ASSESSMENT OF INSTITUTIONAL DATA HETEROGENEITY FOR ARTIFICIAL INTELLIGENCE SYSTEMS
https://doi.org/10.53360/2788-7995-2026-3(23)-23
Abstract
This study quantifies institutional heterogeneity in digital engagement and academic outcomes using anonymized data from Kazakhstani schools, comprising records from two EdTech platforms (Twig-bilim.kz, BilimLand.kz) and entry diagnostic test scores. Applying one-way ANOVA/intraclass correlation (ICC) decomposition and random-intercept mixed-effects modeling, we find that school membership explains 8.8-9.8% of variance in digital platform engagement and 30.0-30.7% of variance in entry test scores, with close numerical agreement between the ANOVA-based variance shares and the mixed-effects ICC estimates. These estimates indicate non-negligible school-level heterogeneity, particularly for entry diagnostic-test scores. The findings indicate that pooled, school-agnostic machine learning models risk systematically miscalibrated predictions for students in schools that deviate from the national average, and argue for schoolaware calibration in the design of AI-based early-warning and adaptive-learning systems intended to support flexible and equitable access to education. The results are discussed in relation to Kazakhstan's ongoing school digitalization initiatives and to the broader literature on algorithmic fairness in education.
Keywords
About the Authors
К. E. IklassovaKazakhstan
Kainizhamal Iklassova – PhD, associate professor, Department of Information and Communication Technologies
150000, Republic of Kazakhstan, Petropavllovsk, Pushkina Str, 86
L. N. Abdullah
Malaysia
Lili Nurliyana Abdullah – PhD, Associate Professor, Department of Mulitimedia, Faculty of Computer Science and Information Technology
Malaysia, Serang, Selangor, 43400 Seri Kembangan
М. Zh. Bazarova
Kazakhstan
Madina Zhomartovna Bazarova – PhD, associate professor of the Department of Computer Modeling and Information Technology
070002, Republic of Kazakhstan, Ust-Kamenogorsk, 30th Gvardeiskoy Divisii Str, 34
References
1. Khalifeh F. Redefining personalized learning in the artificial intelligence era: An updated systematic review from 2019 to 2025 / F. Khalifeh, R. Santiago, R. Palau // Smart Learning Environments. – 2026. – Vol. 13. – Art. 19. https://doi.org/10.1186/s40561-026-00440-6.
2. The impact of the digital divide on synchronous online teaching in Kazakhstan during COVID-19 school closures / A. Amirova et al // Frontiers in Education. – 2023. – Vol. 7. – Art. 1083651. https://doi.org/10.3389/feduc.2022.1083651.
3. How artificial intelligence transforms education in Kazakhstan // The Astana Times. – 2026. – 7 January. https://astanatimes.com/2026/01/how-artificial-intelligence-transforms-education-inkazakhstan/ (date of access: 02.09.2026).
4. Raudenbush S.W. Hierarchical Linear Models: Applications and Data Analysis Methods / S.W. Raudenbush, A.S. Bryk. – 2nd ed. – Thousand Oaks, CA: Sage Publications, 2002.
5. From school AI readiness to student AI literacy: A national multilevel mediation analysis of institutional capacity and teacher capability / X. Guan et al // arXiv. – 2026. – arXiv:2603.20056. – https://doi.org/10.48550/arXiv.2603.20056.
6. To N.P. Fairness for machine learning software in education: A systematic mapping study / N.P. To, P.N. Hung, A. NguyenDuc // SSRN. – 2024. https://doi.org/10.2139/ssrn.4713827.
7. Idowu J.A. Investigating algorithmic bias in student progress monitoring / J.A. Idowu, A.S. Koshiyama, P. Treleaven // Computers and Education: Artificial Intelligence. – 2024. – Vol. 7. – Art. 100267. https://doi.org/10.1016/j.caeai.2024.100267.
8. El Jihaoui M. Towards a fair and transparent early warning system for identifying students at risk of academic failure / M. El Jihaoui, O.E.K. Abra, K. Mansouri // Frontiers in Education. – 2026. – Vol. 11. – Art. 1878686. https://doi.org/10.3389/feduc.2026.1878686.
9. Adaptive learning mechanisms for learning management systems: A scoping review and practical considerations / S. Kucharski et al // arXiv. – 2025. – arXiv:2512.18383.
10. The role of digital technologies in higher education institutions: The case of Kazakhstan / D. Nurtayeva et al // Problems and Perspectives in Management. – 2024. – Vol. 22, № 1. – P. 562-577. https://doi.org/10.21511/ppm.22(1).2024.45.
11. COVID-19 online learning challenges: Kazakhstan secondary schools case study / S. Nurgaliyeva et al // Frontiers in Education. – 2024. – Vol. 9. – Art. 1448594. https://doi.org/10.3389/feduc.2024.1448594.
12. Pokidaev D. Kazakhstan to launch digital schools project / D. Pokidaev // The Times of Central Asia. – 2026. – 2 April. https://timesca.com/kazakhstan-to-launch-digital-schools-project/ (date of access: 02.09.2026).
13. McKinney W. Data structures for statistical computing in Python / W. McKinney // Proceedings of the 9th Python in Science Conference (SciPy 2010). – 2010. – P. 56-61. https://doi.org/10.25080/Majora-92bf1922-00a.
14. Seabold S. Statsmodels: Econometric and statistical modeling with Python / S. Seabold, J. Perktold // Proceedings of the 9th Python in Science Conference (SciPy 2010). – 2010. – P. 92-96. https://doi.org/10.25080/Majora-92bf1922-011.
15. Hedges L.V. Intraclass correlation values for planning group-randomized trials in education / L.V. Hedges, E.C. Hedberg // Educational Evaluation and Policy Analysis. – 2007. – Vol. 29, № 1. – P. 60-87. https://doi.org/10.3102/0162373707299706.
Review
For citations:
Iklassova К.E., Abdullah L.N., Bazarova М.Zh. ASSESSMENT OF INSTITUTIONAL DATA HETEROGENEITY FOR ARTIFICIAL INTELLIGENCE SYSTEMS. Bulletin of Shakarim University. Technical Sciences. 2026;(3(23)):227-235. https://doi.org/10.53360/2788-7995-2026-3(23)-23
JATS XML















