An “Embedded” Measure of Working Memory and Mathematics Achievement in Children Referred for Evaluation of Learning Disability

William Lacey, PhD

Boston Children’s Hospital – Fellow
Lacey_William poster

Scientific Abstract

Background: Working memory, as measured by standard neuropsychological tests such as Digit Span, has been shown repeatedly to be a domain general predictor of mathematics achievement in children and adults. We evaluated a novel measure of working memory embedded in children’s actual mathematics performance as a predictor of their mathematics achievement along with standard neuropsychological observational and test measures of working memory.

Aims: Aims of the present study were (1) To evaluate associations of the working memory descriptor from the Mathematics Diagnostic and Prescriptive Inventory (MDPI-WM) with standard behavioral and test measures of WM and (2) To evaluate MDPI-WM as a predictor of mathematics achievement along with these other measures.

Methods: Data from 82 third to sixth-grade students evaluated in the BCH Learning Disability Program (January 2019 to October 2020) were abstracted from clinical records. WASI-II Full Scale IQ range: 78 to 123.

Working memory measures included the MDPI-WM, a behavioral observation measure (BRIEF Working Memory Parent, BRIEF-WM), and a standard test measure (CMS Numbers, CMS-WM).  The MDPI is a structured mathematics interview that yields quantitative and qualitative features of a Mathematics Learning Profile, including a working memory descriptor based on observations of the child’s behavior while performing mathematical tasks (e.g., need to write down partial sums/differences when performing mental  arithmetic). Scored by an expert system, the MDPI also  generates an overall grade level estimate, indicating the extent to which the child’s achievement is above or below grade placement (MDPI-Ach).

Correlations among all variables were calculated. Multiple regression analyses were conducted with MDPI-Ach the dependent variable and Age, Grade, BRIEF-WM, CMS-WM, and MDPI-WM the independent variables.

Results: There were no significant correlations among the three working memory measures. In the regression model, Age, Grade, and CMS-WM (all p<0.01) and MDPI-WM (p<0.03) were significant predictors but BRIEF-WM was not (p>0.52). The significant predictors were retained in a further model as follows: Age, Grade, and CMS-WM (p<0.01) and MDPI-WM (p<0.04).

Conclusions: Although standardized neuropsychological tests of working memory reliably predict mathematics achievement, they do not fully capture the implementation of working memory skills in the context of actual mathematical performance. The MDPI-WM can provide important qualitative information regarding the role of working memory in mathematics learning disorders not captured by standard neuropsychological tests, with implications for prescriptive intervention.

research Areas


William H Lacey, PhD, Deborah P Waber, PhD, Ellen C Boiselle, PhD, Andrea Ortiz Jimenez

Principal Investigator

Deborah P. Waber, PhD