Dr. Grant B. Morgan, associate professor, published "A Comparison of Diagonal Weighted Least Squares Robust Estimation Techniques for Ordinal Data" in Structural Equation Modeling, Vol. 21, Issue 3, p. 424-438, October 28, 2014, with co-author(s) Dr. Christine DiStefano, University of South Carolina.
Classification, generally, refers to the process of simplifying data by grouping observations that have similarities together. There are many applications of classification in educational and psychological research, and there are many ways to perform classification. My research in this area focuses on the latent variable, or model-based, approach to classification.
Psychometrics generally refers to the measurement of psychological phenomena and involves the application of advanced statistical techniques to psychological instruments. My research in this area focuses on providing validity evidence of existing instruments and exploring applications of existing psychometric ideas and models to new measurement scenarios. Ultimately, my research in psychometrics focuses on reaching correct conclusions about the existence and nature of psychological phenomena of interest.
Latent variable models involve the examination of one or more theorized but directly unobservable variables, such as ability, intelligence, or achievement. A latent variable cannot itself be directly measured, but instruments may be developed that allow one to observe the manifestation of the latent variable. Models that include one or more latent variables require complex mathematical computation to allow researchers to make decisions from the manifestations of the latent variable. My research in this area is related to the accurate estimation of relationships between latent and observed variables.