Paper people versus direct observation: An empirical examination of laboratory methodologies
用信号检测理论检验了纸人实验和直接观察两种实验室方法对个体判断的影响，发现只呈现关键行为（无噪音）时记忆和评价更准确、晕轮效应更少。
Abstract This study used Signal Detection Theory (SDT) measures to test competing predictions derived from stimulus presentation mode and signal‐to‐noise ratio hypotheses about differences between paper person and direct observation laboratory studies of individuals' judgments. Subjects viewed a videotape, read a script, or read a list of critical behaviors depicting one of two hypothetical college lectures. Subjects then completed a recognition test of memory for performance behaviors, and rated the lecturer's performance on five performance dimensions. Results supported the signal‐to‐noise hypotheses: behavioral memory and performance ratings were more accurate, and ratings contained less halo in the performance‐signal‐only (list) condition than in the signal‐plus‐noise (script and videotape) conditions. Implications for the conception of performance in laboratory manipulations and the potential effects of performance‐irrelevant information on rating outcomes are discussed.