Glöckner Andreas, Hilbig Benjamin E, Henninger Felix, Fiedler Susann
Department of Psychology, University of Hagen.
Department of Psychology, University of Koblenz-Landau.
J Exp Psychol Gen. 2016 Apr;145(4):486-508. doi: 10.1037/a0040103.
Previous literature has suggested that risky choice patterns in general--and probability weighting in particular--are strikingly different in experience-based as compared with description-based formats. In 2 reanalyses and 3 new experiments, we investigate differences between experience-based and description-based decisions using a parametric approach based on cumulative prospect theory (CPT). Once controlling for sampling biases, we consistently find a reversal of the typical description-experience gap, that is, a reduced sensitivity to probabilities and increased overweighting of small probabilities in decisions from experience as compared with decisions from descriptions. This finding supports the hypothesis that regression to the mean effects in probability estimation are a crucial source of differences between both presentation formats. Further analyses identified task specific information asymmetry prevalent in gambles involving certainty as a third source of differences. We present a novel conceptualization of multiple independent sources of bias that contribute to the description-experience gap, namely sampling biases and task specific information asymmetry on the one hand, and regression to the mean effects in probability estimation on the other hand.
以往文献表明，一般而言，风险选择模式——尤其是概率加权——在基于经验的决策中与基于描述的决策中显著不同。在两项重新分析和三项新实验中，我们使用基于累积前景理论（CPT）的参数方法，研究基于经验的决策和基于描述的决策之间的差异。一旦控制了抽样偏差，我们始终发现典型的描述-经验差距出现了逆转，也就是说，与基于描述的决策相比，基于经验的决策中对概率的敏感性降低，对小概率的过度加权增加。这一发现支持了以下假设：概率估计中的均值回归效应是两种呈现形式之间差异的关键来源。进一步的分析确定，在涉及确定性的赌博中普遍存在的任务特定信息不对称是差异的第三个来源。我们提出了一种新颖的概念化方法，即导致描述-经验差距的多个独立偏差来源，一方面是抽样偏差和任务特定信息不对称，另一方面是概率估计中的均值回归效应。
