Smith Carter L, Hantula Donald A
BioVid Corporation, Princeton, New Jersey.
Behav Res Methods. 2008 Nov;40(4):940-53. doi: 10.3758/BRM.40.4.940.
An experimental comparison of two commonly used delay-discounting procedures (binary choice and fill in the blank) and modes of administration (paper and pencil and computer based) was conducted. Statistically significant main effects were found for task type--steeper discounting was observed in the binary-choice task--but not for mode of administration. As simple evidence of validity, hyperbolic curves consistently provided a better fit to the data than did exponential curves for both tasks. Further, magnitude effects were also observed across conditions. Correlational results varied largely as a function of the discounting index (either k or area under the curve) under consideration. Across the two tasks, discounting indices showed modest levels of reliability [r(AUC) = .33; r(k) = .75]. The findings pointed to refinements for both the methodology and criteria that are used to study delay discounting and raised questions about the commonly assumed relationship between discounting and the construct of impulsivity.
对两种常用的延迟折扣程序（二元选择和填空）以及施测方式（纸笔和基于计算机）进行了实验比较。发现任务类型有统计学上的显著主效应——在二元选择任务中观察到更陡峭的折扣——但施测方式没有。作为有效性的简单证据，对于这两项任务，双曲线曲线始终比指数曲线更能拟合数据。此外，在不同条件下也观察到了量级效应。相关结果在很大程度上因所考虑的折扣指数（k或曲线下面积）而异。在两项任务中，折扣指数显示出适度的可靠性水平[r(AUC)=0.33；r(k)=0.75]。研究结果指出了用于研究延迟折扣的方法和标准的改进之处，并对折扣与冲动性结构之间通常假定的关系提出了疑问。
