The impact of decision support system features on user overconfidence and risky behavior

There is considerable research on how Decision Support Systems (DSSs) enable users to make better decisions. However, there is less focus on the possibility that some of their features may introduce biases and encourage suboptimal user behavior. We report on an experimental study that examines three DSS features that are generally considered beneficial to the user: the degree of choice the system provides to the user, the presence of competition among concurrent users, and the use of training to increase system familiarity. We hypothesize that the three DSS features may increase risky behavior (measured as the amount invested in a stock investment task with random outcomes) and overconfidence, conceptualized as the illusion of control, the phenomenon whereby people believe their chances of success at a task are greater than would be warranted by objective analysis. Our results confirm the effects of the three DSS features on risky behavior but only degree of choice impacts overconfidence. Moreover, overconfidence does not appear to mediate the impact of the DSS features on risky behavior. Finally, we hypothesize and confirm that, controlling for the effect of actual performance, overconfidence increases user satisfaction with the decision-making process and outcome.