PSYCHOMETRIC VERIFICATION OF THE AUTOMATED TEST ITEM MANAGEMENT SYSTEM

Abstract

This article examines the results of a comprehensive modernization of the automated test item management system used in national examination procedures, including the Unified National Test and the Comprehensive Test for master’s degree admission. The main focus is on the impact of implementing technical and organizational solutions on the psychometric characteristics of tests and the overall improvement of educational measurement quality. The study analyzed measures such as the introduction of caching mechanisms, upgrading computing infrastructure, applying database sharding technology, integrating a DLP system to ensure information security, as well as using the automated item quality control tool CheckerWord. To evaluate the effectiveness of the implemented changes, a psychometric analysis was conducted using the Rasch model and Winsteps software (version 5.8.3.0). The results showed that modernization had a positive effect on test reliability, participant differentiation, and item fit to the model. Particular attention was given to the protection of personal data and the stable functioning of the database. The findings confirm that a comprehensive approach combining both technical and psychometric measures contributes to ensuring the validity, reliability, and reproducibility of educational assessments within the national quality assurance system.