This post is the first in a series of four sharing new insights from the Canopy project using data published in fall 2019. While the findings reflect trends in school design before the COVID-19 pandemic, they nevertheless suggest patterns that may prove informative as schools nationwide reconsider elements of their designs.

Where are schools reimagining teaching and learning across the country, and what approaches are they adopting? The Canopy project, a collaborative initiative led by the Christensen Institute, set out to answer these questions. The project uncovered a set of schools—many of which have not received widespread recognition—and revealed the practices these schools use in pursuit of student-centered learning. 

The Canopy uses a set of ‘tags,’ or a common lexicon of keywords and phrases, to indicate the presence of innovative approaches and practices in each school. (The entire set of 88 tags and their descriptions can be found by downloading the dataset on the Canopy website.) 

What’s compelling is not just the individual practices schools report implementing, but which practices appear to be implemented together. Uncovering the ways schools combine practices helps suggest categories of practice, which is one step towards building common understanding of, and eventually evidence for, the innovative approaches that drive student outcomes in specific circumstances. 

Using a methodology called exploratory factor analysis, we sorted the tags schools cited into five cluster groups based on how often they come up together. The five clusters do not, in themselves, offer answers when it comes to what practices individual schools implement. Instead, like distinguishing the shapes of clouds in the sky, the clusters help us make sense of bigger thematic categories in school innovation among all schools in the Canopy. And rather than these categories of practice being determined from the top-down by experts or thought leaders, they emerge from the bottom-up—from data on school designs reported by schools themselves. Here’s what surfaced in our analysis:

Four intuitive categories

It was relatively easy to recognize themes in the first four clusters and assign each one a category name using language that is common across the education sector. 

1. Blended Learning: This set of practices mostly features strategies for personalization powered by technology and data. The tags most strongly associated with this cluster were station rotation, real-time data use, and data analytics drive instruction

2. Project-Based Learning: This set of practices relates to hands-on, active instructional methods where projects play a large role. The tags leading the pack in this category were projects as primary method of learning, interdisciplinary/multidisciplinary, connecting local and global, design thinking, service learning, exhibitions, and real-world problem solving.

3. Competency-Based Education: This set of practices is linked to assessment, grading, and systems that measure learning based on competency rather than seat time. The tags most strongly associated with this cluster included competency/mastery-based education, grading policies focus on mastery, learner agency, flexible assessment schedule, competency framework, and advancement on mastery

4. Equity and Social-Emotional Learning: This set of practices supports students’ overall well-being, especially for marginalized students most likely to benefit from integrated supports. The tags with the strongest relationship to this cluster were dialogue circles, family and community support services, supports for high-poverty and homeless students, practice of culturally relevant pedagogy, and wraparound services and integrated student supports. Other tags in this cluster represented practices more explicitly tied to social-emotional learning (SEL), such as commitment to whole child or SEL in strategic plan, and SEL integrated across academics

The category names above are not perfect—they are the names we chose to use as shorthand to describe each category. Some people might prefer the term “inquiry-based learning” to “project-based learning,” for example. Nevertheless, we were struck by how each cluster of practices has a consistent theme. 

The mystery fifth category

The last cluster of tags in our analysis initially mystified us, featuring a range of practices that didn’t seem intuitively related. Some, like career training and preparation and measures for college readiness, suggest a category like “college and career readiness.” Others point to shifting boundaries between school and the real world, like credit for learning outside the classroom and community and business partnerships. A number of practices enable more flexibility and customization of learning, such as supports for students off-track for graduation, take-home devices, and individual learning paths. And some practices promote supportive relationships like advisories and 1:1 mentoring. We found ourselves challenged to name this category, eventually landing on “Flexible Pathways to College & Career.”

A lightbulb went off when we calculated which schools—elementary, middle, or high—were more likely to cite practices from each of the five categories. The fifth category was the only one with a stronger positive correlation with high schools, and a negative correlation with elementary schools. A second look at the practices in that category confirmed that many of them are relevant for older students on the brink of transition to postsecondary options.

There’s a lot of energy around redesigning high schools, with the XQ Super School project as the most high-profile example. And “college and career ready” has been a rallying cry among education systems for years now. But the variety of practices featured in the fifth cluster of Canopy tags suggest that there’s still much to learn about the various approaches schools take toward supporting older learners.

Learning from five categories of innovative practice

Our analysis suggests several questions we propose as provocations to the sector:

  • Are there distinct models being developed by innovative high schools seeking to prepare students for postsecondary options? For example, Rooted School, a charter high school in New Orleans, organizes behind the purpose of closing the wealth gap for African Americans. Loudoun County Public Schools in Virginia focuses on empowering students to make contributions to the world. Both school systems offer work-based and flexible learning opportunities for students, but the design and implementation of those programs may vary.
  • Where and how do elementary and middle schools implement some of the practices that, based on the data, appear more commonly in high schools? For example, Cajon Valley Union School District introduces the “World of Work” to students as early as kindergarten.
  • How do practices supporting the most marginalized students intersect with social-emotional learning (SEL)? While all students can benefit from practices designed to build social-emotional skills, many of the practices in the Equity and SEL category are specifically designed to support marginalized students. Further research should investigate the intersections among these practices. 
  • How do we facilitate synergies among practitioner communities seeking to deepen their work in these five areas of innovation? Some practitioner communities already align their work with the categories we identified, such as how The Learning Accelerator aggregates resources for blended and personalized learning. In other cases, there may be an opportunity for existing but disconnected communities to collaborate if the practices they promote are being implemented together in schools. For example, the final cluster of tags suggests that high school designs may blur the boundaries between Career and Technical Education (CTE), “real world learning,” and alternative pathways for youth.

Debating categories and terminology may feel abstract, but we can’t identify what leads to impact without clarity on the “what.” If education stakeholders collaborate to define schools’ practices and categorize them consistently, we’ll be one step closer to identifying what works in which circumstances toward student-centered learning outcomes.


  • Chelsea Waite
    Chelsea Waite