A study of more than 350,000 students in England has found that teachers predict higher A-Level grades for girls than boys who go on to achieve the same exam results. The findings raise fresh questions about the role of predicted grades in university admissions.
The research, by academics at the Universities of Cambridge and Bristol, found a consistent “pro-girl” gap in teacher predictions across 10 A-Level subjects. Depending on the subject, girls were between 3% and 11% likelier than boys to have their grades overpredicted, and between 15% and 41% less likely to have them underpredicted.
This does not mean that teachers deliberately favour girls. Instead, the researchers suggest that because girls typically have stronger all-round academic records, teachers may pick up on characteristics like motivation, diligence and classroom engagement when forming expectations about their future performance.
This could create a “halo effect”, where positive impressions about these students spill over into predictions about their academic potential.
Indeed, when the researchers compared students who not only achieved the same A-Level results, but also had similar records of prior academic performance, the picture changed. Within these groupings, boys received higher predicted grades in scientific STEM subjects, while girls retained a smaller advantage in non-STEM subjects, such as English Literature.
The findings, published in the Economics of Education Review, have implications for the UK’s university admissions system. Although university places are ultimately awarded based on students’ final A-Level results, initial offers are made months earlier and rely heavily on predicted grades.
Organisations including the educational charity The Sutton Trust have called for reforms to the system. In 2020, the Government consulted on moving towards Post-Qualification Admissions, where university decisions would be made after students received their final grades – but ultimately decided against the change.
Dr Konstantina Maragkou, from the University of Cambridge’s Faculty of Education and the study’s lead author, said the debate around predicted grades should be revisited.
“It is very unlikely that teachers deliberately predict better results for girls, but they may be relying on an impression of student characteristics which can lead to gendered predictions,” she said. “We need to think carefully about how fair that is for young people when the stakes are so high.”
The study used data from three university application cycles between 2017 and 2019. By comparing boys and girls with the same A-Level outcomes, the researchers identified a substantial gender gap in predicted grades.
On average, girls received predictions between one tenth and one fifth of a grade higher than boys. For example, girls were predicted, on average, around 0.17 of a grade higher in Physics and 0.13 higher in Sociology. “Differences of this size may appear small, but they can be enough to move a student across a grade boundary – for example, from a B to an A, or an A to an A*,” Maragkou said.
The researchers tested whether the findings could be explained by other differences between boys and girls and accounted for a wide range of other possible factors, including socioeconomic background, ethnicity, class size and school attended. The results remained unchanged in each case.
The researchers then explored what might explain this gender gap. One possibility is that teachers’ predictions are influenced by a conscious or unconscious perception of girls’ abilities compared with boys. Another is that their predictions incorporate broader information about the students. Previous studies have suggested that teachers could be influenced by observable characteristics such as a student’s work ethic, diligence, engagement, or everyday conduct at school.
To investigate this, the researchers used earlier Key Stage 2 and GCSE results to measure students’ broader academic profiles. Crucially, they looked beyond performance in the subjects they later chose at A-Level, capturing what they describe as “cross-disciplinary competence”. For example, they examined the earlier English performance of students who went on to take STEM A-Levels, and the earlier maths performance of students who chose non-STEM subjects.
This substantially changed the results. When comparing boys and girls with both identical A-Level outcomes and comparable prior academic records, the prediction advantage for girls fell by about 70% in non-STEM subjects and reversed in STEM subjects in favour of boys. “Accounting for prior academic record did not eliminate the gender gap; it just changed it,” Maragkou said.
The researchers also found that the gender gap was particularly pronounced among lower-achieving students, possibly because predictions are more difficult when performance is less certain.
The authors stress that their explanation for the gender differences requires further research. However, they note that official guidance states that “predicted grades should be independent of student behaviour or background.”
The findings may also have implications for other countries where teacher assessments form part of student evaluations, although the UK remains unusual in using predictions in university admission decisions.
Maragkou said that any attempt at reform should involve teachers, policymakers and researchers. “Teachers are not supposed to base predicted grades on students’ character, but they are probably not doing so consciously,” she said. “We also need to ask whether it is always wrong, conceptually, to consider broader qualities that affect academic success. These are difficult questions and they deserve careful consideration.”
