Kowal Benjamin P, Yi Richard, Erisman Amanda C, Bickel Warren K
Center for Addiction Research, Fred and Dierk's Research Laboratories, Psychiatric Research Institute, University of Arkansas for Medical Sciences, Little Rock, AR 72205, USA.
Behav Processes. 2007 Jun;75(2):231-6. doi: 10.1016/j.beproc.2007.02.005. Epub 2007 Feb 8.
Two algorithms are commonly applied in computerized temporal discounting procedures (Decreasing Adjustment and Double-Limit Algorithms); however, the degree to which the two algorithms produce similar patterns of discounting is unknown. The present experiment compared the two common algorithms across sign (gains and losses) and magnitude ($10 and $1000) conditions. Twenty participants made choices between larger later and smaller sooner alternatives that were presented by each of the algorithms in separate conditions. Strong correlations were found between the two measures; however, the Decreasing Adjustment Algorithm tended to produce lower indifference points and higher rates of discounting than the Double-Limit Algorithm. Both algorithms found significant magnitude effects. Less consistent results were found when comparing the two algorithms across sign. The present results suggest that researchers should apply caution when making comparisons between outcomes of delay discounting studies that have used the two different algorithms. However, the interpretation of findings from individual studies is probably not strongly affected by the use of different computer algorithms.
两种算法通常应用于计算机化的时间折扣程序（递减调整算法和双限算法）；然而，这两种算法产生相似折扣模式的程度尚不清楚。本实验在正负（收益和损失）和金额大小（10美元和1000美元）条件下比较了这两种常用算法。20名参与者在由每种算法在不同条件下呈现的较大延迟和较小即时选项之间做出选择。两种测量方法之间发现了很强的相关性；然而，递减调整算法往往比双限算法产生更低的无差异点和更高的折扣率。两种算法都发现了显著的金额大小效应。在跨正负比较两种算法时，结果不太一致。本研究结果表明，研究人员在比较使用两种不同算法的延迟折扣研究结果时应谨慎。然而，个体研究结果的解释可能不会受到使用不同计算机算法的强烈影响。
