Gyeongcheol Cho, The Ohio State University

2024 Dissertation Prize

Structural equation modeling (SEM) enables empirical testing of hypothetical relationships between variables, including constructs. Traditionally, SEM has assumed all constructs to be latent, existing independently of their indicators, and represents them by a (common) factor. This domain is known as factor-based SEM. However, some constructs, such as socioeconomic status and genes, are not inherently latent but rather correspond to a summary or cluster of their indicators. To deal with such constructs, component-based SEM has emerged, wherein constructs are represented as composite indexes of indicators, termed components.

In this talk, I begin with a systematic comparison of the two SEM domains and briefly introduce three novel SEM methods—structured factor analysis (SFA), convex generalized structured component analysis (convex GSCA), and deep learning generalized structured component analysis (DL-GSCA). SFA tackles two long-standing issues in factor-based SEM—improper solutions and factor score indeterminacy—by simultaneously estimating model parameters and the probability distribution of (candidate) factor scores using a single cost function. Conversely, convex GSCA and DL-GSCA extend the capabilities of component-based SEM; the former generates interpretable components that preserve the original scales of indicators, while the latter employs deep learning to identify non-linear components that maximize predictive power for target outcome variables. I will demonstrate the technical foundations of these methods and their practically utility through empirical data analyses.

about the speaker

Dr. Gyeongcheol Cho is an Assistant Professor in the Department of Psychology at The Ohio State University. He earned his B.A. in Economics with a minor in Mathematics from Korea University in South Korea in 2017 and obtained his Ph.D. in Quantitative Psychology and Modeling from McGill University in Canada in 2023. His research focuses on advancing quantitative methods to explore and confirm hypothetical relationships between behavioral, psychological, and biological variables, with an emphasis on the measurement of theoretical constructs. Technically, his work encompasses factor/component analyses, structural equation modeling, and the integration of these methods with deep learning artificial neural networks. He has authored 20 publications in academic journals and developed user-friendly software, GSCA Pro (www.gscapro.com) and SFA Prime (www.sfaprime.com), to facilitate advanced statistical analysis.