Recently, with the rapid development of social network services, political entities have employed social media to conduct political propaganda; they disseminate official announcements, promote candidates, and even attack the opposites. However, to influence public opinion, some of them will create sockpuppets to participate in discussions. To better understand the phenomenon, this study attempts to measure the behavioral differences between sockpuppets and normal users using network structure analysis and user behavior analysis. We observe and realize the behaviors of sockpuppets by collecting a dataset from PTT, a famous social forum in Taiwan. We propose three feature categories including authorbased, commenter-based, and network-based features. From the analysis results, we find that sockpuppets are more active and attractive than ordinary users. Moreover, we observe that sockpuppets play an essential role in information flow from the results of the network structure of user relationships.