Refining and Validating Paul Nation’sVocabulary Size Test for TOEFL Candidates inPakistan: An Item Discrimination and Predictive Validity Studyvocabulary size test

Abstract

The purpose of this study is to refine and validate Paul Nation's Vocabulary Size Test (VST) to better meet the vocabulary assessment needs of TOEFL candidates. There is also a lack of research regarding the predictive validity of the VST in the context of TOEFL exams. The primary objectives of this study were to (1) conduct item discrimination analysis to evaluate how effectively individual items of the VST distinguish among students with different levels of vocabulary proficiency, (2) assess the internal consistency reliability of the refined test, and (3) examine the predictive validity of the test in relation to students’ performance in the TOEFL exam. The study employed a combination of purposive and convenience sampling and involved 336 TOEFL candidates from various institutions in Punjab (Pakistan). The original VST developed by Nation and Beglar (2007) was administered, and item analysis was carried out using facility and discrimination indices. Items with facility values between 0.30 and 0.70 and discrimination values of 0.40 or above were retained, resulting in a refined 58-item version of the test. Reliability analysis yielded a Cronbach’s alpha of α = 0.94, indicating high internal consistency. To evaluate predictive validity, a simple linear regression was conducted using scores from the refined VST and TOEFL exam results obtained from a subsample of 30 TOEFL candidates. The results showed a strong, statistically significant relationship (Model 2: R² = .90, Adjusted R² = .90, p < .001). These findings have important implications for test developers, educators, and researchers in Pakistan, offering a more reliable vocabulary assessment tool with direct relevance to TOEFL preparation and predictive modelling of  TOEFL performance