Developing an Original Multi-Level Modelling Framework for Educational Inequality Analysis in Nigerian States

Educational inequality in Nigeria operates at multiple levels simultaneously (individual, household, school, local government, and state) and requires multi-level modelling frameworks that can decompose variance and identify level-specific interventions, yet such frameworks remain undeveloped for the Nigerian context. This study developed an original multi-level modelling framework for educational inequality analysis in Nigerian states. A methodological framework development approach employing hierarchical data analysis was used, combining systematic review of multi-level educational modelling (64 publications from 2018 to 2024), empirical development and testing using Learning Assessment data from NECO and UBEC covering 48,000 students, 2,400 schools, and 87 LGAs across six states representing all geopolitical zones, and expert validation. Variance decomposition confirmed that 34.7 percent of test score variation was attributable to school-level factors, 18.3 percent to LGA-level factors, and 12.1 percent to state-level factors, with only 34.9 percent attributable to individual and household factors. Cross-level interaction analysis showed that the effect of household socioeconomic status on academic performance was significantly moderated by school infrastructure quality and local government education spending. The original Multi-Level Educational Inequality Framework for Nigeria (MEIF-Nigeria) specifies model structure, variable operationalisation, and interpretation guidelines for each level, including provisions for handling missing data patterns characteristic of northern state records. Expert review by 16 education data science and measurement specialists confirmed the framework's methodological originality and policy utility.

Keywords: multi-level modelling, educational inequality, Nigeria, school effectiveness, variance decomposition