Juslin P, Olsson H, Winman A
Uppsala University, Uppsala, Sweden
Organ Behav Hum Decis Process. 1998 Jan;73(1):3-26. doi: 10.1006/obhd.1998.2749.
In this comment on a recent article in this journal on calibration of subjective probabilities by Suantak, Bolger, and Ferrell (1996) we point to methodological problems with the data presented in the article, with computer simulations we demonstrate that the results claimed by the authors to refute the ecological models (Gigerenzer, Hoffrage, & Kleinbölting, 1991; Juslin, 1994) are actually predicted by them, and we respond to the criticisms of the notion of "ecological cue validity." It is further argued that contrary to the claims by the authors: The decision variable partition model advocated by the authors fails to account for the results reported in the article, but there is one error model-the combined error model (Juslin, Olsson, & Björkman, 1997)-that predicts the observed symmetric hard-easy effect with crossover at proportion correct .75. Finally, an analysis of 44 cognitive and 21 sensory tasks demonstrates that there is a difference in calibration for sensory and cognitive tasks. Copyright 1998 Academic Press.
在这篇对《苏安塔克、博尔格和费雷尔（1996年）在本期刊发表的关于主观概率校准的近期文章》的评论中，我们指出了该文章所呈现数据的方法学问题。通过计算机模拟，我们证明作者声称用以反驳生态模型（吉仁泽、霍夫拉格和克莱因博尔廷，1991年；尤斯林，1994年）的结果实际上是由这些模型预测出来的，并且我们回应了对“生态线索有效性”这一概念的批评。进一步论证的是，与作者的主张相反：作者所倡导的决策变量划分模型无法解释该文章所报告的结果，但有一个误差模型——组合误差模型（尤斯林、奥尔松和比约克曼，1997年）——能预测在正确比例为0.75时观察到的具有交叉的对称难易效应。最后，对44项认知任务和21项感觉任务的分析表明，感觉任务和认知任务在校准方面存在差异。版权所有1998年学术出版社。
