Abstract

Background: Despite the evolution of the International English Language Testing System (IELTS) to accommodate the evolving needs of test-takers and institutions, concerns have been raised regarding its reliability in assessing English language proficiency. Purpose: This study aimed to investigate the reliability of the International English Language Testing System (IELTS) design as a measure of English language proficiency. Methodology: To explore this issue, the research reviewed relevant literature and employed various methods, including frequency analyses, data analytics, and machine learning algorithms based on mathematical principles. Unlike previous studies, this work analysed a dataset of 33,505 authentic band scores, allowing for a comprehensive examination of IELTS test score patterns. Findings: The findings revealed that the concept of reliability is complex, and the statistical methods used by IELTS test providers, based on classical test theory, may provide approximate rather than precise scores. Relying solely on the reputation and authority of the IELTS test is insufficient to establish its reliability, as it involves logical fallacies and arguments from authority. Research implications: It is necessary to conduct further independent research to ensure the IELTS test's reliability and address the concerns expressed by researchers and test-takers.