In this paper, we conduct agent-based simulation experiments for network formation analysis. In the published papers, Bala and Goyal (2000) have constructed a mathematical model leading a star network to be strict Nash equilibrium. However, Berninghaus et al. (2007) have conducted the laboratory experiments using human subjects basing on the mathematical model, and the result of the experiments indicates that human subjects do not always make decision just as the mathematical model predicted. In this paper, we propose a simulation model using the adaptive artificial agents to clarify the reason of the deviation from the mathematical predictions.