Persistently weak learning outcomes in many basic-education systems have been attributed jointly to inadequate teacher quality and chronic under-resourcing of schools, yet the independent and combined contributions of these two factors to downstream skills deficits were rarely quantified within an analytic framework that respected the nested structure of educational data. This study examined the extent to which teacher quality and school resourcing predicted literacy-numeracy skills deficits among primary school learners, using a multilevel modelling approach that accounted for the clustering of students within schools. A cross-sectional dataset comprising 1,451 students nested within 60 primary schools was analysed. Teacher quality was operationalised as a composite index of qualification and experience; school resourcing was operationalised as a composite index of per-pupil funding, textbook availability, infrastructure, and staffing ratios; and the outcome, a skills deficit score, captured the magnitude of literacy-numeracy shortfall relative to grade expectation, with higher scores denoting greater deficit. Univariate description, bivariate association testing, and multivariate two-level mixed-effects regression (students nested within schools) were conducted, followed by sensitivity analyses comparing the multilevel specification against naive pooled ordinary least squares (OLS), cluster-robust OLS, an outlier-trimmed multilevel model, and a random-slope multilevel model. The null model intraclass correlation coefficient was 0.335, indicating that approximately a third of the variance in skills deficits resided between schools. In the adjusted multilevel model, teacher quality (b = -0.281, SE = 0.015, p < .001) and school resourcing (b = -0.525, SE = 0.202, p = .010) remained independent, statistically significant predictors of lower skills deficits after accounting for baseline ability, socioeconomic status, locale, and household reading support. Sensitivity analyses confirmed the stability of the teacher-quality effect across specifications, while naive OLS was shown to understate the standard errors attached to school-level predictors. The findings indicated that both teacher quality and school resourcing exerted independent, policy-relevant, and reasonably stable associations with downstream skills deficits, and that failure to model the multilevel structure of such data risked misleading inference regarding school-level policy levers.