1 paper
Jeonghyeon Moon, Jiwon Kim, Yeheum Lah +2
LLMs are increasingly used to simulate human survey responses, but prior work has mainly evaluated replication using mean-level or aggregate agreement, offering limited insight int…