Utilizing rich end user context information is viewed as one of the necessary approaches in developing more personalized mobile services and user experiences. The practical impact of end user context research and new opportunities in the field provided by emerging data collection methods such as handset-based measurements (i.e., collecting usage data directly from the end users' devices) have inspired new highly interesting large scale empirical context studies, but also brought quite diverse usage of the term context itself. Proper discussion and usage of context requires an unambiguous statement of how the term is understood in the particular case. On one hand the term should be positioned with the existing and commonly understood general definitions, but on the other hand it should also be acknowledged that especially an empirical research paper, or a context-aware service, can grasp only some specific aspects or elements of context. This paper proposes a context classification framework that aims to clarify the use of the term context in handset-based related end user studies. The framework is partly based on the experimental experience accumulated in our own handset panel studies. While helping researchers to plan context data acquisition and communicate and position the end user context elements used, the framework helps other stakeholders, such as application developers and service providers, to identify and utilize research and data most relevant for their particular needs. The paper also demonstrates the expressivity of the framework by examples.