EVALUATING TEST-TAKER ABILITIES: INSIGHTS FROM NATIONAL CERTIFICATE ASSESSMENTS

Abstract

This paper explores innovative methodologies for assessing test-taker abilities, with a specific focus on the application of the Rasch model in national certification processes. Our organization is committed to ensuring fair, reliable, and valid assessment practices that accurately reflect the diverse competencies of test-takers. By leveraging the Rasch model, we enhance the precision of ability estimations, providing nuanced insights into individual performance that go beyond the capabilities of traditional classical best theory. The study presents practical examples from pilot testing in biology for 9th-grade students, demonstrating the model’s effectiveness in addressing key assessment challenges, including item bias, score interpretation, and measurement precision. Additionally, the paper discusses how modern test theories complement Rasch-based analyses, contributing to improved test design, calibration, and fairness. The findings highlight the importance of continuously refining test items and conducting robust statistical evaluations to maintain fairness in assessments across diverse educational contexts. The analysis includes classical test theory metrics such as distractor analysis, item-total correlation, item difficulty, and Cronbach’s alpha for four test forms, interconnected by six common items which allow for meaningful comparison of scores across test-takers who have answered different sets of questions. Furthermore, the applications of the Rasch model are demonstrated through the evaluation of item fit to the model, the construction of a Wright Map, and the analysis of test characteristic curves, all of which support evidence-based decision-making in national assessment systems.