This is the summary of an article by Marco A. Janssen, which introduces a model to understand collective action in social networks. You can get the pdf of the original behavioral targeting article here: Targeting Individuals to Catalyze Collective Action in Social Networks.
Our society faces a lot of problems that require collective action, including reducing carbon footprints to respond to the environment’s drastic climate change. Collective action is studied by social scientists, trying to understand this phenomenon in a large scale.
Government rules and legislation doesn’t necessarily cause behavioral change. The authors of this study would like to look for behavioral studies whose findings may solve collective action problems. One result states that people tend to participate more in collective action through social influence, e.g. reputation and social pressure.
Thus, the ability to target persons who are influential may cause a lot of cooperative behavior. In fact, this kind of cascading can be seen in social networks. Many studies regarding spreading social network influence is related to viral marketing, where the main focus is information spreading.
It has been shown that behavioral changes spread much more rapidly in a clustered network, rather than a random network. This is different to information spreading, that’s why behavioral change diffusion is different from the diffusion of spreading information. Peer Pressure plays a role here; a new behavior is only adopted when a user sees that most of his or her neighbors have adopted that behavior.
The paper discusses a rigorous mathematical and computational model of agents creating decisions and contributing to a problem on collective action. The aim is to compare various strategies of increasing collective action, among others.
The model consists of agents, located in a network, making decisions between A and B.
The network neighbors affects the decision of the agent. A behavior, therefore, has a both and individual and a social influence part, taking into account an agent’s personal preference as well.
Social influence increases when more of an agent’s neighbors exhibit a certain behavior. In addition, homophily effects are also considered. Interventions are also added, which may influence an agent to change its behavior from A to B.
The paper aims to determine the most influential agents to adopt singled out behavior that leads to cascading adoption. The study can extend to make the simulation closer to human behavior, such as changes in product preferences. Eventually the effectiveness of this model can be tested against empirical data.
Some interesting findings in the results of this study include the following. If agents are given feedback regarding the adoption in their network, rather than distributing that information on a global scale, the behavior’s desired adoption increases by up to 400 percent. This means that targeted information actually increases rates of adoption.