Test Format and Local Dependence of Items Revisited: A Case of Two Vocabulary Levels Tests

Abstract

Local item dependence (LID) is one of the most critical assumption in the Rasch model when it comes to the validity of a test. As the field of vocabulary assessment is calling for more clarity and validity for vocabulary tests, such assumption becomes more important than ever. The article offers a Rasch-based investigation into the issue of LID with the focus on the two popular formats of Vocabulary Levels Tests (VLT): multiple-choice and matching. A Listening Vocabulary Levels Test (LVLT) and an Updated Vocabulary Levels Test (UVLT) were given to a single cohort of 311 university students in an English as a Foreign Language (EFL) context. The analyses of raw score and standardized residuals correlations were conducted. The findings found no relationship between the 4-option, multiple-choice format of the LVLT and item local dependence. However, results from score and standardized residuals correlations analyses showed a strong link between the 3-item-per-cluster, matching format and item local dependence. The study calls for greater attention to the format of future vocabulary tests and support the use of meaning-recall formats in vocabulary testing.