Abstract
Prior research has revealed robust and consistent relations between spatial and mathematical skills. Yet, establishing a causal relation has been met with mixed effects. To better understand whether, to what extent, and under what conditions mathematics performance can be improved through spatial training, we conducted a systematic meta-analysis of the extant literature. Our analysis included 29 studies that used controlled pre-post study designs to test the effects of spatial training on mathematics (N = 3,765; k = 89). The average effect size (Hedges's g) of training relative to control conditions was .28 (SE = .07). Critically, there was also evidence that spatial training improved individuals' spatial thinking (g = .49, SE = .09). Follow-up analyses revealed that age, use of concrete manipulatives, and type of transfer ("near" vs. "far") moderated the effects of spatial training on mathematics. As the age of participants increased from 3 to 20 years, the effects of spatial training also increased in size. Spatial training paradigms that used concrete materials (e.g., manipulatives) were more effective than those that did not (e.g., computerized training). Larger transfer effects were observed for mathematics outcomes more closely aligned to the spatial training delivered compared to outcomes more distally related. None of the other variables examined (training dosage, spatial gains, posttest timing, type of control group, experimental design, publication status) moderated the effects. Additionally, analyses of publication bias and selective outcome reporting were nonsignificant. Overall, our results support prior research and theoretical claims that spatial training is an effective means for enhancing mathematical understanding and performance. However, our meta-analysis also highlights a poor understanding of the mechanisms that support transfer. To fully realize the potential benefits of spatial training on mathematics achievement, more theoretically guided studies are needed.