In this paper we discuss how apparently simple indicators such as gender differences need to be interpreted with extreme care. In particular, we consider how the assessment framework (for instance, stimulus for the assessment, types of texts, question format) and the methodology (for instance, design, definition of the target population) of international surveys may have a potential impact on the results and on the indicators. Through analysis of PISA data we show how increases or decreases in the gender gap can, at least partially, result from variations in the framework or the methodology of the respective assessments. Discussant with a response to the paper presentations (to be listed as a paper with this name): Bob Lingard (University of Edinburgh, Scotland): PISA in relation to discourses about gender: Relates the PISA study and the presentation to the discourses about gender and educational policy.