Five different identification problems in mixture models are made explicit. Necessary and sufficient relationships among these problems of identification are analyzed using the concepts of weak and strong identification. This analysis is first particularized under a normality assumption and then used for LISREL models.
Mouchart, M., & San Martin, E. (1998). Identification problems in a class of mixture models with an application to the LISREL model (CORE Discussion Papers 1998/25). https://hdl.handle.net/2078.5/34521