Working memory deficits in Math learning difficulties: a meta-analysis

In the present study, we made a quantitative synthesis of the literature comparing children with Math learning difficulties to average-achieving, age-matched children on measures of working memory in view of Baddeley and Hitch's multicomponential working memory model. Our meta-analytic investigation drew from the literature a number of 18 studies that matched the inclusion criteria. We hypothesized that all three components of Baddeley and Hitch's model should contribute to the poor Math performance displayed by children with Math learning difficulties, and that this relation varies as a function of age. Also, we hypothesized that children with Math learning difficulties would present more accentuated working memory deficits for numerical material and/or processing. Based upon Cohen's criteria, the results indicate a large effect size, in favor of the controls for the relationship between Math performance and central executive component, as well as Math performance and the visual-spatial sketchpad, more accentuated in younger ages. Only a moderate effect size emerged in the case of Math performance and the phonological loop (PL). With respect to numerical versus non-numerical working memory deficits, a large effect size emerged for numerical central executive measures, along with a moderate effect size for the PL numerical measures, and a weak effect size for non-numerical PL. The results indicate that Math difficulties are attributable to a central executive deficit or delay, more accentuated for numerical stimuli and/or processing and to a visual-spatial working memory deficit.