David Clarke (Melbourne) will use comparative data from the LPS study (mathematics classrooms) to discuss how differences can be created and ignored by the categories we use to generate and analyze international data. The pursuit of commensurability in international comparative research by imposing general classificatory frameworks can misrepresent valued performances, school knowledge and classroom practice as these are actually conceived by each community and sacrifice validity in the interest of comparability. The “validity-comparability compromise” is proposed as a theoretical concern with significant implications for international cross-cultural research. The presentation will draw on current comparative classroom data to illustrate a variety of aspects of the issue and its consequences for the manner in which international research is conducted and its results interpreted. The contrast and unexpected similarities offered by comparative analyses, he argues, reveal and challenge existing assumptions, theories and conceptual frameworks and make essential a reconstruction of some of our basic dichotomies as complementary elements in more integrated and inclusive theories.