Socially Embedded Knowledge Networks and the Making of Opinion Leaders: Evidence from Twitter

This paper focuses on knowledge markets exploring how network relationships between knowledge consumers impact the equilibrium number of opinion leaders. Both a theoretical model and empirical analysis show that there’ll be more opinion leaders in a knowledge market if the most active knowledge consumers occupy more central positions in a social network connecting consumers. The model formalizes the following story. Knowledge consumers are embedded in network relationships through which they influence each other on which opinion providers they pay attention to. If the most active (thus most capable to influence) knowledge consumers occupy more central network positions, consumer attention gravitates toward some opinion providers, and this turns more opinion providers into opinion leaders. The model inspires and is supported by empirical analysis using a Twitter network and associated tweets. First, unsupervised machine learning is used to define knowledge markets: topic modeling finds 45 topics in tweets, network community detection yields 4 nearly isolated Twitter sub-networks, and a knowledge market is then defined by a combination of one topic and one sub-network. Second, with each knowledge market being a unit of observation, we define variables and test our theoretical predictions. This is the first paper to formally define opinion leaders, knowledge markets, and consumer attention. While the existing literature emphasizes the role of opinion providers’ network positions on the making of opinion leaders, this work shows the network positions of active consumers matter because active consumers serve as a propagation machine.