The Influence of Early Respondents: Information Cascade Effects in Online Event Scheduling

Abstract

Sequential group decision-making processes, such as online event scheduling, can be subject to social influence if the decisions involve individuals’ subjective preferences and values. Indeed, prior work has shown that scheduling polls that allow respondents to see others’ answers are more likely to succeed than polls that hide other responses, suggesting the impact of social influence and coordination. In this paper, we investigate whether this difference is due to information cascade effects in which later respondents adopt the decisions of earlier respondents. Analyzing more than 1.3 million Doodle polls, we found evidence that cascading effects take place during event scheduling, and in particular, that early respondents have a larger influence on the outcome of a poll than people who come late. Drawing on simulations of an event scheduling model, we compare possible interventions to mitigate this bias and show that we can optimize the success of polls by hiding the responses of a small percentage of low availability respondents.