Building an adaptive test model for English reading comprehension in the context of online education

With the ongoing progress of economic globalization, international communication has witnessed a notable surge. Proficiency in English has become indispensable for individuals seeking to comprehend international information and engage in economic, cultural, and scientific exchanges. At the heart of English instruction lies English reading, and using models to test scientific English reading comprehension abilities can significantly enhance the quality and effectiveness of learning. This study aims to augment the quality of adaptive English reading comprehension tests and accurately gage the assessment of college students' English reading abilities. In this research endeavor, the difficulty of objective questions in English reading comprehension was quantified by amalgamating deviance entropy and fuzzy hierarchical analysis. Subsequently, a hybrid test model grounded in Rasch was devised with multiple adaptive algorithms. The test results unveiled that the weighting values of the objective question difficulty model, based on fuzzy hierarchical analysis, changed, ranging from 0.551 to 0.562 for content and from 0.449 to 0.438 for question type. The mean ability values of the 40 students spanned from − 3 to + 3 logits, with the dwell time exhibiting a distribution pattern. Compared with the traditional adaptive test model, the hybrid adaptive test model employed 22.25 questions, resulting in an overlap rate of 0.21. The time taken to assemble the paper was a mere 0.05 s, accompanied by a reduction in difficulty by 0.32, leading to an impressive test efficiency of 98%. The findings of this study underscore that the adaptive model developed herein offers superior ability assessment and effectively enhances students' English reading comprehension learning.