Bridging assessment and learning: a cognitive diagnostic analysis of a large-scale Spanish proficiency test

Abstract

The literature has highlighted the importance of Cognitive Diagnostic Assessment (CDA) of large-scale, high-stakes language tests to assess the individualized strengths and weaknesses of every learner. However, there are relatively few studies on the accuracy and applicability of feedback information in follow-up teaching and learning practices. Using both the diagnostic results of 1933 test takers derived from the Generalized Deterministic Inputs, Noisy and Gate (G-DINA) model of a national Spanish test (EEE) and qualitative data from the test takers’ literature reviews in their bachelor thesis drafts, the precision of the diagnostic feedback was examined to verify its usefulness for the improvement of academic reading. The results showed that the G-DINA model had an appropriate model fit to the test performance data, and that the CDA is able to identify the specific skill profiles of each test taker, which were not always consistent with the scores provided by classical test analysis. The triangulation of the diagnostic reports and the literature reviews from the learners’ thesis drafts, clearly showed that CDA used for a large-scale test can assess reading skills accurately and the feedback is valuable for improving the future academic reading and thesis revision.