[Paper Review] The Role of Social Networks in Information Diffusion

Eytan Bakshy, Itamar Rosenn+2 more|arXiv (Cornell University)|Jan 19, 2012

Opinion Dynamics and Social InfluencePhysics and Astronomy28 references137 citations

TL;DR

This study uses a large-scale randomized experiment on Facebook with 253 million users to isolate the causal effect of social signals on information diffusion. It finds that exposure to friends' sharing behavior significantly increases both the likelihood and speed of information sharing, and while strong ties are individually more influential, weak ties—due to their greater abundance—drive the majority of novel information diffusion.

ABSTRACT

Online social networking technologies enable individuals to simultaneously share information with any number of peers. Quantifying the causal effect of these technologies on the dissemination of information requires not only identification of who influences whom, but also of whether individuals would still propagate information in the absence of social signals about that information. We examine the role of social networks in online information diffusion with a large-scale field experiment that randomizes exposure to signals about friends' information sharing among 253 million subjects in situ. Those who are exposed are significantly more likely to spread information, and do so sooner than those who are not exposed. We further examine the relative role of strong and weak ties in information propagation. We show that, although stronger ties are individually more influential, it is the more abundant weak ties who are responsible for the propagation of novel information. This suggests that weak ties may play a more dominant role in the dissemination of information online than currently believed.

Motivation & Objective

• To isolate the causal effect of social network signals on information sharing behavior in a real-world online environment.
• To resolve the confounding effects of homophily and shared information sources in observational data on social influence.
• To empirically measure the relative roles of strong and weak ties in spreading novel information online.
• To determine whether exposure to friends' sharing behavior on Facebook's newsfeed causally increases user sharing behavior.
• To quantify how tie strength influences the propagation of information in large-scale social networks.

Proposed method

• Conducted a large-scale randomized field experiment on Facebook with 253 million users.
• Randomized exposure to social signals (friends' sharing behavior) on the newsfeed, creating two conditions: exposed and not exposed.
• Compared the rate and likelihood of information sharing between exposed and non-exposed users to isolate causal effects.
• Used observational data from the 'no feed' condition to estimate baseline sharing due to homophily and external influences.
• Measured tie strength based on frequency of interaction and used it to compute the overall contribution of strong vs. weak ties to information diffusion.
• Analyzed temporal correlation in sharing behavior and the cumulative effect of multiple sharing friends on user sharing propensity.

Experimental results

Research questions

• RQ1 Does exposure to friends' information sharing on Facebook's newsfeed causally increase the likelihood and speed of user sharing?
• RQ2 To what extent do strong ties versus weak ties contribute to the diffusion of novel information in online social networks?
• RQ3 How does the presence of multiple friends who share the same content affect an individual’s likelihood to share?
• RQ4 What is the relative contribution of social influence via the newsfeed versus independent sharing due to shared information sources or external communication?
• RQ5 How does tie strength correlate with both influence and access to novel information in real-world social networks?

Key findings

• Users exposed to friends’ sharing behavior on the newsfeed were significantly more likely to share information and did so more quickly than those not exposed.
• The probability of sharing in the 'no feed' condition—representing homophily and external influence—served as an upper bound on non-causal sharing, while the difference between conditions provided a lower bound on causal influence via the newsfeed.
• Although strong ties are individually more influential, weak ties are responsible for the majority of information diffusion due to their greater abundance.
• Multiple sharing friends have a cumulative causal effect on an individual’s propensity to share, indicating a simple contagion process.
• Exposure to social signals on the newsfeed significantly increases the spread of information, demonstrating a causal role of social networks in information diffusion.
• The results generalize Granovetter’s theory of weak ties to everyday information diffusion, showing that weak ties are critical for spreading novel information in large online networks.