Partial Least Squares Structural Equation Modeling for Language Education Research

Abstract

Partial least squares structural equation modeling (PLS-SEM) is a statistical analysis technique that has been widely adopted in marketing and strategic management disciplines, yet there is a paucity of output in higher education research, and particularly so in language education. Indeed, social science (including education) researchers have traditionally relied on first-generation techniques such as regression-based approaches and various types of factor analysis (Fornell, 1982, 1987). For the last two decades, however, second-generation techniques such as PLS-SEM has become increasingly popular. The PLS-SEM approach offers multivariate perspectives to predict complex causal relationships, making it highly suitable for developing theories in exploratory research. In this presentation, I will, from a language teacher-researcher perspective, illustrate how a PLS-SEM-based educational study can be conducted and offer solutions to problems that may arise. Examples used in this presentation are taken from our in-house Learning Transfer pilot project that aims to investigate first-year undergraduate students’ perceptions of the skills taught in a general English for Academic Purposes (EAP) course, called ‘Core University English’ (CUE), offered by the Centre for Applied English Studies (CAES), the University of Hong Kong (HKU). This study is one of the first attempts to adopt the PLS-SEM approach in language educational research.