Applying Agent-Based Models to the Prediction of Intention and Behavior

While the statistical approach can often help develop models that have predictive power, it rarely yields the insights required to build models that try to explain the underlying processes and mechanisms driving psychological or social phenomena. The Agent-Based Modeling [ABM] approach has been considered as a potentially useful approach to complement the existing literature and improve the explanatory ability of models. This contribution applied agent based modeling to social psychological models to examine whether they generated better predictions of intentions and behavior, in comparison to standard statistical approaches. Standard step-wise robust regression analysis and agent-based simulations were applied to consumer perception of products and reported behavior, taking into account population heterogeneity. Three key findings emerged. First, the agent-based simulation significantly improved the prediction of behavior from intention. Second, the improvement of prediction was inversely proportional to the complexity of the underlying social psychological model. The improvement was more noticeable for behavior, which is not usually well explained with standard statistical estimations. Finally, the introduction of varying degrees of rationality in agents’ behavior led to an improvement in the predictive ability of the simulations. The results of this exploratory study are discussed in the perspective of the potential of agent-based modeling as an alternative perspective to evaluating social psychological models.