Opinions and beliefs are essential ingredients in the diffusion of innovation. We present a framework to model and simulate diffusion processes of opinions and beliefs in interpersonal communication networks. We introduce an algorithm to create stylized networks with attributes of real world interpersonal communication networks. We also introduce a simple, but expandable model for simulating the dynamics of the diffusion processes of opinions and beliefs. We apply network multiagent simulations

more »

... to show the importance of local clusters of connected agents for keeping opinions and beliefs endemic in a social system. We also argue that due to the structure of interpersonal networks, these local clusters have the capability to propagate opinions and ideas into the entire system. 3 influence to diffusion patterns. Therefore, the algorithmically created networks have to fit the attributes of real world interpersonal networks [Hamill and Gilbert, (2009)] . In a nutshell, to analyze the diffusion processes of opinions and beliefs in interpersonal networks, we need an adequate network model of real world interpersonal networks and a congruent diffusion model to describe the propagation of opinions and beliefs in these networks. With these models, we have the ability to study which structural patterns influence which kind of diffusion positively. This is important in the context of diffusion intervention. For example, technology manager and marketers are interested in how to better start diffusion campaigns. Political players need assessment in detecting which protests can lead to uprisings, and which cannot because of unstable underlying networks. Siebers et al. [2010] stated the lack of "frameworks or methodologies to guide researchers and analysts through the agent-based modeling and simulation process." This is what we are offering with this article. We are going to present a framework to model diffusion processes of opinions and beliefs in interpersonal communication networks. In section 2 we elaborate the network model for the simulation experiments. We discuss the structural attributes of networks, which are formed by personal communication between the people of a real world social system and we offer an algorithm to create stylized networks having these attributes as well as ways to calibrate this algorithm using real world data. In section 3, we discuss differences between the diffusion of infectious diseases, information, and opinions and beliefs and present the dynamic model for our simulation experiments. In section 4, we simulate these models using multi-agent simulations. We show that locally connected clusters of infected agents play an important role for the stabilization of new opinions and beliefs and their propagation through entire social systems. Section 5 discusses the results and interpretations of the simulation experiments and gives guidelines for increasing the chance of successful diffusion processes of opinions and beliefs in interpersonal communication networks. Finally, in section 6, outlooks and future work are presented. Modeling the Structure of Interpersonal Communication Networks The members of a social system and their connections, which are used for communication, are essential elements in Rogers' [1995] definition of the diffusion of innovation. These diffusion elements can be modeled using network analytical models [Valente (1995) ]. Social networks [Wasserman and Faust, (1995) ] consist of a set N of agents and a set E of connections (edges) between these agents. Because the network model is independent from its content, the entities of a network are often called nodes. Nodes are connected by edges representing corresponding interactions between nodes. As nodes in our networks represent humans, we use the term agent to describe a single entity; the terms edges, connections, and links are used interchangeably. The edges in interpersonal communication networks are constructed by human-to-human interaction, which can be face-to-face, e-mail, or telephone communication. We explicitly exclude mass media communication from the considerations of this article. Social network analysis created a vast amount of methods [Wasserman and Faust, (1995) ] and theories [Freeman, (2004) ]. To describe the structure of interpersonal communication networks and their dynamics we need the following definitions. Two agents a and b in a network are directly connected if there exists an edge e(a,b); these