Two essays on individual vs group behavior and feedback structure in Judgmental forecasting

Abstract

Forecasting is necessary for organizations to prepare for the future for effective operations and this forecasting process largely relies on human judgment. It is therefore, important to examine ways to improve judgmental forecasting accuracy. This dissertation consists of two essays focusing on ways to improve judgmental forecasting performance. In the first essay, “Individual and Group Differences in Advice Taking Behavior: Perspectives in Judgmental Forecasting,” I examine advice taking behavior when individuals and groups make judgmental forecasts given external forecast advice. Using a controlled laboratory experiment, I investigate the differences between individuals and groups in the degree of advice taking and the ability to discern advice quality. Participants in an individual condition make decisions individually without any interactions with others whereas participants in a group condition interactively discuss with other two participants and agree on single decisions. I examine two studies regarding the distribution of advice quality

in Study 1, participants are randomly assigned to one of the two advice quality conditions where the same quality advice, either unreasonable or reasonable forecast advice, is given, and in Study 2, participants receive variable advice quality. In both studies, I find that groups are less receptive to advice than individuals and this relationship is mediated by confidence level

groups feel more confident than individuals, which decreases reliance upon advice. We also find that groups, who feel more confident, are able to discriminate the quality of advice better than individuals. In Study 2, I explicitly measure the disagreement level in a group and find that higher disagreement level in a group leads to lower confidence level, which then increases group’s reliance upon advice.

In the second essay, “The Effect of Relative Performance Feedback on Judgmental Forecasting Accuracy,” using a controlled laboratory experiment, I test the effect of providing relative performance feedback on judgmental forecasting performance. I show that relative performance feedback that provides rank information compared to other competing participants improves forecasting accuracy compared to the case without such feedback. We also find that the effect of relative performance feedback depends on the content of the feedback information as well as whether it is tied to additional financial reward. Relative performance feedback becomes more effective when subjects are informed that they rank behind other competing participants, compared to when subjects are informed that they rank above other competing participants. This finding is consistent with loss aversion: Low-ranked individuals view their performance as a loss and work harder to avoid this. By contrast, leading participants tend to slack off. Lastly, I find that the addition of monetary reward paid to top performing individuals reduces the effect of relative performance feedback, particularly for bottom performing individuals.