Rapid Review of Data Education within Mathematics 5-16

Appendix 1 Primary anchoring experiences (extracts from ACME PEY document) 57 Appendix 2. ProDaBi -Project Data Science and Big Data in German schools 60This July 2026 version has corrected a previous error where Data Science Learning Progressions (DSLP) were attributed as Data Education for Everyone (DE4E).pg. 4 pg.8 LimitationsThe 'evidence' we draw on in this report is not primary research evidence.We are aware of no trials that demonstrate the superiority of one approach to curriculum inclusion and organisation over another, and certainly not in the new field of data education.Given the rapidity of the review, we have not sought results of finer-grained experiments.Instead, we have drawn on sources that have been implemented at scale, and for some years, e.g. the curricula of New Zealand, Scotland and Singapore.While none of these approach the scale of England's education system, they do operate across a range of schools.There are problems in comparing performance internationally (not least that this ignores known effects of social disadvantage) but we note that these sources cover a range.In PISA 2022, students in Singapore performed better at uncertainty and data questions than English students do, and English students outperformed New Zealand and Scottish students (Ingram et al., 2023).This is welcome but, in England, out of the four PISA mathematics domains, uncertainty and data showed the largest gap between the highest-and lowest-performing students, indicating that many students are being left behind in data education.As well as scale, the other quality of the sources we use is their roots in iterative design research involving teachers, researchers and statistics curriculum developers.Notable examples are GAISE II (Bargagliotti et al., 2020), the New Zealand (NZ) curriculum (Tahurangi, 2026), Scotland's Data Skills and Data Literacy curriculum (Data Education in Schools, 2024) and work on learning pg. 9 progressions led by Data Science for Everyone www.datascience4everyone.org/(Data Science Learning Progressions, 2025).A related limitation, that we have not addressed, is that we have not considered the connections between mathematics and AI that are less relevant to statistics.Notably, AI and Machine Learning (ML) make use of mathematical concepts such as vectors, circles and decision trees, and there is design research (e.g.ProDaBi, please see Appendix 2) trialling lessons for school-age children that make these connections explicit.