NMIP Summer 2026 Recap: Math Learnings from the Field
The National Math Improvement Project was genuinely national this summer. Between June and August, we met in New York, visited classrooms in Los Angeles, presented in Atlanta and Orlando, and virtually connected with leaders from all six of our districts. Here is where we went, what we published, and what our districts are working on heading into the new school year.
Defining Fluency
We opened the summer in New York, with teams from New York City Public Schools and the School District of Philadelphia and Dr. Jennifer Bay-Williams, on one of the most consequential questions in math instruction: What is math fluency, and how do students develop it?
The conversation kept returning to a false choice. Teachers report feeling forced to pick between basic fact fluency and conceptual understanding, as though one comes at the expense of the other. What made the session valuable is that it was not a presentation. The districts put draft materials in front of peers and asked what was confusing, and worked through real challenges consultancy-style together. Two weeks later, we picked up the same thread virtually, with Chicago, Houston, and Miami-Dade joining and sharing their fluency strategies, pathways redesign, and acceleration structures.
Atlanta: Math, Technology, and the Illusion of Inclusion
NMIP leaders presented during multiple sessions at the Council of the Great City Schools’ Curriculum, Research, and Instructional Leaders Conference, including Math Belongs to Everyone, on family and community engagement, and Gatekeeper to Gateway, on expanding Algebra I access.
The family engagement conversation produced the reframe that has stuck with me most: Districts often treat families as learners who need to be taught the new math. The panel argued for treating them as experts instead — on their own children, if not on the curriculum. Parents do not need to re-teach rational functions at the kitchen table. They need the right questions to ask at school, transparency about changes before those changes arrive, and a chance to see what genuine math learning looks like in the classroom.
We also convened an emerging technology panel that touched on much of the debate we’re hearing in the back-to-school news cycle:
"Screen time is not one thing." The better question is not more screens or fewer, but whether the screen time adds instructional value or could be paper and pencil with no loss.
Coherence beats guardrails. Routing every tool purchase through one team keeps schools from being pulled in five directions.
Most students using AI are not cheating. They are filling gaps where instruction did not land.
The line that got the room: if a partner is not there after the contract is signed, they are a vendor, not a partner.
CGCS added a special education track this year, where one session delivered the summer's sharpest phrase by Dr. Nathalie Nerée, Chief of Special Education, School District of Philadelphia — "dispelling the illusion of inclusion" — alongside a push to think past compliance and train whole schools rather than build separate models.
Los Angeles: Proximity to Classrooms
Summer school does not typically conjure joy. That is, however, exactly what we found across two days of NMIP school visits in Los Angeles Unified.
We heard from a panel of students who built guitars from scratch, where a fretboard off by a millimeter means the instrument won’t tune; and who competed internationally in designing, building, and racing hydrogen-powered radio-controlled cars against university teams, and described catching a typo in a manufacturer-supplied equation because they knew enough to know the answer was wrong — a better argument for conceptual understanding than most white papers.
The district shared their redesigned math pathway model built on “compacting, never skipping,” paired with a placement dashboard that surfaces students who qualify for acceleration but were not programmed into it. The insight travels: the binding constraint on acceleration is frequently a programming or counseling oversight, not student readiness.
They shared their emphasis on cognitively demanding instruction, with a clean line drawn between rigor as the what and CDI as the how.
A principal panel inverted the usual dynamic and turned questions back on the district leaders in the room. More than one called instructional coaches the highest-leverage hire available to them; the hard part is getting strong coaches into the highest-need schools.
The week also coincided with new published evidence. The Columbia Center for Public Research and Leadership released a new case study featuring LAUSD, documenting how the district brought nearly every middle school onto a shared math HQIM over six years through voluntary, off-cycle adoption — no mandate, one cohort at a time. On the state assessment in grades 6-8, math proficiency rose in every adopting cohort, with gains ranging from roughly 2-10 percentage points, depending on the cohort and its baseline. Early analysis from implementation partner Leading Educators found students at strongly-implementing schools gained at 3-5x the rate of comparison sites, with the effect strongest among multilingual learners and students with disabilities — the two groups most often assumed to struggle with problem-based instruction.
The mechanisms the case study describes are core ones for our districts: a shared curriculum, PL built around live teaching, teacher practice data shared (and separated from evaluation), and coaching concentrated where need was greatest.
Later in July, NMIP leaders carried this conversation to Solving the Math Problem at the ISTE+ASCD Annual Conference in Orlando.
What We Published
Fluency at Scale: What Math and Literacy Leaders Need to Know, alongside the April Instructional Spotlight from NMIP and CGCS, examines what it takes to build lasting fluency in math and literacy as a foundation for deeper learning. It was co-authored by Dr. Jennifer Bay-Williams with colleagues from Teaching Matters and Student Achievement Partners.
The Backdrop: A Rebound and a Cliff
The Education Scorecard, released in May by researchers at Harvard, Stanford, and Dartmouth, found:
Math rebounded first. After the pandemic, the annual rate of math improvement returned to pre-2013 levels between 2022 and 2024, while reading kept declining through 2024.
The recovery was U-shaped. The highest- and lowest-income districts improved most. Middle-income districts — 30 to 70 percent subsidized lunch — improved least.
Federal relief did the work. Without it, the researchers estimate districts above 40 percent subsidized lunch would have shown no math recovery at all. The highest-poverty districts remain roughly half a grade equivalent behind their own 2019 levels.
Absenteeism is still a headwind. 23 percent of students were chronically absent in 2024-25, against 15 percent before the pandemic.
That was the context for our August virtual meeting and ‘doing more with less.’ The dollars that produced the rebound are gone, and the systems serving the students who need the most are the ones now asked to sustain gains with stagnant investment.
States, meanwhile, had their most active legislative session for math in recent years:
Automatic enrollment is becoming standard. Georgia, Arizona, Iowa, and Tennessee now require high-performing students to be auto-enrolled in advanced math — a decisive move away from opt-in, with bipartisan support. This is the policy version of what LAUSD's dashboard does operationally, and HISD and Miami-Dade were some of the first systems to implement.
Screening and tiered support are arriving in math. Oklahoma, New Mexico, and Mississippi paired universal screening with tiered intervention; Michigan extended MTSS. The clearest instance of math following literacy's arc and in alignment with Excel in Ed’s Math Matters recommendations, including the model policy for K-8 screeners.
Pathways are multiplying. Connecticut now requires at least four distinct high school pathways; Indiana requires a data science pathway; Georgia's Math Matters Act directs new advanced course standards; Oklahoma and Maine added required math years. In large part due to the Dana Centers Launch Years initiative.
Professional learning is getting policy attention. Connecticut, Mississippi, New Mexico, and Oklahoma are pairing instructional requirements with approved provider lists, statewide PD, and updated certification.
Funding tied to math, however, stays largely embedded in broader STEM and intervention appropriations, and states are facing flat budgets. Legislatures are prioritizing math in statute while districts are forced to tighten belts for the next budget sessions.
Looking Ahead
One more finding from the Scorecard is worth naming. Having identified more than a hundred districts improving faster than their demographic peers, the researchers' recommendation was not a new program or mandate: it was to pair those districts with comparable systems and allow them to visit each other and compare notes.
That is the founding premise of the National Math Improvement Project. Districts do not need direction; district leaders seek structured access to peers who have solved a version of the same problem, and the time to learn from one another in person and separate from their day-to-day responsibilities.
