EVALUATING THE VALIDITY OF MEARA AND ROGERS’ VOCABULARY APTITUDE TEST: A RASCH MEASUREMENT MODEL ANALYSIS

Abstract

Language aptitude tests, such as the LLAMA_ B3, developed by Meara and Rogers (2022), may play an important role in vocabulary learning research. Following Bokander & Bylund's (2019) advice, the construct validity of this test is examined here. Data collected from 314 participants were analyzed using Rasch analysis. All the eigenvalues fall below 2, supporting the assumption of uni-dimensionality. Q3 findings help us claim that the assumption of local independence has been met. This study found a .96 and a .84 reliability for the items and the participants, respectively. There was a convergent estimate of the results (Variance-Covariance Matrix: +2.002 and -2 log likelihood=+11.458.300). Accordingly, the test is valid for the present sample of students. As far as the analysis shows, the items represent a single underlying construct, meaning that the LLAMA_ B3 is fundamentally coherent. Psychometrically, a single trait seems to have been identified. It can be concluded that both the items and participants behaved predictably, indicating that the test is certainly worth further investigation and refinement.