LLM-Driven Probabilistic Reasoning Interventions: Effects on Senior Secondary Students’ Data Literacy and Statistical Thinking
- 1 College of Mathematics and Statistics, Northwest Normal University, Lanzhou, China
- 2 College of Mathematics and Statistics, Northwest Normal University, Lanzhou, China
- 3 No. 3 Primary School, Lanzhou, China
Abstract
Against the integration of generative AI into mathematics education, large language models (LLMs) reshape the instructional design of probability and statistics, a core branch of stochastic mathematics focusing on uncertain reasoning. Drawing on statistical cognition theory and computational mathematics pedagogy, this study constructs an LLM-assisted cognitive intervention framework targeting probabilistic reasoning bias correction. This research aims to examine how AI-powered instructional scaffolding mitigates intuitive cognitive biases, improves adolescents’ probabilistic reasoning proficiency, and fosters data literacy. A controlled experimental design was adopted, with 124 Grade 11 students from intact randomly allocated classes split into an LLM intervention group and a conventional lecture control group. Data were collected via pretest-posttest assessments of statistical thinking, probabilistic reasoning error diagnostic tests, and standardized self-report scales, followed by structural equation modeling (SEM) quantitative analysis. Three key findings emerged: First, LLM-supported dynamic questioning, counterexample generation, and personalized error remediation significantly elevated formal stochastic reasoning performance and reduced heuristic misconceptions. Second, the intervention group achieved statistically meaningful gains in statistical perception, probabilistic modeling, and data interpretation relative to the control group. Third, cognitive engagement fully mediates the association between LLM instruction and data literacy growth, with no statistically significant direct effect of LLM intervention on data literacy. This study innovatively embeds generative AI into senior secondary probability teaching, delivers a replicable instructional framework for cultivating student data literacy, and provides empirical and theoretical evidence for AI-enabled mathematics education.
- Makramalla, M., Coles, A., le Roux, K. and Wagner, D. (2025) Mathematics Education for Sustainable Futures: A Strengths-Based Survey of the Field to Invite Further Research Action. Educational Studies in Mathematics , 119, 535-556. https://doi.org/10.1007/s10649-025-10389-x
- Rashid, S. (2026) The Role of Large Language Models in Education Simulations: A Systematic Mapping Study. Master’s Thesis, University of Turku.
- Zhu, H., Li, T., He, P. and Zhou, J. (2025) Enhancing Automated Grading in Science Education through LLM-Driven Causal Reasoning and Multimodal Analysis. Proceedings of the Thirty - Fourth International Joint Conference on Artificial Intelligence , Montreal, 16-22 August 2025, 10352-10360. https://doi.org/10.24963/ijcai.2025/1150
- Ye, Y., Li, W., Wen, Z., et al . (2026) AgentSchool: An LLM-Powered Multi-Agent Simulation for Education. arXiv: 2605.30144.
- Henkel, O., Horne-Robinson, H., Dyshel, M., et al . (2024). Learning to Love Edge Cases in Formative Math Assessment: Using the AMMORE Dataset and Chain-of-Thought Prompting to Improve Grading Accuracy. arXiv: 2409.17904.
- Ma, Y., Hu, S., Wang, Y., et al . (2024). When LLMs Learn to Be Students: The SOEI Framework for Modeling and Evaluating Virtual Student Agents in Educational Interaction. arXiv: 2410.15701.
- Kim, J.Y. (2025) Enhancing Visual Analytics through Foundation Models: Strategic Implementation and Visualization Literacy Assessment Frameworks. Ph.D. Thesis, Seoul National University.
- Kashani, M.M. (2025) Statistical Assessment of Plans via Probabilistic Optimization of Reliability. Ph.D. Thesis, IT-Universitetet i København.
- Yang, Z. (2026) Personalized Machine Learning and Large Language Model Agents for Digital Health Monitoring and Intervention. Ph.D. Thesis, University of California.
- Perez Verástegui, J.F., Ortega Rojas, Y.K., Casazola Cruz, O., et al . (2025) Guide to the Use of Generative Artificial Intelligence in Education and Research. UNESCO Press.
- Abedi, S. (2025) Integrating DMN and LLM for Automated Student Feedback and Support. Ph.D. Thesis, Politecnico di Torino.
- Yang, Y., Shin, A., Kang, M., et al . (2024) Easy Come, Easy Go? Examining the Perceptions and Learning Effects of LLM-Based Chatbot in the Context of Search-as-Learning. arXiv: 2410.01396.
- Li, X. and Wang, Y. (2023) Developing a Measurement Scale for Senior High School Students’ Data Literacy. Chinese Journal of Mathematics Education , 16, 45-52.