These two distinct activities have been inadvertently mixed in comparative criminology thereby causing a great deal of confusion. The present paper argues that attention by comparative researchers to this distinction between theory testing and theory construction will contribute to the resolution of one major methodological problem: that of research site selection. We suggest that in light of the distinction drawn, the testing of the universality of a proposition is most appropriately done in research comparing Western and non-Western nations ('dissimilar design'). But comparative research aimed at the specification of a proposition is best conducted by comparing Western nations ('similar design'). (Publisher abstract)