Factorial Validity and Invariance of the MUET Writing Rating Scale: Empirical and Theoretical Correspondence

Abstract

Purpose â€“ The validity of any performance assessment depends substantially on the rating scale used in the assessment procedure. It delineates the theoretical construct being measured and influences how performances are judged and interpreted. This study examined the factorial validity and invariance of the rating scale used for the assessment of the essay writing component of the Malaysian University English Test (MUET) in terms of the correspondence to empirical data and the theoretical construct of ESL writing.   Methodology â€“ To determine the factorial validity of the MUET essay writing rating scale, a measurement model was developed based on the criteria description of the scale. Model-data fit was tested using CFA. The data comprised 392 essays, taken from a university English proficiency examination. Factorial invariance was tested through consecutively more restrictive models.   Findings â€“ 3-factor structure model was found to produce the best fit to the data. Factorial invariance was tested by examining the comparability of the structure and values of parameters within the model across two groups. A more restrictive model was subsequently tested, where variances and covariances of the latent constructs together with factor loadings were constrained to be equal for the two groups. The resultant goodness-of-fit indices showed a better fit to the data, providing support for the factorial invariance of the 3-factor structure model.   Significance â€“ The results support a 3-factor structure model, congruent with the theory of ESL writing. However, further validation needs to be carried out across different ESL populations to ensure accurate representation and assessment of ESL writing performance.