ABSTRACT
School-based improvement programs represent a core strategy in improving education because they can leverage pre-existing social and organizational structures to promote coordinated and comprehensive change across multiple facets of schooling. School-based programs are generally designed to be implemented by intact schools/districts, frequently making it infeasible or atheoretical to assign students within the same school to different conditions while ensuring study validity. Rather, studies frequently assign intact schools/districts to treatment conditions to accommodate the multilevel structure of schooling and the theory of action underlying many school-based programs. In planning such studies, effective and efficient design requires plausible values of the variance partition coefficients and the variance explained by covariates during the design stage. Using representative samples of each country, we develop empirical estimates of design parameters within and across 15 countries that are intended to inform and facilitate the efficient design of multisite cluster-randomized studies of school improvement in sub-Saharan Africa.
Notes
1. R2 values are not well defined in multilevel models because variance can increase with the introduction of covariates. For this reason, our approach follows the literature by using pseudo-R2 values to describe the reduction in variation (Hedges & Hedberg, Citation2007; Jacob et al., Citation2010; Spybrook & Raudenbush, Citation2009).
Additional information
Notes on contributors
Ben Kelcey
Ben Kelcey is an assistant professor in the College of Education, Criminal Justice, & Human Services at the University of Cincinnati. His research interests include the development of measurement and quantitative research methods to understand effective teaching and schooling.
Zuchao Shen
Zuchao Shen is a doctoral student in the Quantitative Methods in Education Program at the University of Cincinnati. His research interests include school effectiveness research methods and policy.