Regional inequities in school attendance in Somalia: evidence from spatial statistics and multilevel modeling

School attendance in Somalia remains critically low and is marked by significant regional disparities, hindering national development and perpetuating cycles of poverty. This study investigates the individual, household, and geographical determinants of school attendance and maps its spatial patterns across Somalia to inform targeted interventions. Utilizing data from the 2023 Somalia Integrated Household Budget Survey (SIHBS) covering 13,329 children, this research employed multilevel logistic regression and spatial statistical techniques (Global Moran’s I, Local Moran’s I, and Getis-Ord Gi*). Results revealed profound geographical clustering of school non-attendance (Global Moran’s I = 0.597, p < .001). Significant hot spots of non-attendance were identified in southern and southwestern regions, including Bakool (76.6% non-attendance) and Shabeellaha Dhexe (87.3% non-attendance), while cold spots with better attendance were found in areas like Woqooyi Galbeed (43.1% non-attendance) and Sool (42.9% non-attendance). Multilevel models indicated that older age (attendance drops from 93.9% for 5–10 year-olds to 35.9% for 21–25 year-olds), being female (β = -0.307, p < .001), qat consumption (β = -3.422, p < .001), and working for pay (β = -1.955, p < .001) were significantly associated with lower school attendance. Conversely, literacy (β = 5.816, p < .001) was a strong positive predictor. A substantial portion of the variance in attendance was attributable to household-level differences (ICC = 52.5% in the full model). The findings underscore that school attendance in Somalia is shaped by a complex interplay of multi-level factors and exhibits stark spatial inequities. This evidence highlights the necessity for geographically targeted, context-specific policies that address both individual/household vulnerabilities and broader structural barriers to improve educational access and equity across Somalia.