Contextualized Mathematics Self-Efficacy and Mathematics Self-Concept: A Multilevel Structural Equation Modeling Moderated Mediation Analysis

Abstract

Contextualized relationships between mathematics self-efficacy and mathematics self-concept is investigated in their effects on mathematics achievement. Drawing data from Trends in International Mathematics and Science Study 2011 (TIMSS) administered to eighth graders in the United-States, multilevel structural equation modeling (MSEM) moderated mediations are implemented; indirect and contextual effects are evaluated, as well as their effect size. School level moderation qualifies the mediations, thanks to a school-SES composite. In order to do so, several reliability indices (α, ω, H) are compared and adjusted to the clustered nature of the data while non-linear principal component analysis (NLPCA) is conducted to obtain predictors’ factor scores. The Johnson-Neyman technique allows for insightful visualizations of the proposed moderation. Main results put forward an original perspective on the big-fish-little-pond effect where it is shown that a larger pond might not be so bad after all.