Education Commons^™

Structural Equation Models With Small Samples: A Comparative Study Of Four Approaches, Frances L. Chumney

College of Education and Human Sciences: Dissertations, Theses, and Student Research

The purpose of this study was to evaluate the performance of estimation methods (Maximum Likelihood, Partial Least Squares, Generalized Structured Components Analysis, Markov Chain Monte Carlo) when applied to structural equation models with small samples. Trends in educational and social science research require scientists to investigate increasingly complex phenomena with regard for the contextual factors which influence their occurrence and change. These additional layers of exploration lead to complex hypotheses and require advanced analytic approaches such as structural equation modeling. A mismatch exists between analytic technique and the realities of applied research. Structural equation modeling requires large samples in general …