Educational debates over the past few months have focused on young people who are not in education, employment or training (NEET), and the cycle of compulsory GCSE maths resits. Too often we start discussing these problems around age 16. The latest findings in our annual review of mathematical education (RoME 2026) show that we need to prioritise tackling these issues earlier in secondary education. In our Pipeline and Introductory reports, we highlighted how the roots of inequalities in mathematical education reach a long way back. Quite when and how these can best be tackled is unclear, but a focus on key stage 3 is critical and is often overlooked amidst the legitimate emphases on strong foundations in primary and advanced pathways post-16. Over the past two years, the Observatory for Mathematical Education has been following about 27,000 pupils in 148 secondary schools across England as they progress through years 7 and 8. Surveys have explored their attitudes and experiences of school maths. The analysis identifies three things that help to explain outcomes at GCSE: confidence, understanding and curriculum. One of the clearest findings from RoME 2026 relates to gender differences. By year 8, girls are more likely than boys to report anxiety about maths, even after accounting for differences in prior attainment. Beyond ability These findings reflect patterns seen in both PISA and TIMSS and raise important questions about what happens during the early years of secondary school. If pupils with similar levels of previous attainment experience mathematics so differently, we need to look beyond ability alone to understand emerging confidence gaps. The findings suggest the aspiration of teaching-for-mastery in England’s mathematics curriculum might not be the experience of all pupils, particularly those with low prior attainment who are most likely to miss the all-important GCSE grade 4. Mathematics is cumulative. Mastering one idea provides important foundations for future learning. In our large national sample, more than 30 per cent of year 8 pupils who had not met the expected standard in maths at key stage 2 said they did not fully understand content before the class moved on. Additionally, 23 per cent of year 8 teachers said they regularly moved to a new topic before all pupils were ready. These figures were slightly higher than those reported for year 7 last year. The cumulative effect of this experience is a problem and something that needs addressing. The review reports the maths topics that teachers are prioritising in lessons. Ratio, proportion and rates of change are critically important concepts and skills that should be mastered by 16 for all manner of educational, work and life applications. Last year, in RoME 2025 ratio featured in just 2 per cent of the most recently reported year 7 lessons. In RoME 2026 this had increased to 11 per cent of lessons, which is welcome, although somewhat surprisingly teachers reported algebra to be a focus of 40 per cent of lessons. Important questions These figures are not, by themselves, a measure of curriculum quality. They do, however, raise important questions about the prioritisation of different areas of mathematics. These findings come at an important moment. The prime minister has identified vocational and technical education as a priority, with renewed focus on ensuring the education system works for all young people. That ambition is welcome, and successful skills development requires strong mathematical foundations. At the same time, the growth in numbers studying A-level maths has been remarkable, yet participation gaps in advanced mathematics, particularly by gender and socioeconomic background, also require better understanding and action. Among year 13 pupils studying the subject at A-level, the 38 per cent who are female are notably less positive, with 24 per cent reporting that they enjoy maths less than they did in year 12, compared with 16 per cent of boys. To tackle the barriers that impact young people post-16, we need better evidence on when and how they get established. RoME 2026 brings together such evidence, but is just one part of a complex picture. As the Observatory’s cohort studies build, they will enable better understanding of learner pathways, and the institutional, pedagogical and social factors that are associated with successful outcomes for different learners. The review’s findings have important implications for educational policy and practice, at the various scale readers work at within our education system. And, as with all good research, it raises many new and interesting questions that require further investigation.