Model selection in exploratory latent class and mixture models: What's to be preferred?

Applications of exploratory latent class and mixture models are greatly increasing across social, behavioral and medical research. However, I will argue that these models are not being used optimally, either for the particular substantive domain or in terms of fitting with general principles in science. Following a brief review of general research approaches, preferences and principles, typical model selection practice used with latent class and mixture models is located in this broad context. It is argued that confirmatory and null-hypothesis testing approaches have been automatically and inappropriately imported to the growing body of exploratory latent class and mixture model applications. Current practice follows from a null-hypothesis testing tradition, emphasizing parsimony and simplicity. However, latent class and mixture models follow more naturally from a model fitting framework, which defaults toward more complexity, rather than simplicity. A rationale for favoring a higher degree of complexity and a hypothesis generative approach to the use of these techniques is given.