ABSTRACT
One way of interpreting the relationship measured between variates in multidimensional data is to use a model assuming a conceptual, hypothetical variate (latent factor) of a lower dimension mutually related to the observed variates. Such models are collectively referred to as latent structure models and have their origin in sociologist Lazarsfeld (7). The latent class model in sociology, the factor analysis model in psychology, and the latent trait model in education and ability measurement fields are particularly common; they have been subjected to many studies historically and are high in demand in the field of practice.
