Tag
该论文对决策模型（decision models）和最新生成式 LLM 作为合成问卷受访者的能力进行了心理测量审计，在 14 个决策模型与 47 个生成式模型上测试后发现，模型在样本层面的模拟保真度尚可，但无法替代真实受访者的个体回答，且新一代模型未必优于旧版本。
A Pew Research Center study finds that methods to remove bogus respondents from online opt-in polls, such as trap questions and automated prescreening, can reduce error but do not provide a surefire solution, with voter file matching potentially increasing error.
This study evaluates silicon sampling methods, showing that fast modes outperform slow modes in efficiency and fidelity while highlighting limitations in accurately representing opinion variance.
Jenova AI's Survey Design Consultant tool helps improve survey quality by analyzing questions through the Tourangeau cognitive model, recommending validated scales, and ensuring optimal length.