Project

Evaluation of the Networks for School Improvement (NSI) Initiative

Educators are leaning into network-based continuous improvement (CI) to tackle long-standing teaching and learning problems. Although there are studies of specific networks, there are limited evaluations that examine a large group of varied school improvement networks.

In 2018, the Gates Foundation established the Networks for School Improvement (NSI) to increase the proportion of students, especially those who are Black, Latino, or experiencing poverty, who are on track for high school graduation and college enrollment.

Learn more in this blog post from the Gates Foundation, Lessons Learned to Inform Student and Teacher Success.

The initiative supports intermediary organizations that work with networks of schools in using CI methods to identify and test strategies that improve teacher practices and student supports; it mainly took place from the 2018–19 through the 2024–25 school years.

This rapid scale-up of networked CI provided a unique opportunity to understand how intermediaries develop and implement school networks, how schools in these networks implement CI to improve student outcomes, and what the impact of these efforts are.

The Summative Evaluation of NSI

AIR, RAND, and Mathematica received funding from the foundation to conduct a summative evaluation of the initiative.

Schools’ use of CI (led by AIR)
Intermediaries’ approach to developing and supporting school networks (led by RAND)
The relationships between network activities and schools’ use of CI (co-led by AIR and RAND)
The impacts of the NSI initiative on student outcomes and the relationship between those impacts and NSI implementation (led by Mathematica)

The three research organizations jointly developed a summative brief that synthesizes findings across the individual evaluation reports.

School-Level Implementation of Continuous Improvement

AIR examined the implementation of CI at the school level and explores the aspects of implementation at the school level that may be related to school and network factors.

Final Report | Appendices A through G (April 2026, PDF)

Our report draws on data from 25 NSI, spanning the 2020–21 to 2024–25 school years, using artifacts collected from CI teams, surveys of school leaders, interviews with intermediary staff, and case studies from a sample of participating schools.

Analyses focused on the following:

Implementation of key features of CI (e.g., inquiry cycles, root cause analysis, theory of practice improvement, aims, change ideas, and data use);
How CI teams address and incorporate equity;
Variation in implementation across the 5 years of each grant;
The conditions that enable schools to implement CI;
How CI practices are spread within a school or beyond; and
The sustainability of implanting CI after grant funding ended.

Key Findings

Much of the literature on networked improvement emphasizes a focus on iterative inquiry cycles using the Plan–Do–Study–Act (PDSA) model. While many NSI and schools adopted this model, some relied on other approaches to regular data use, including goal-setting routines and early warning routines.
Many of the inquiry cycles that schools launched were incomplete—they did not include all the Plan, Do, Study, and Act phases.
There was considerable variation across NSI and schools in the implementation of CI.
Although there was some progress in CI implementation over the grant years, progress varied across NSI.
Principal support and stable school teams, among other factors, are associated with positive implementation, but many schools lacked these conditions.
CI practices that are easiest to spread within a school and beyond are likely those that can be completed without conducting an inquiry cycle (e.g., root cause analysis).
After foundation funding ended, about half of the schools maintained at least some CI practices and their CI team.

What We Learned

We found that participating schools adopted some of the key features of CI, but implementation varied substantially across NSI and participating schools.

Our results underscore the importance of robust documentation of CI work, ongoing professional development to support both CI practices and the interventions being tested, and leadership support to maintain and spread effective practices.

We recommend future research should examine strategies to enhance implementation of CI at the school level, mechanisms behind sustainability, and the long-term effects of CI on student outcomes, particularly for historically underserved groups.

Networks, Coaching, and the Implementation of Continuous Improvement in Schools

AIR and RAND examined the association between coaching (a network support) and network cohesion with school-level CI implementation of CI. The analyses are exploratory, taking advantage of the opportunity to link data on coaching and cohesion from RAND’s evaluation with data on CI implementation in schools from the evaluation led by AIR.

Final Report (April 2026, PDF)

Key Findings

Although coaching was widely used to support school CI team understanding and use of CI, there was negative, but modest, association between the frequency of coaching and schools’ CI implementation. Although the results were initially unexpected, interviews with network hub staff demonstrated that more coaching was often provided to school CI teams with greater needs.
Network cohesion—as measured by the density of network connections among participating schools and organizations—was related to schools’ implementation of CI practices. In particular, it was associated with fewer, but more complete inquiry cycles, more adoption of change ideas, but fewer elements of network aim statements implemented.
Network hubs were intentional in establishing cross-school connections within their networks, which created opportunities for data sharing and progress monitoring, peer accountability, and comparing successful change ideas.

What We Learned

The results suggest that building dense network connections among schools may be an important avenue to facilitate school improvement.

We recommend future research should explore whether coaching best supports capacity building for CI when it is purposefully tailored to the needs, contexts, and learning trajectories of school CI teams. Additionally, we recommend further exploring the role of hubs within an improvement network and how their role influences how schools engage in continuous improvement.