Using Large Language Models as Formative Assessors Evaluating Students' Mathematical Modelling Competence with ChatGPT

Supplementary Materials Due to journal page limitations, the complete record of ChatGPT interactions—including prompt engineering iterations, rubric development discussions, and detailed assessment justifications for each student work sample—is provided as online supplementary material. The supplementary material includes: (1) initial conceptual dialogues defining mathematical modelling and modelling competence; (2) refinement of the modelling cycle framework; (3) collaborative rubric development with ChatGPT; (4) anonymized student work samples from the school bus parking task; and (5) ChatGPT's complete assessment justifications for all six student groups across three competencies. Readers are encouraged to consult these materials for a comprehensive understanding of the prompt engineering methodology and assessment protocol.