Use and misuse of receiver operating characteristic analysis in eyewitness identification.
Use and misuse of receiver operating characteristic analysis in eyewitness identification.
复制标题
接收器操作特征分析在目击者识别中的使用和误用。
DOI:
10.1016/j.jarmac.2021.06.003
复制
发表时间:
2021
影响因子:
4.2
通讯作者:
Moody, Sarah A.
中科院分区:
文献类型:
--
作者:
Yang, Yueran;Moody, Sarah A.
Comments on an article by MB Kovera and AJ Evelo (see record 2021-79725-001). Kovera and Evelo Kovera and Evelo noted the importance of understanding social context effects on eyewitness behavior as well as social context’s decline in eyewitness research due to the rise of receiver operating caracteristic (ROC) analysis. ROC analysis should not dim the importance of social factors in eyewitness research. Instead, the proper use of ROC analysis should consider how social context influences eyewitness behavior, which in turn influences the legal system’s objective to maximize the expected utility of investigator classification decisions when using eyewitness evidence. Two misconceptions have hindered the proper use of ROC analysis in eyewitness research. The first misconception is that an ROC curve describes eyewitness operating characteristics. Rather, an ROC curve describes investigator operating characteristics when investigators use eyewitness evidence to judge suspects’ guilt. The second misconception is that ROC analysis relies on AUCs to assess performance. Rather, EU lines should be used jointly with ROC curves to quantify and compare investigator performance. ROC analysis’s reliance on the expected utility metric also enlightens other possibilities to improve investigator performance beyond just developing alternative lineup procedures. The legal system also needs to consider other parameters involved in estimating the expected utility, including investigators’ actual operating points, the prior probability of guilt, and the utilities of classification outcomes.(PsycInfo Database Record (c) 2022 APA, all rights reserved)