Math Data Collaborative
2019 to 2020

Despite overall increases in mathematics scores in recent years, achievement gaps related to social class and race persist and have been largely unchanged over the past 15–20 years. These gaps have been particularly persistent in the middle-grade years: While math achievement gaps among the nation’s 4th-graders have narrowed somewhat over this time, achievement gaps evident in 8th grade have not narrowed. A central hypothesis, supported by empirical evidence, is that students who struggle in the middle grades need help specifically with the transition from arithmetic to abstract mathematics. To build adequate grasp of the critical pre-algebra concepts—rational number, ratio, and proportion concepts—students need opportunities to engage in skills and habits of thinking that mathematicians use in their work: puzzling and persevering, seeking and using structure, using tools strategically, describing repeated reasoning, and communicating with precision.

The past decade has seen tremendous investments being made in curricula and pedagogies aimed at helping middle-grades students develop these skills in order to close the achievement gap, with little evidence of widespread impact. There is a clear need for interventions that can support middle-school students who struggle with mathematics.

Researchers from Education Development Center are conducting a two-year pilot study of a Middle Grades Math Data Collaborative,funded by Schmidt Futures. The Collaborative will engage a postdoctoral data scientist who will work with research staff at EDC and at NWEA in mining “big data” consisting of national student interim assessment data, with the goal of understanding the growth patterns of specific mathematics skills for a wide national cohort of struggling middle-grades students. Researchers will codify growth patterns for classrooms or schools in mathematics learning, and will develop metrics, models, and tools required for rapid improvement of student outcomes.

This project seeks to

 1. identify key mathematics skills and concepts that underlie students’ abilities to transition to pre-algebra, and concepts that are typical stumbling blocks;

2. develop profiles of performance variance among disadvantaged students in geographically and demographically varied school and districts; and

3. serve as a proof of concept for how large-scale interim assessment data and machine learning can be used as infrastructure to identify promising interventions and practices to improve student learning.

The ultimate goal is to bring new, actionable insights into how educational systems can more effectively meet the needs of struggling mathematics learners.