A mediator-guided LLM simulation selects survey items whose virtual convergent validity matches human ground truth better than random or LLM-judge baselines.
Jongwook Han, Dongmin Choi, Woojung Song, Eun-Ju Lee, and Yohan Jo
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Psychometric Item Validation Using Virtual Respondents with Trait-Response Mediators
A mediator-guided LLM simulation selects survey items whose virtual convergent validity matches human ground truth better than random or LLM-judge baselines.