Conceptualizing a Nonlinear Meta-Analysis in Education Research

• In Event: When Sharp Statistics Meet Complex Data

Abstract

Within the area of education research, many critically important instructional effects appear to unfold nonlinearly over time. Despite the growing methodological literature on nonlinear growth methodologies for use with original data, the meta-analytic literature—that describes methods for modeling effect sizes across many published studies—has not yet incorporated nonlinearity in a meaningful way. Here, we parameterize a nonlinear mixed effects model for that purpose, and demonstrate the first nonlinear meta-analysis in education research on a dataset of effect sizes from special education reading intervention studies. We show that the nonlinear meta-analysis model estimated substantively relevant parameters that would not have been available with any other existing method, and highlight future avenues of methodological research and substantive application.

Authors

• Denis Dumas, University of Denver

• Daniel McNeish, Arizona State University

• Garrett J Roberts, University of Denver