任务框架和序列趋势对趋势阻尼的影响

The Effect of Frame and Series Trend on Trend Damping

以往研究发现,下降序列的趋势阻尼大于上升序列,但该结论可能掺杂了任务框架这一额外变量。基于此实验1选取可控事件,实验2选取不可控事件作为背景,探讨了任务框架和序列趋势对趋势阻尼的影响。研究发现:不管在可控事件还是不可控事件中,序列趋势对趋势阻尼的影响都受到任务框架的调节,收益框架中下降序列的阻尼大于上升序列,损失框架中两者差异不显著。

Generally, individuals make forecasting in two ways： statistic forecasting and judgmental forecasting. Judgmental forecasting is characterized by systematic biases because of its subjectivity. Trend damping, one of these biases, refers to that individuals tend to underestimate future values for upward trend, and overestimate them for downward ones when forecasting from time-series with noise. In other words, people underestimate the steepness of the series＇ trend. Previous research found that damping effect in upward trend was larger than that in downward trend. However, most experiments on trend damping require participants to forecast quantities for which values are better than lower ones, such as the sales of a good. It is possible that downward trends represent a situation of perceived losses, and that upward ones represent perceived gains. And many studies have found that the way a problem is expressed, or framed, can dramatically influence judgment. Some research further showed that trend damping in the frame of loss was larger than that in the flame of gain. This means that previous results may mix the effect of frame. Based on the shortcoming in previous studies, present research aimed to investigate the effect of frame and trend on trend damping through two experiments. Previous research suggested that the effect of frame and series trend on trend damping may result from two reasons. First, participants＇ forecasts may have been subject to an optimistic bias. Second, individuals may have expected actions to be taken to reverse downward trends but not to reverse upward ones. Optimistic bias may still exist under uncontrollable events whereas reverse expectation may not. However, research suggested that optimistic bias and reverse expectation may weaken under uncontrollable events. Since then, experiment 1 examined the effect of frame and trend on trend damping under controllable events, and experiment 2 explored the effect of frame and trend on trend damping under uncontrollable events. Among these two experiments, the materials were time series which were constructed using power-law functions, of the general form： y=100＋300x（x/48）k ＋error. The dependent variable was the D-value of predictive value and truth-value. Results showed that： （1） The truth-values were significantly bigger than predictive values for upward trends, whereas significantly smaller than predictive values for downward trends, in other words, significant damping effect was occurred. （2） The damping effect was greater when the slope of time series was bigger. （3） The interaction effect of gain-loss frame and series trend was significant under controllable and uncontrollable situations. The simple effect analysis both suggested that the damping effect was greater in downward trends than in upward ones in the frame of gain, but there was no difference between both trends in the frame of loss. In the frame of gain, the downward trends are always negative, and the upward ones are always positive. According to the optimistic bias, individuals may expect that the negative thing was less likely to occur, which may result in greater damping effect in downward trend. The reverse expectation suggested that individuals may hypothesize some actions to be taken to reverse the negative thing, which may also result in greater damping effect in downward trend. Adaptation account of trend damping proposed that people more frequently experience data series that are increasing than data series that are decreasing. As a result, individuals develop expectations about how series typically change. These expectations may influence their forecasts; as a consequence the damping effect in the frame of loss has no difference.