Spotlight On Evidence: Investing in Formative Assessment to Improve Teaching & Learning

The recently released 2024 NAEP scores are a reality check for educators, districts, and states. Instead of seeing an expected boost in achievement as the country moves further away from the pandemic, reading scores in both fourth and eighth grade actually declined by two points from 2022. And even though there was a small uptick in math scores, it remains that only 39 percent of fourth graders and 28 percent of eighth graders are proficient in math. Perhaps even more concerning is the fact that any observed improvements were driven by higher-achieving students, leaving struggling students falling even further behind.

As education systems grapple with how to move forward, they are increasingly looking at outliers and bright spots. Many in the field have pointed to states like Mississippi and Louisiana—places that did see significant improvements in NAEP scores—to try to uncover what policies they had in place that can explain these gains. Two things that come up repeatedly as an explanation are the use of evidence-based curricula and aligned professional development, something Overdeck Family Foundation has previously spotlighted as having a strong research base for boosting student outcomes.

Less discussed in these conversations, but also potentially critical for moving the needle on student outcomes, is the role formative assessments can play in improving student learning by guiding instruction.

To date, there is evidence across studies finding that use of high-quality formative assessment is associated with improvements in students’ math, reading, and science achievement. The research finds that to be most effective, formative assessments should be embedded seamlessly into the existing curricula, be aligned with teacher professional learning, and provide real-time information to teachers and students that can directly inform future instruction.

Our foundation has been funding formative assessments since 2022, because we hypothesized that they were critical tools for delivering on the promise of truly personalized learning. As the evidence grows, we continue to believe that high-quality formative assessments, especially when paired with HQIM and aligned professional development, can help address urgent challenges teachers face in the classroom, leading to improved instruction and student outcomes.

To yield useful information, formative assessments should help increase teachers’ knowledge of individual student’s abilities and misconceptions, which can otherwise be difficult to attain.

It makes intuitive sense that teachers who know more about their students’ academic abilities can help them learn more effectively. There is less clarity, however, about how much teachers vary in their understanding of students’ skills and how much that impacts student achievement. A recent project by researchers at Harvard University sought to build evidence to address this challenge. In a study of 284 math teachers, researchers found that teachers’ accuracy in assessing students’ skills and knowledge was teacher-specific. In other words, some teachers—typically those with stronger mathematical knowledge themselves—were good at understanding students’ skills, while others were not. This had implications for student learning: teachers’ accuracy in understanding student skills predicted stronger student achievement as measured on state standardized tests. Teachers’ knowledge of student misconceptions appeared to be a particularly important area for further work and improvement.

These findings align with earlier research identifying teacher knowledge of student misconceptions as being critical for supporting science achievement, over and above teachers simply having a strong command of the subject matter. Yet it is often quite challenging for educators to accurately assess individual student misconceptions because they are difficult to directly observe in a classroom and vary by student. Effective formative assessments can address this challenge by capturing nuanced data on how each student answers questions and performs on tasks, in addition to what his or her final answer or product looks like, subsequently making that information accessible to educators. Further innovations in technology and advances in AI may not only help capture this information in real time, but also provide targeted guidance to teachers on what to do with it, making it easier for educators to offer just-in-time support and scaffolding. As mentioned earlier, to be most effective, these assessments would need to be paired with HQIM and aligned professional development.

High-quality formative assessment can be particularly impactful for students who are most in need of help.

Some early research on assessment found that impacts were isolated to higher-achieving students. The evidence on the benefits of formative assessments for lower performing students has been less robust, but the increased gaps in achievement and the decreased performance of lower achieving students show the urgent need to better understand potential impact.

Some approaches are showing promise. For example, we recently invested in ASSISTments, a formative assessment platform primarily for grades three through eight that leverages artificial intelligence (AI) to provide immediate feedback to students on their math assignments. The platform is designed to align directly with high-quality curricula like Illustrative Mathematics and has participated in two large-scale, externally-conducted randomized controlled trials (RCTs) that build on decades of R&D. In the first RCT, researchers from SRI conducted the study in Maine, enrolling 43 schools, 87 teachers, and 2,769 seventh grade students. After one year of using ASSISTments, researchers found improvements of 0.18 SDs on seventh graders’ math achievement, which translates into about six to seven months of learning. Critically, students most in need of help benefited the most: students scoring at or below the median in math achievement prior to using ASSISTments saw impacts of 0.29 SDs, or almost a year of learning in math.

Having access to high-quality assessment data—and support for using it—can improve educator practice and student outcomes.

Given this evidence, the field is now grappling with how to develop and scale tools that help all teachers more accurately and effectively understand what their students do and do not know in an effort to improve student learning.

Overdeck Family Foundation and peer funders like the Gates Foundation and Charles and Lynn Schusterman Family Philanthropies have invested in the Achievement Network (ANet), a national provider of professional learning, technical assistance, and assessments that has found promising evidence of impact on teaching and learning. For example, an RCT conducted by researchers at Harvard University found that having access to ANet’s assessment platform doubled the amount of time teachers reported reviewing and using data. Critically, in schools the research team assessed as having a high readiness to implement the platform, these impacts also trickled down to students. Teachers’ satisfaction with and use of assessment data positively predicted growth in students’ math achievement. And, after two years of implementation in those schools, ANet’s assessment approach yielded improvements in students’ math (0.18 SDs) and ELA (0.12 SDs) achievement. At the same time, null effects on student achievement in schools that had lower readiness to implement ANet’s approach highlight the critical role that school culture and leadership play in creating a data-driven culture where assessment can deliver on its goal of boosting achievement.

Our foundation is now supporting ANet to enhance its assessment platform so that teachers can reduce the time spent analyzing student data and increase the time spent on direct instruction. We believe these improvements will serve to make these assessment data more useful, scalable, and impactful, a hypothesis that the organization will test through careful research.

As we’ve explored before, evidence-based practices like the science of reading, high-quality instructional materials, high-impact tutoring, and aligned professional learning are all effective tactics for driving positive outcomes (likely most effective when done in tandem). Formative assessments are another critical piece of the puzzle, and one that is similarly rooted in evidence. Recent advances in AI technology hold particular promise for seamlessly embedding assessment into instruction in addition to making assessments personalized for each student and easier for teachers to understand and act upon.

As we continue deepening our investment in this space, we’re committed to learning more not only about what these technologies look like in practice, but whether our hypothesis that personalization can measurably improve teaching and learning outcomes in the medium and long term pans out. We invite other funders and partners to join us in these efforts by reaching out to research@overdeck.org.

Header image courtesy of ANet