Criteria for Analyzing a Test Measuring Learning Progress

Evaluating learning progress is a vital element of educational interventions for students with learning disabilities. Measuring change imposes considerably different requirements on test construction compared to traditional psychometric diagnostic instruments. The present paper discusses four theoretical challenges for test construction, namely a high reliability, unidimensionality, constant item difficulty, and high test fairness. A procedure is proposed for analyzing tests to fit these four criteria making use of item analyses, confirmatory factor analyses, Rasch modeling, and analyses of differential item functioning. The suggested procedure is exemplified on a newly developed test for measuring language proficiency of students with learning difficulties on the basis of c-tests. The results disclose c-tests as highly suitable for measuring differences in general language development.