A linear logistic test model (LLTM) application in foreign language testing

This study applies the Linear Logistic Test Model (LLTM) to the reading comprehension subtest of an English certification exam developed by the Foreign Language Assessment Program (PELEx, for its acronym in Spanish). A Q-matrix operationalized cognitive and linguistic predictors, such as paraphrasing and inferential reasoning, to explain item difficulty. Delta squared results showed that the Q-matrix accounted for 77% of the variance in item difficulty, with "Inferences" and "Subtask Complexity" being key contributors. While overlaps in item difficulty coefficients reflected the nested nature of the Common European Framework of Reference for languages (CEFR) levels, this progression aligns with the framework's principles. The findings show that LLTM can make item difficulty more interpretable in reading assessment while also helping identify which item features merit further refinement in future test development.