Multilevel Modelling of Educational Achievement in Primary Schools Across the Six Geopolitical Zones of Nigeria

This study applies multilevel statistical modelling to analyse educational achievement in primary schools across all six geopolitical zones of Nigeria, using a hierarchical data structure in which pupils are nested within schools and schools are nested within local government areas, enabling the simultaneous estimation of individual, school-level, and geographic effects on learning outcomes. Existing analyses of primary education quality in Nigeria have predominantly relied on single-level regression methods that fail to account for the non-independence of observations within schools, potentially producing misleading estimates of school and contextual effects. Pupil-level literacy and numeracy test score data from the 2022 National Assessment of Learning Outcomes conducted by the Universal Basic Education Commission are used, covering a nationally representative sample of 18,000 pupils in 900 schools across 90 local government areas. Three-level hierarchical linear models are specified and estimated, with pupil-level predictors including gender, age, attendance rate, and household socioeconomic status, school-level predictors including pupil-teacher ratio, teacher qualification, and infrastructure quality, and local government area-level predictors including per-capita government spending on education and urbanisation rate. Intraclass correlations are computed at school and local government area levels to quantify the proportion of total achievement variance attributable to each hierarchical level. Results indicate that 38 percent of variation in numeracy scores is attributable to school-level factors and a further 14 percent to local government area-level factors. Keywords: multilevel modelling, educational achievement, hierarchical linear models, primary schools, Nigeria