The use of virtual reality (VR) in archaeological research is increasing year over year. Nevertheless, the influence of VR technologies on researchers' perception and interpretation is frequently overlooked. These device-induced biases require careful consideration and mitigation strategies to ensure the integrity and reliability of archaeological research results. Our aim is to identify potential interpretation biases introduced by the use of VR tools in this field, by analyzing both eye-tracking patterns and participant's behavior. We have designed an experimental protocol for a user study involving an analysis task on a corpus of archaeological artifacts across different modalities: a real environment and two virtual environments, one using a head-mounted device and the other an immersive room. The aim of this experiment is to compare participants' behavior (head movements, gaze patterns and task performance) between the three modalities. The main contribution of this work is to design a methodology to generate comparable and consistent results between the data recorded during the experiment in the three different contexts. The results highlight a number of points to watch out when using VR in archaeology for analysis and interpretive purposes.