mine. This is because as individuals we tend to look at what area we are strongest in, and decide on that basis what educational and occupational options to pursue. What the literature reveals is a consensus that there are important differences between males and females in personal academic strengths.

Stoet & Geary analysed PISA data for 475,000 students from 67 countries and found that more males than females had science and mathematics as academic strengths; and more females than males had reading as a personal academic strength. Other researchers have found that students with high math and high verbal scores (HM/HV) choose STEM careers at lower rates than students with high math and moderate verbal scores (HM/MV). The reason being that high math plus high verbal scores means one has more options beyond math fields where one can succeed in. They also found that HM/HV students consists of more females (63%) and that HM/MV consists of more males (70%). Researchers in Sweden who performed a longitudinal analysis on the educational and career outcomes of 167,776 individuals in Sweden and found that more boys than girls (65.5% vs 34.5%) displayed what they called a technical/numerical academic strength; while more girls than boys (67.8% vs 32.2%) displayed a verbal/language academic strength. Therefore gender differences in intra-individual academic strength explain some of the differences in GEEMP outcomes and cannot be ignored.

Abilities play an important but partial part in explaining disparities – vocational interests are just as important and in this area gender differences emerge. Researchers Su, Rounds and Armstrong conducted the first comprehensive meta-analysis of differences in interests between men and women. The study looked at vocational interests which they defined as “the expression of personality in work, hobbies, recreational activities, and preferences”. The concluded the following:

“The present study, however, revealed substantial sex differences in vocational interests. The largest difference between men and women was found along the Things–People dimension, with men gravitated toward things-oriented careers and women gravitated toward people-oriented careers.”

They also evaluated sex differences particularly in STEM vocational interests and found that men have a moderately higher interest in Science and Mathematics compared to women. The largest difference was found in engineering interest with men having a much higher interest than women. If you took the top 25% of people with the highest interest in engineering the ratio of females to male would be 1: 5. Another study investigated gender differences in interests within STEM fields themselves. In their study they found that specific STEM fields are different in terms of how people-orientated or thing-orientated they are. Health and life sciences are more people-orientated; while fields such as engineering and physics are more thing-orientated, the other STEM fields will fall in between.

They found that firstly, the engineering field has the largest gender difference in basic interest – more men than women had an interest in engineering fields. The largest difference where more women had a higher interest than men in the field was Social and Medical sciences. Secondly, they found that the greater the male representation in a discipline the more things-orientated is the discipline; the reverse also held true the greater the female representation in a STEM discipline the higher along the people-orientated scale is the discipline. Studies have also found that sex differences in vocational interests that are present in the general population are also present in individuals with high math abilities (top 3%, 1%, 0.5%, 0.01%) with the ability to succeed in any STEM field. Even amongst this group of highly gifted individuals, males are more likely than females to rank things-orientated interests first. According to the researchers, “This preference difference between men and women, which is also conspicuous in intellectually precocious samples, undoubtedly contributes to the preponderance of females with profound mathematical gifts who choose to become physicians rather than engineers and physical scientist”. Therefore gender differences in interests are an important factor in determining STEM outcomes.

The last factor which influences STEM outcomes is lifestyle preferences and values which refers to “how people perceive and prefer to structure their lives in the broader context of family, personal development, career, social relationships, and community”A longitudinal study that followed the educational and occupational outcomes of mathematically highly gifted individuals from a young age found that they differed in work-life preferences. Men had allocated an average of 11 hours more towards their careers than women had; and also indicated that they planned to allocate more time towards careers than women. Women had allocated planned to allocate more towards family and non-career pursuits. When the survey asked how much time individuals were willing to allocate towards their ideal jobs, men were willing to allocate more time than women. The ratio of men to women increased when the hours per week individuals were willing to allocate towards their jobs increased; more men than women are willing to work 80 hours per week. Men in the group were found to prefer working full-time, being successful at work, earning a high income more than women; while more women than men preferred working part-time, having strong friendships, not working outside the home, and having time for friends and family. These differences might not fully explain why there are disparities in STEM fields but they play a part in explaining why men are more likely to be amongst the highest achievers in the field. The study also notes that both men and women in the group were equally satisfied with their life outcomes.

In conclusion then, it is clear that the discrimination narrative fails to account for disparities in STEM. Differences in choices which are influenced by ability levels, ability patterns, interests and life-work preferences go a long way in explaining why we do not see equal representations in the different STEM fields. One cannot simply ignore the wealth of data showing how choices men and women make are simply different. The underlying assumption behind most gender inequality discussions, namely that; equal opportunities must result in equal outcomes has been shown, I hope, to be grossly inadequate at best, and blatantly false at worst. We cannot reduce the issue to simply being about “male dominance, patriarchy and companies providing baby care in the office”, as Pillay put it in his now infamous article. The real picture is far more complex and nuanced then we would love it to be. In the words of psychologist Susan Pinker:

“I absolutely agree with and promote equal access to opportunities and education. But equal access to opportunities and education does not determine an equal result. Assuming that women are simply a tamped down, smothered version of men—and would always choose what men choose if they only had the chance—is neither respectful of women’s autonomy nor supported by the data.”