Four Best Practices for Meta-Analysis: A Systematic Review of Methodological Rigor in Mathematics Interventions for Students With or at Risk of Disabilities

Abstract

Meta-analysis methodology has evolved with the development of more robust statistical techniques; however, few reviews in special education have focused specifically on methodological rigor in meta-analyses. In this study, we examined 29 meta-analyses of mathematics interventions published from 2000 to 2022 to determine the extent to which researchers have applied four best practices in meta-analyses focused on mathematics interventions for students with disabilities. Our findings were (a) studies used three primary moderator techniques: meta-regression ( k = 10), subgroup analysis ( k = 8), analysis of variance ( k = 3), and both subgroup analysis and meta-regression ( k = 1); (b) only one study considered small sample corrections for hypothesis tests; (c) few researchers handled the dependence between multiple effect sizes ( k = 3); and (d) the funnel plot was commonly used to detect publication bias ( k = 8). Based on our findings, we make recommendations for methodological considerations for future meta-analyses.

Meta-analysis of single-case design research: Application of multilevel modeling.

“…Furthermore, several related methodological limitations exist when synthesizing single-case data across studies. In many single-case design research, the number of subjects (e.g., three) and measurement occasions per phase (e.g., five) are smaller than the group studies or time-series analysis, resulting in inflated Type I errors (Park et al, 2023). Sampling errors of the observed outcome variables for the same study are likely correlated in meta-analysis studies (Hedges et al, 2010).…”