The Math And Science Gender Gap Shrinks When Tests Are Longer


Stephen Luntz

Freelance Writer

clockSep 3 2019, 16:18 UTC

The famous gender gap for maths and science exams disappears as they get longer. LStockstudio/Shutterstock

Teenage girls do worse on average than boys on most maths and science tests, a finding which has attracted plenty of attention with efforts to explain and address. However, one aspect of this has been missed until now: the difference diminishes as the test goes on. In longer tests, girls make up most of the difference, and often overtake boys. The finding suggests that efforts to address imbalances in hiring intakes for STEM (science, tech, engineering, and math) positions might be missing a very simple solution.


Every three years 15-year-old students in 74 countries take the Programme for International Student Assessment (PISA). In a new study, Dr Pau Balart of the University of the Balearic Islands and Erasmus University PhD student Matthijs Oosterveen compared the performance of boys and girls for the 2006-2015 tests.

As in many previous studies, Balart and Oosterveen confirmed girls did better on average than boys on reading tests, but the reverse was the case for math and science. The fact the math and science gender gaps are shrinking and vary dramatically by country suggests it is a product of environment rather than biology, although there are still people who fiercely argue the opposite.

In Nature Communications Balart and Oosterveen report the gap narrowed on the later questions in the test, and in a few countries even reversed, with girls outperforming boys towards the end. In reading tests the genders started out equal; the much-noted female superior performance only showed up later on.

To test their theory that these results reflect better female performance over longer time periods the pair turned to a database of tests that lasted longer than PISA's two hours. They found that, at least for math, the gender gap decreased with greater test length.


The authors offer several hypotheses, including differing approaches to planning for and partaking in a test by boys and girls, self-discipline and ability to maintain attention, and levels of confidence, although so far they reach no firm conclusions.

This isn't the first time women's strength on longer testing has been noted. Oxford University added an extra 15 minutes to its maths and computer science exams, and found female students' performance improved. However, this has been looked upon as suggesting a deficit in women – that they were more likely to be thrown off by time pressures – rather than greater capacity to maintain focus or motivation.

The findings raise the question of how long the ideal test should be. As Balart and Oosterveen note, most people's working day is a lot longer than a two-hour test. Longer assessment may be a better predictor of workplace performance.


Universities have adopted many responses to major gender imbalances in their science and engineering intakes, including targeted advertising and additional mentoring. More controversially, a handful of institutions have introduced different entry requirements for courses where men dominate. This study suggests, however, a simpler option would be to simply make everyone sit longer tests.