Empirical Research on Improving of Validity and Reliability in Educational Assessment Using Cognitive Intelligence Models

Abstract

Purpose: The primary objective of this study was to explore the application of cognitive intelligence models in educational assessment, focusing on how these models can improve assessment validity, reliability, and feedback effectiveness. The research aimed to evaluate whether AI-powered assessments could better capture student performance, provide personalized feedback, and contribute to more accurate educational evaluations. Methodology: An empirical research design was employed to investigate the impact of cognitive intelligence models on educational assessments. Data were collected from a sample of students using both AI-powered assessments and traditional paper-based exams. The study involved comparing key performance indicators such as assessment validity, reliability, and feedback effectiveness. Statistical analyses, including Cronbach’s alpha for reliability and t-tests for performance comparison, were used to assess the significance of the results. Results: The findings indicated a significant improvement in the validity and reliability of AI-powered assessments. The experimental group using AI assessments scored 85.6% on average, while the control group using traditional exams scored 78.2%, with the difference being statistically significant (p < 0.01). AI assessments demonstrated higher reliability (Cronbach’s alpha = 0.92) compared to traditional exams (Cronbach’s alpha = 0.81). Personalized feedback from the AI system was highly rated by students, with 85% reporting it to be actionable and beneficial for improving learning outcomes. Additionally, AI assessments proved effective in predicting future student performance with an accuracy of 87%. Conclusion: The results demonstrate that cognitive intelligence models significantly enhance the quality of educational assessments, providing more valid, reliable, and personalized evaluations. These models offer advantages over traditional assessment methods by adapting to individual student needs and providing real-time, actionable feedback. Despite the potential benefits, challenges related to data privacy, algorithmic bias, and the role of human educators must be addressed to ensure the ethical and effective implementation of AI in educational settings.