The fast growth of Internet of Things systems has transformed the way policy decisions are made based on the data produced by these devices. These data are prone to errors and anomalies and can present privacy and security issues which can affect policy decisions. Therefore, it is necessary that, at each layer of Internet of Things architecture, the data undergo a process to ensure its quality. This paper presents a framework for building a trust management system for Internet of Things devices using anomaly detection in different layers of IoT architecture and formulation of a reputation index.