How to elicit and cease herding behaviour? On the effectiveness of a warning message as a debiasing decision support system

Abstract

Behavioural economics has been argued to be a productive basis for decision support system (DSS) research. Whereas traditional economics assumes that individuals make decisions independently of others, behavioural economists have shown that humans tend to follow the crowd in their decisions (i.e., exhibit herding bias). However, the literature is silent on how convincing the information on the decisions of the crowd needs to be to elicit herding bias and on whether herding can be reduced (i.e., debiased) by presenting a warning message. This paper addresses both questions in the contexts of financial decisions that were guided by two DSSs in the form of simulation tools. In particular, we conduct a randomised controlled trial with 768 respondents randomly assigned to peer information. The results indicate that the intervention successfully elicited herding bias and that herding occurs when respondents are informed that at least 50% of other people made a particular decision. The results further show that a DSS in the form of a warning message is not sufficient to debias herding. In conclusion, these findings showed that individuals are easily influenced by erroneous peer information and that this effect is robust against debiasing using a warning message. Hence, DSS developers need to consider more intense debiasing strategies to overcome herding.