English Proficiency Homogenization Using Norm-referenced and Criterion-referenced Tests: A Structural Equation Modeling Approach

Abstract

This correlational research used the association between norm-referenced (NRTs) and criterion-referenced tests (CRTs) to predict CRT scores on the basis of NRT scores, homogenize English proficiency of university students, and design a structural equation modeling approach between NRTs and CRTs. The participants were 210 allied health EGP (English for General Purposes) students who were assigned to three levels of language proficiency using Cambridge Placement Test. Researcher-made midterm and final exams, focusing on grammar, vocabulary, and reading comprehension were conducted. Results showed significant positive correlations between the NRT and CRTs. Structural Equation Modeling (SEM) analysis indicated significant paths from NRT as the dependent latent variable to CRTs as independent latent variables. The scores on the components of the three latent variables including vocabulary, grammar, and reading, within three assessments (placement, midterm, and final) were considered as observed variables. Significant paths between NRT and CRTs suggested that complex interrelations between components of NRT and CRTs can be used to homogenize university students’ proficiency in academic English courses using NTR scores to overcome problems related to the individual differences. Consequently, in academic English courses, groupings based on university students’ language ability using NRT scores would override groupings solely based on students’ academic majors.