The emergence of online social networks has greatly facilitated the diffusions of information and behaviors. While the two diffusion processes are often intertwined, “talking the talk” does not necessarily mean “walking the talk”—those who share information about an action may not actually take part in the action. This study aims to understand whether the diffusion of information and behaviors are similar, and whether social influence plays an equally important role in these processes. Integrating text mining, social network analyses, and survival analysis, this research examines the concurrent spread of information and behaviors related to the Ice Bucket Challenge event on Twitter. We show that the two processes follow different patterns. Unilateral social influence contributes to the diffusion of information, but not to the diffusion of behaviors; bilateral influence conveyed via the communication process is a significant and positive predictor of both diffusion of behaviors and information. Based on the Bass diffusion theory, we find that the influence from bilateral social connections is a more significant predictor than the influence from unilateral social connections. In addition, when jointly modeling the two adoption behaviors, the prediction accuracy of behavior adoptions is significantly improved. These results have important implications for applying theories of social influence, social networks, and contagion to better understand individuals’ behaviors in passing information and taking actions in a social context.
