Purpose - The purpose of this paper is to explore how hubs' social influence on social network decisions can cause the behavior of information cascades in a market. Design/methodology/approach - The authors establish understanding of the fundamental mechanism of information cascades through a computational simulation approach. Findings - Eigenvector centrality, betweenness centrality, and PageRank are statistically correlated with the occurrence of information cascades among agents; the hubs' incorrect decisions in the early diffusion stage can significantly cause misled shift cascades; and the bridge role of hubs is more influential than their pivotal position role in the process of misled shift cascades. Originality/value - This implication can be extendable in the field of marketing, sequential voting, and technology, or innovation adoption.