Sunbelt XXXI International Network for Social Network ... - INSNA

Coalitions In Exchange Networks: Some New ResearchBonacich, Phillip F.; Bienenstock, Elisa J.Mathematical and Statistical Network ModelsExchange, CoalitionsFRI.PM2Bonacich and Bienenstock (1997) proposed a model for coalition formation among actors in exchange networks. The model produced disjoint sets of actors inany exchange network who could improve their bargaining position if permitted to form coalitions that could act as bargaining units. Bonacich (2000) providedexperimental evidence for a modified version of the theory. However, there were some anomalous results in the experiment that suggested modifications inthe model: the Bonacich and Bienenstock model gave incorrect predictions for “weak power” networks; actors could sabotage coalitions of their opponents byco‐opting some of their members. Simpson and Macy (2001) also proposed a coalition model for exchange networks that can be integrated with theBonacich‐Bienenstock model. In this paper we describe a better theory that synthesizes the original Bonacich and Bienenstock theory, the anomalousexperimental results, and the Simpson and Macy model and propose an improved experiment. P. Bonacich and E.J. Bienenstock. 1997. “Latent Classes inExchange Networks: Sets of Positions with Common Interests,” Journal of Mathematical Sociology, 22:1‐28 P. Bonacich. 2000. “Patterns of Coalitions inExchange Networks: An Experimental Study.” Rationality and Society. 12:353‐373. B. Simpson and M. W. Macy. 2001. “Collective Action and Power Inequality:Coalitions in Exchange Networks.” Social Psychology Quarterly. 64:88‐100.Co‐evolution Model For Dynamic Social Network And BehaviorTong, Liping; Shoham, David; Cooper, RichardMathematical and Statistical Network ModelsExponential Random Graph Model, EM Algorithm, Co‐Evolution Model, Actor‐Based Stochastic Modeling, Continuous Behavior Variable, Longitudinal SocialNetworkSAT.PM2Individual behavior, such as screen time, physical activity, and eating habit, can be strongly influenced by the behavior of their friends. Meantime, the choice offriends can also be influenced by the preference of their behavior. To study the interdependence of social network and behavior, Snidjers et al. has developedthe actor‐based stochastic modeling (ABSM) methods for the co‐evolution process, which turns out to be useful when dealing with longitudinal social networkand behavior data when behavior variables are discrete and have limited number of possible values. Unfortunately, since the evolution function for behaviorvariable is in exponential format, the ABSM can generate unrealistic results when the behavior variable is continuous or has a large range. To realistically modelcontinuous behavior variable, we propose a co‐evolution process so that the network evolution is based on an exponential random graph model and thebehavior evolution is based on a linear model.