Impact Evaluation · 2017–2025

Across ten districts and eight years of data, student math achievement improved.
Here is what the evidence shows.

From a high-poverty rural supervisory union in Vermont to a Colorado district serving more than 15,000 students, we kept finding the same pattern. This evaluation is our most complete account of what it is, what drives it, and where the evidence remains limited.

Authored by John Tapper, PhD, founder and Chief Executive Officer, All Learners Network · December 2025

10

District partnerships

in the evidence base, across five states, 2017 to 2025

23/23

Thompson School District schools

showed statistically significant math improvement over three years

+23pts

Above state math averages

Worcester County, Maryland, six years after partnering with ALN

3×

State proficiency rate for students with IEPs

C.P. Smith Elementary, Vermont, on state math assessments

For district leaders and school administrators

About this evaluation

This evaluation synthesizes student achievement data, teacher surveys, classroom observations, and administrator interviews from ten school districts and supervisory unions in Colorado, Maryland, Massachusetts, New Hampshire, and Vermont, spanning 2017 to 2025. Achievement data comes from state assessments, iReady diagnostics, and the Vermont Comprehensive Assessment Program (VTCAP). The evidence points in an encouraging direction. We share the full picture, including the limitations.

The evaluation was authored by John Tapper, PhD, ALN's founder and Chief Executive Officer. We publish it in full, including its limitations section, because we believe that intellectual honesty is part of how this work earns trust with the district leaders and educators we partner with.

The central question: when districts commit to sustained professional development in mathematics grounded in High Leverage Concepts, formative assessment, and differentiated instruction, what actually happens to teaching practice and student outcomes?

What this evaluation covers

Quantitative findings
Student achievement data from state assessments and iReady diagnostics across ten sites, including effect sizes and year-over-year growth patterns.
Qualitative findings
Evidence of instructional change in three areas: differentiation and grouping practices, formative assessment use, and conceptual focus versus curriculum coverage.
Cross-site patterns
Three patterns that appeared across sites: the timeline of change, disproportionate gains for historically underserved students, and the role of implementation conditions.
Limitations and methodology
A transparent account of what the evidence can and cannot prove, including the absence of comparison groups and variable data quality across sites.
What the evidence shows

Three patterns across ten sites


Finding 01

How long before a district sees results?

In site after site, ALN implementation produced a consistent two-year pattern. Instructional practices shifted in Year 1 as teachers began using Math Menu structures, formative assessment protocols, and conceptual frameworks built around High Leverage Concepts. Measurable student achievement gains followed in Year 2.

YEAR 1
Instructional change
Teachers shift to Math Menu, formative assessment protocols, and High Leverage Concepts focus. Practice changes are measurable before achievement gains appear.

YEAR 2
Student achievement gains
Measurable gains in student math outcomes emerge, consistent across state assessments, iReady diagnostics, and VTCAP.

 

This pattern held across different geographic contexts, student populations, and assessment instruments. One notable exception: Keene School District, which invested a full year in coaching infrastructure before beginning classroom implementation, showed 24% district-wide proficiency gains in what was effectively their first implementation year. The evidence suggests that a robust planning year may compress the timeline.


Finding 02

What about the students who need it most?

Perhaps the most notable finding across sites: students who typically show the smallest response to instructional reforms showed the largest gains here. Students receiving special education services, students from economically disadvantaged backgrounds, and English language learners demonstrated accelerated growth compared to general education peers, across multiple districts, assessment types, and years.

This pattern runs counter to common findings where educational interventions produce the largest effects for already-advantaged students. The evidence includes:

  • C.P. Smith Elementary, Vermont: students with IEPs achieved VTCAP math proficiency rates exceeding 50%, compared to approximately 18% statewide for comparable populations.

  • Keene School District, New Hampshire: students from economically disadvantaged backgrounds showed a 35% proficiency increase, compared to a 24% district-wide gain.

  • Winooski Public Schools, Vermont: English language learners in responding grade cohorts demonstrated statistically greater growth than native English speakers, in Vermont's most linguistically diverse district.

The evaluation suggests that ALN's emphasis on multiple representations, conceptual access points, and instruction matched to each learner's mathematical thinking may reduce the barriers that typically disadvantage these student groups.


Finding 03

What makes it work, and what gets in the way?

Beyond whether achievement changed, the qualitative evidence examines how and why. Three factors consistently distinguished sites with stronger outcomes from those with weaker ones.

Embedded coaching alongside workshops. Sites that paired sustained instructional coaching with professional development workshops consistently outperformed sites that received workshops alone. Thompson School District's district-wide consistency, supported by monthly embedded coaching visits to all 23 schools, provides the clearest evidence.

Administrative support for depth over coverage. Teachers who received explicit backing from building and district leaders to prioritize conceptual depth over curriculum pacing showed stronger implementation and better student outcomes. Winooski's within-district variation tells this story directly: the grade levels with principal support for slowing down showed large gains, while the grade level facing pacing pressure showed no significant change, in the same district, with the same professional development.

Internal capacity built to last. Worcester County's six-year sustained improvement, well beyond active external support, coincided with the district's decision to train building-level coaches, revise benchmark assessments, and align pacing guides to High Leverage Concepts. When districts build internal systems that support the instructional model, improvement persists through staff turnover and leadership change.


District by district

The four most robust evidence sites

Colorado · 2022–2025

Thompson School District

23/23
Schools with statistically significant improvement


Serving more than 15,000 students across 23 elementary and middle schools in Northern Colorado, Thompson implemented ALN professional development district-wide in 2022. All 23 schools showed statistically significant gains on iReady mathematics diagnostics across three consecutive years, with elementary effect sizes ranging from 0.66 to 0.91.

Maryland · 2017–2023

Worcester County Public Schools

+23pts
Above state math averages, six years after partnering with ALN


Five high-poverty elementary schools where PARCC scores initially trailed state averages. By 2023, six years after the initial partnership, Worcester County exceeded state math averages by 23 percentage points, maintaining that performance through staff turnover and leadership changes.

New Hampshire · 2022–2024

Keene School District

+24%
District-wide math proficiency in first implementation year


Keene invested the first year of partnership in infrastructure: developing 12 instructional coaches, aligning assessments, and building administrator understanding of the instructional model. In the first year of broad classroom implementation, district-wide proficiency increased 24%, with economically disadvantaged students showing a 35% proficiency gain.

Vermont · 2023–2025

Winooski Public Schools

0.619
Effect size (large) for Grade 7 cohort on VTCAP


Vermont's most diverse district, with 44% English language learners and 68% economically disadvantaged students. Students who began ALN-trained instruction in Grade 5 showed large, statistically significant growth on VTCAP by Grade 7. English language learners in responding cohorts showed greater gains than native English speakers.

In their own words

What teachers say changed

"I used to look at a worksheet and just count how many they got right. Now when I do a work sort, I'm looking at how they got their answers. Did they use a visual model? Did they rely on memorized procedures? Are they showing multiplicative thinking or are they still additive? That tells me so much more about what they actually understand and what they need next."

Fourth-grade teacher

Keene School District, New Hampshire

"Instead of taking my kids out to drill addition facts, I'm in the classroom during Math Menu. Sometimes I'm working with students with IEPs on place value concepts, but sometimes I'm working with students who just need more time on that concept, whether they have an IEP or not. And I get to see what grade-level work looks like, so I can build toward that instead of just practicing procedures in isolation."

Special education teacher

Worcester County Public Schools, Maryland

Intellectual honesty

What this evidence can and cannot prove

We believe the limitations of a piece of research are part of its credibility. This evaluation's conclusions are described as "suggestive evidence of positive association" rather than as proof of causation, for good reasons. The most significant constraints on causal inference are:

  • No comparison groups. No studies employed randomized control designs or matched comparison schools. Concurrent changes in curriculum, leadership, or policy could contribute to outcomes alongside ALN implementation.
  • Selection effects. These are districts that chose to partner with ALN, often because they had identified significant needs. Districts with different characteristics or readiness levels might experience different outcomes.
  • Variable data quality. Some sites provide strong quantitative data with limited qualitative evidence. Others provide the reverse. Few provide both. This limits conclusions about which specific practices drive results.
  • Publication bias. This evaluation documents partnerships where data was collected and analyzed. Sites where partnerships concluded without formal impact analysis are not represented here.

Despite these limitations, the magnitude and consistency of achievement gains, across diverse contexts, assessment instruments, and student populations, exceeds what measurement error or regression to the mean would predict. ALN has engaged program evaluators for comprehensive analysis across current partner districts in 2024-2025, which will provide more rigorous evidence with comparison conditions and systematic fidelity measurement.

Full evaluation

Read the complete findings

The full evaluation includes detailed quantitative findings for each site, teacher and administrator interview data, analysis of implementation conditions, and the complete methodology and limitations discussion. No form required.
All Learners Network · December 2025 · Authored by John Tapper, PhD
Also free

Get the High Leverage Concept Maps

The evaluation references High Leverage Concepts throughout, because they are central to what teachers learn to focus on. Our free High Leverage Concept Maps show exactly what those concepts are and how they build across PreK-8 grade levels. Enter your email and we will send them right over.

Download the PreK-8 High Leverage Concepts

Next Step

See what this could look like in your district

The evidence is encouraging. What matters most is whether it translates into your context, with your students and your team. Let's talk through what that looks like.