Assessing the association between parents’ math talk and children’s math performance: A preregistered meta-analysis

Abstract

The home math environment has gained considerable attention as a potential cause of variation in children’s math performance, and recent research has suggested positive associations between parents’ math talk and children’s mathematical performance. However, the extent to which associations reflect robust causal effects is difficult to test. In a preregistered meta-analysis, we assess the association between parents’ math talk and children’s math performance. Our initial search identified 24,291 potential articles. After screening, we identified 22 studies that were included in analyses (k = 280 effect sizes, n = 35,917 participants). A multilevel random effects meta-analysis was employed, finding that parents’ math talk is significantly associated with children’s math performance (b = 0.10, SE = 0.03, p = .002). We tested whether associations differ as a function of sample characteristics, observation context, observation length, type of math talk and math performance measured, and modeling approaches to math talk variable analysis. In addition, we tested whether associations are robust to the inclusion of strong baseline covariates and found that effects attenuated when children’s domain-general and/or prior math abilities are included. We discuss plausible bounds of the effects of parents’ math talk on children’s mathematical performance to inform power analyses and experimental work on the impact of parents’ math language on children’s math learning.