Examine the Longitudinal Association Between Prior and Subsequent Mathematics Using Meta-Analytic Structural Equation Modeling Approach

The current meta-analysis investigates the longitudinal association between prior and subsequent mathematics performance, involving mathematics measured at three time points, and to identify potential factors that could moderate this association, including age, time lag, and types of mathematics. Our analysis included 105 studies, comprising 111 independent samples and involving 413,541 individual students. Our findings revealed that early mathematics knowledge at Time 1 was associated with later performance at Time 3, both directly and indirectly through intermediate performance at Time 2, after controlling for working memory and language comprehension. Notably, the stability of the association between Time 1 and Time 3 mathematics remained robust across increasing age and time lag, while the association between Time 1 and Time 2 weakened as the time lag grew. Additionally, stronger associations were observed when mathematics at Time 1 and Time 3 involved similar or more comprehensive tasks. These findings suggest snowballing effect is more likely to occur within similar types of mathematics tasks, whereas a steppingstone effect is more likely between prior foundational mathematics and later higher level mathematical concepts.