Validating Reading Comprehension Assessment Under the GDINA Model

Abstract

Cognitive diagnosis models (CDMs) are latent variable models mainly developed to assess students’ specific strengths and weaknesses in a set of skills or attributes within a particular domain. In this study, the reading comprehension assessment is diagnostically designed, constructed, and developed from the very first step. The predetermined attributes or sub-skills are explicitly defined in the construction phase as they should align with the instructional goals. Using R package CDM, the Generalized-DINA model (GDINA) was applied to the reading comprehension assessment. A total of 900 Year 4 primary students from the Eastern District of Pulau Pinang national and vernacular schools sat for this assessment. Through the cognitive analysis, the study is expected to provide detailed diagnostic feedback on students’ strengths and weaknesses in the underlying skills identified in the reading comprehension assessment. Such detailed information can help teachers in classroom teaching, designing remedial courses, and developing material according to the student's needs.