Contemporary Considerations in Latent Variable Modeling

This three-paper dissertation examines contemporary approaches that education, social, and behavioral researchers use in measuring latent constructs, ensuring equity, fairness, and unbiasedness, and minimizing regression coefficient bias and uncertainty in the relationship between an L2 latent reflective predictor variable and a global outcome variable. Specifically, the first paper focuses on ensuring equity, fairness, and unbiasedness of the English and Spanish versions of the Authoritative School Climate Survey, while the second paper examines the problem of inflated factor correlation in confirmatory factor analyses when compared to exploratory structural equation modeling using three rotations. The third paper uses simulated conditions to examine bias that can be introduced when using the doubly manifest approach to estimate the relationship between an L2 latent reflective predictor variable and a global outcome variable, and under what conditions this approach is feasible compared to the generally accepted doubly latent approach.