EVALUATING THE PREDICTIVE EFFECTIVENESS OF AI-GENERATIVE FORMATIVE ASSESSMENT ON STUDENTS’ ACHIEVEMENT IN EDUCATIONAL STATISTICS.

Abstract

This study examined the predictive power of Artificial Intelligence (AI)-generated formative assessments on students’ academic achievement in Statistics among NCE students at the Federal College of Education (Technical), Isu, Ebonyi State. The study combined theoretical and empirical approaches to explore how AIbased formative assessment tools influence and forecast students’ learning outcomes. Theoretically, the study was grounded in Constructivist Learning Theory and the Learning Analytics Framework, highlighting the role of adaptive feedback in promoting learning. A correlational and quasi-experimental design was employed. The population consisted of 230 NCE students offering Statistics, out of which a stratified random sample of 100 students participated. Two instruments were used: the AI-Generated Formative Assessment Test (AIFAT) and the Statistics Achievement Test (SAT). Data were analyzed using Pearson correlation, linear regression, and independent t-tests. Findings revealed a strong Reliability analysis using Cronbach’s Alpha produced coefficients of 0.86 for AIFAT and 0.79 for SAT, indicating high internal consistency and positive correlation (r = 0.73, p < 0.05) between AI-generated formative scores and academic achievement. Regression analysis indicated that AI-generated assessments predicted 54% of the variance in students’ achievement (R² = 0.54).. The study concludes that AI-based formative assessment systems enhance prediction accuracy, personalized learning, and academic performance. It recommends the integration of AI-driven formative tools in teacher education to improve assessment efficiency and student success in Statistics.