Cluster profiles of student mathematics performance are dominated by overall ability, with most clusters being linear rescalings of average exam performance¶

Claim · All claims
Evidence · 1 study (2 entries) · associational r2 · q2
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 2 of 2 entries) · not reviewed by a person

Subclaims¶

q2 i? Most BMM clusters are strongly linearly correlated with each other and with average exam score, indicating overall ability rather than distinct skill profiles drives cluster shape. → Benjamin Mawdsley 2026 q2 i? A small number of cluster pairs show near zero or slightly negative correlation, suggesting small atypical subgroups that may benefit from increased personalisation. → Benjamin Mawdsley 2026 (2)

Evidence¶

Benjamin Mawdsley 2026¶

Benjamin Mawdsley, Tom Quilter, Richard Turner, Sarah Jackson, and Paul Edwards. (2026). Archetypes or ability? Clustering for modelling student mathematical competence. https://arxiv.org/abs/2607.26063 doi:10.48550/arxiv.2607.26063

q2 · i? · associational · r2

Analysis of Bernoulli Mixture Model clusters fitted to mock exam data from the platform. The authors use Pearson correlation between all pairs of clusters, finding most clusters have linear correlation above 0.6, showing "performance on each questions typically scales linearly with overall exam performance".

"The dominant factor is ability and performance on each questions typically scales linearly with overall exam performance."

Benjamin Mawdsley 2026 (2)¶

Benjamin Mawdsley, Tom Quilter, Richard Turner, Sarah Jackson, and Paul Edwards. (2026). Archetypes or ability? Clustering for modelling student mathematical competence. https://arxiv.org/abs/2607.26063

q2 · i? · associational · r2

In the same cluster-correlation analysis of BMM outputs, the authors report that "a small number of cluster pairs have a near zero or slightly negative correlation", identifying rare atypical profiles; Appendix B gives an example of two differing high-performer populations.

"A small number of cluster pairs have a near zero or slightly negative correlation, indicating that there may be small subsets of the population that have atypical behaviour and may benefit from increased personalisation."

Discussion¶

Learner Variables¶

• General Cognitive Ability — predictor: learners who differ on it differ in outcomes

Related Claims¶

• Secondary academic performance is predicted by cognitive ability more strongly than by curiosity — the general claim this finding bears on, with the studies behind it
• Bernoulli Mixture Model fits select around 10-12 clusters per question, with no runs finding high cluster counts optimal, indicating specialised skill sets do not generally exist — related
• Explainable models including logistic regression, kNN and a Bernoulli Mixture Model all reach around 78% accuracy in predicting question-level pass/fail, with well-calibrated probabilities — related
• Using performance on each individual question as features yields only a small improvement over a single-ability logistic regression model — related
• Discussion thread length is strongly linearly related to the number of players contributing, indicating collective rather than small-group collaboration — related
• Prior mathematical accuracy is the dominant propensity factor: over half of January accuracy variance and under one-fifth of fluency variance predicted June performance — related
• Latent profile analysis of pre-service teachers' CT skills identifies three profiles (Novice, Developing, Proficient), revealing heterogeneous, non-linear skill acquisition — reports the opposite
• Not all students would benefit from improving the same competencies, suggesting an individualized approach to education may be preferable to a one-size-fits-all approach — reports the opposite