Study warns PCA misestimates personality dimensions in LLM embeddings

A new study in generative psychometrics compared PCA with Exploratory Graph Analysis (EGA) across three LLMs including GPT-4o, finding that EGA perfectly recovered the six-factor personality structure while PCA almost entirely failed, prompting researchers to call for abandoning PCA on LLM embeddings.

2026-10-06 ~ 2026-10-06 · 2 related posts

• PCA vs EGA on LLM embeddings: EGA recovers 6-dim structure 97.6–100%, PCA nearly 0% — GolinoHudson · 2026-10-06
• Researchers warn: PCA misestimates personality structure in LLM embeddings, EGA works better — GolinoHudson · 2026-10-06