Longitudinal Meta-Analyses of Retest Effects and Mean-Level Development of School Achievement as Assessed by Standardized Tests

Standardized school achievement testing is an important tool for monitoring learning success, evaluating educational programs, and informing educational policy. However, robust evidence for typical academic performance trajectories across the school years, which could inform educational interventions and monitoring programs, remains lacking as no meta-analysis has yet synthesized the available longitudinal findings on the mean-level changes in standardized school achievement tests. Additionally, the retest effect, an important consideration when interpreting the results of repeated testing, has barely been investigated in the context of school achievement testing. This article presents two meta-analyses of longitudinal studies that address these gaps. Based on an extensive literature search for peer-reviewed journal articles and dissertations in PsycINFO and ERIC, we identified 75 records (e = 569, total n = 223,964) for Meta-analysis 1 on the retest effect, and 224 records (e = 3,022, total n = 767,692) for Meta-analysis 2 on the mean-level changes. Meta-analysis 1 revealed a very small and short-lived retest effect (Δ = .005, [95% CI -.035, .045]), indicating limited practical implications, particularly for longer test-retest intervals. Meta-analysis 2 found a medium-sized yearly mean-level gain (Δ = 0.513, 95% CI [0.434, 0.593]). Yearly gains varied significantly by grade level, with much larger gains in earlier grades compared to later grades. Additionally, trajectories varied by school stage and achievement domain. Thus, rather than applying general conventions for educationally significant effect sizes, the effects of interventions should be interpreted relative to the expected yearly gains in the respective school year and domain. The study was funded by a research grant from [blinded]. OSF registration: https://osf.io/687xj?view_only=5a89536e3c954e1593629c547998340e.