From Test Performance to Language Use: Using Self-Assessment to Validate a High-Stakes English Proficiency Test

Despite the profusion of studies into the factor structure of language tests, limited research is currently available on the relationship between test performance and language use in TLU domains. Utilizing the structural equation modeling approach, this study set out to investigate the factor structure of a high-stakes university-based English proficiency test, and then modeled the relationship between test takers’ performance and their ability to use language in TLU domains. A self-assessment (SA) inventory was developed and validated to capture test takers’ ability to use language in TLU domains. The results showed that the higher-order factor model best fit both the test and SA data. Structural regression analysis indicated a moderately strong relationship between students’ general EFL ability and the SA latent factor. The results of this study lend empirical support to the construct and predictive validity of the test, both of which are deemed crucial to its overall validity argument. The results also shed light on the utility of SA to capture language use in TLU domains as well as the use of SA for test validation purposes.