How Reproducible Are LLM-Generated Q-Matrices? Evidence from Downstream Classification

Minkwon Kim


Abstract
Regenerating LLM Q-matrices ten times per model on TIMSS 2011 items, we find two runs of the same model reassign 61–88% of student mastery profiles. Greedy decoding removes this instability; disagreement with expert judgment (63–76%) survives. Majority voting fixes neither. Report distributions, not single runs.
Anthology ID:
2026.aimecon-main.73
Volume:
Proceedings of the Artificial Intelligence in Measurement and Education Conference (AIME-Con): Full Papers
Month:
October
Year:
2026
Address:
Wyndham Grand Pittsburgh Downtown, Pittsburgh, Pennsylvania, United States
Editors:
Joshua Wilson, Christopher Ormerod, Magdalen Beiting-Parrish
Venue:
AIME-Con
SIG:
Publisher:
National Council on Measurement in Education (NCME)
Note:
Pages:
649–655
Language:
URL:
https://aclanthology.org/2026.aimecon-main.73/
DOI:
Bibkey:
Cite (ACL):
Minkwon Kim. 2026. How Reproducible Are LLM-Generated Q-Matrices? Evidence from Downstream Classification. In Proceedings of the Artificial Intelligence in Measurement and Education Conference (AIME-Con): Full Papers, pages 649–655, Wyndham Grand Pittsburgh Downtown, Pittsburgh, Pennsylvania, United States. National Council on Measurement in Education (NCME).
Cite (Informal):
How Reproducible Are LLM-Generated Q-Matrices? Evidence from Downstream Classification (Kim, AIME-Con 2026)
Copy Citation:
PDF:
https://aclanthology.org/2026.aimecon-main.73.pdf