Do “better” teachers and classroom resources improve student achievement? A causal comparative approach in Kenya, South Africa, and Swaziland

We use the 2007 SACMEQ data to make traditional “upwardly biased” estimates of teacher and classroom resource correlates of 6th grade student achievement in Swaziland, Kenya, and South Africa using an OLS model, and a “less biased causal” approach using a student fixed effects model. Our fixed effects model exploits the fact that most students in all three countries have different teachers for reading and mathematics. Each student is therefore subject to the “treatment” of different teacher characteristics and classroom resources, yielding a relatively unbiased but rather “stringent” estimate of teacher and classroom effects. Our results suggest that: (a) several important identifiable teacher characteristics and classroom resources affect student achievement in each country; that (b) those characteristics and resources may differ from one national context to another, between male and female students, and across socioeconomic groups of students; and that (c) the “upwardly biased” results generally differ from the “less biased causal” results. We discuss and attempt to explain these differences.