Data informativity is a crucial property to ensure the consistency of the prediction error estimate. This property has thus been extensively studied in the open-loop and in the closed-loop cases. In this paper, we consider data informativity in the case of dynamic network identification. In particular, we derive a number of data informativity conditions for the prediction error identification of a particular module of a dynamic network using a direct identification approach. An optimal experiment design approach is also proposed to distinguish between the different situations leading to data informativity.