ABSTRACT
This chapter focuses on exploring and examining additional extensions of the latent change score (LCS) framework. It introduces multiple ways for regularization to be used with the LCS model. The chapter proposes the incorporation of regularization for constraining parameters within the LCS model to allow for simpler, more flexible model testing. With longitudinal data, although a number of statistical frameworks are available for analysis; the use of structural equation modeling has become increasing popular. Hypotheses about change trajectories, determinants of change, along with other components can be formulated as a sequence of models, each representing a specific theoretical formulation. Regularization allows for more flexibility, in both simplifying the model testing process, while promoting the incorporation of more complex specifications. Across both univariate and bivariate LCS models, using both frequentist and Bayesian regularization allowed for a simplified process in choosing a best model, while also increasing the plasticity of the LCS model to incorporate additional parameterizations.
