Ji Zeng
OEAA psychometrician. Present April 23, 2026, and presenter of Session 5 — Simulation Studies on Subtest Indicators on April 24 — the full morning of day two.
Author of CT-PO vs SBAC Simulation Results Deck, the most technically substantial artifact in the corpus: a comparison of a Classification Tree with Proportional Odds against the Smarter Balanced subscore indicator method, with K-modes as reference, across two batches of simulated conditions plus real-data explorations at grades 5, 8, and 11.
The work criticises the existing subscore literature on two grounds — that it treats subscore reporting as prediction of true subtest performance, and that it assumes multivariate normality (citing Sinharay 2014 and Gorney & Sinharay 2025) — and demonstrates empirically that Michigan’s science subtest distributions are not normal.
The committee described the work as “very sophisticated” and offered support: Support Ji Zeng’s work through direct collaboration.
copula-approach-to-subtest-indicators is a response to her session, addressed to her.
Related. M-STEP Subscore Reporting