The stability of students’ academic achievement in school: A meta-analysis of longitudinal studies

Abstract

Assessments of school achievement play a central role in education. For example, they are used in educational diagnostics to inform long-term placement decisions and are an important criterion in large-scale educational monitoring. Findings on the rank-order stability of students' school achievement are highly relevant to educational research and practice. While low stability undermines the validity of using academic achievement for diagnostic purposes, near-perfect stability may indicate insufficient educational mobility. However, the rank-order stability of school achievement is very rarely studied explicitly. Therefore, we know little about the conditions under which this stability is higher or lower. The present meta-analysis of longitudinal studies reports the rank-order stability of teacher-assigned school marks and school achievement tests. A total of 1990 test-retest correlation coefficients were compiled, involving 740,610 individual participants from 363 longitudinal studies. The mean rank-order stability, estimated for Grade 5 and a test-retest interval of two years, was ρ = .70 (95 % CI: .69, .72). Stability was lower for school marks (ρ = .67) than achievement tests (ρ = .72). Stability remained relatively constant across grade levels and decreased with increasing test-retest interval. The decrease in stability with increasing interval was greater for school marks than for achievement tests: Both indicators were comparably stable in very short intervals, but achievement test results were much more stable in the long-term. The stabilities of school marks and achievement tests varied between both school domains and countries.