Sternberg Saul, Backus Benjamin T
Department of Psychology, University of Pennsylvania.
Graduate Center for Vision Research, State University of New York, College of Optometry.
Psychol Rev. 2015 Oct;122(4):830-7. doi: 10.1037/a0039658.
It is sometimes suggested that reaction time (RT) distributions have the same shape across conditions or groups. In this note we show that this is highly unlikely if the RT is the sum of the stochastically independent durations of 2 or more stages (sequential processes) (a) that are influenced selectively by different factors, or (b) 1 of which is influenced selectively by some factor. We provide an example of substantial shape differences in RT data from a flash-detection experiment, data that have been shown to satisfy requirement (a). Ignoring these requirements, we also note that in a large range of instances reviewed by Matzke and Wagenmakers (2009) in which the ex-Gaussian distribution was fitted to RT data from different conditions in the same experiment, most sets of distributions fail to satisfy even a weak requirement for shape invariance. In the Appendix we describe the Summation Test for selectively influenced stages with independent durations (Roberts & Sternberg, 1993), and provide an example of its application. (PsycINFO Database Record
有时有人认为，反应时（RT）分布在不同条件或组间具有相同的形状。在本论文中，我们表明，如果反应时是2个或更多阶段（序列过程）的随机独立持续时间之和，且满足以下条件之一，那么这种情况极不可能发生：（a）这些阶段受到不同因素的选择性影响；或者（b）其中一个阶段受到某个因素的选择性影响。我们给出了一个闪光检测实验中反应时数据存在显著形状差异的例子，这些数据已被证明满足条件（a）。忽略这些条件，我们还指出，在Matzke和Wagenmakers（2009）回顾的大量实例中，将前高斯分布拟合到同一实验中不同条件下的反应时数据时，大多数分布集甚至都无法满足形状不变性的弱要求。在附录中，我们描述了针对具有独立持续时间且受选择性影响阶段的求和检验（Roberts & Sternberg，1993），并给出了其应用示例。（PsycINFO数据库记录）
