Is teaching data the key to gender equality in STEM?

This year’s press release by the Joint Council for Qualifications on A-Level entries led with an arresting claim: ‘Female Science Entries Overtake Males for First Time’.

Is the effort over many decades to recruit more young women into STEM subjects finally paying off? The answer, unfortunately, is no. The JCQ was brandishing an honest trifle to deceive us in deeper consequence, as a quick look at the small print confirms. Science is defined as biology, chemistry and physics, and doesn’t include maths, further maths, ICT, computing, or design and technology.

Where are we

Why this peculiar definition? Young women have always predominated in biology. Boys still outnumber girls in physics by four to one, but a slight uptick in female entries to chemistry nudges the combined total just over 50%. This is disingenuous. Any reasonable definition of science subjects finds a substantial majority of boys studying them, at a time when 55% of all A-Levels across all subjects are taken by girls. Looking at JCQ data, the outstanding feature of the gender distribution in STEM subject entries over the last two decades has been its stubborn resistance to change.

An exception is computing, where the good news is that the proportion of girls studying the subject is rising, but the bad news is that this is only to a dismal 13%. If we include psychology (not normally defined as a STEM subject) things improve a little because three quarters of psychology candidates are young women, and the subject has doubled its numbers over the last two decades: girls in science rise from 46.6% in 2001 to 49.0% this year.

We’re in the middle of a data revolution

This situation ought to cause more concern than it does. As the data revolution takes hold, as working with data becomes more ubiquitous and as data science moves centre stage, the latter will have a dire problem with diversity if only one in five of those with university entry level qualifications in physics, one in seven in computing science, or two in five in maths are women.

The great turn-off happens at 15, as young people choose which A-Levels to study, and due to funding pressures and league table performance competition, choice has been reduced in most state sector schools. Girls comprise close to half the GCSE entries for every major science subject except computing, but the problem lies with the proportion who choose to or are able to continue their study.

Looking at JCQ data, fewer than one in six girls carry on in maths, physics and computing: barely half the proportion of boys who do so. It is therefore hardly surprising that when we look at the subjects they study at university, young women form 15% of entrants in computing science and engineering, 38% in maths and 45% in physical sciences. In 2018 the Institute of Physics reported the astonishing statistic that one quarter of English schools had precisely one girl studying A-Level physics. Astonishing, that is, until you discover that almost half (44%) did not have any girls at studying the subject at all, according to the Institute for Fiscal Studies. The Institute for Fiscal Studies also found that independent schools are much better at persuading girls to study science.

And the gender gap in STEM gets even worse after higher education. Analysis of the gender of those registering patents by Pauline Beck & Christopher Harrison at the Intellectual Property Office shows that in Britain in 2016 a miserable 7% of patents were filed by women.

Time to change tack?

This is a scandal and ought to cause far more concern than it does. Policy makers should recognise that decades of work with role models, champions, ambassadors, and publicity campaigns has achieved almost nothing. Success will not come from yet more of the same. Perhaps it is time to try another track, one that works with, rather than against, the grain of the data revolution. Stop trying to shepherd young women towards STEM disciplines that they either choose to avoid or schools are unable or unwilling to teach. Instead bring the STEM skills into the disciplines they do wish to study. Every university discipline now uses data in abundance. Dealing with data requires competence in maths and statistics, logical thinking that underpins physics, and often the kind of holistic attention to context and problem formulation that is vital in biology and chemistry.

Not only could this be done, it has been done. As the recent Royal Society report on data science noted, over the last seven years, the Q-Step programme in the social sciences has turned hundreds of young women into experts in quantitative data analysis. Moreover, because of their social science background, they combine their skills in computing and statistics with an ability to communicate how they work to different audiences: the so-called unicorn formula that employers crave. Gender was not one of Q-Step’s objectives, rather its objective was to raise the standard of statistical expertise in UK university social science. However, its success in attracting women to take up STEM skills perhaps shows us a better way of tackling the STEM gender gap. Taking the skills into the subjects young women choose to study may be much more effective than further fruitless attempts to change young women’s subject choices.