We know more about how to improve math outcomes than current policy debates suggest

This post is adapted from an email originally shared on March 8, 2024. If you would like to receive future emails, you can sign up here.

I recently watched a 7th-grade math lesson that did a better job than I ever did as a teacher asking kids relatable theoretical probability questions. How would you represent the probability of a 6-foot-tall 7th grader? How would you represent the probability of getting a test in school in any given week? Making sense of where students were coming from was a fascinating puzzle. Why did one student think the probability of there being a 6-foot-tall 7th grader was absolutely certain (6/6) while another student thought it was unlikely and round to 0/6? Why did everyone represent all of their probability fractions using a denominator of 6? (The experimental probability lesson the day before involved dice, of course.) Watching a lesson with great content and instruction reminds me, without question, that every student can be a math person.

Unfortunately, math learning results far from match that clear potential. Looking at math learning outcomes closely is the most heartbreaking thing a lover of mathematics can do. The overall outcomes are gutting; certain student subgroups have starkly disparate outcomes — with single digit proficiency quite common — and the COVID drops were stark.

These outcomes — and the sense that “we did a lot around reading so we need to pay equal attention to math” — is leading math to get worthy attention by state leaders, policy makers, and education reporters right now. The approach to both reading and math policy conversations misses the mark in ways that will limit their long-term impact.

When policy makers ask, “what should we do to improve reading outcomes?,” there are a few clear and obvious things that most experts agree would make a positive difference. In foundational skills in reading, there are teacher training, curriculum, and assessment moves based on well established research that seem like they will help. As I have written about before, I worry that the reading policy reforms project too much confidence that a narrow set of (important) strategies will strengthen outcomes in a subject that is highly interconnected and notoriously hard to move. I worry they focus too much on foundational skills (decoding) alone and lack attention to implementation — especially around school leadership team capacity building — that will make or break success. I worry the reading reforms are set up for disappointment and even backlash when those plays don’t lead to immediate positive growth.

In contrast, when policy makers ask, “what should we do to improve math outcomes?,” answers feel murkier. Many have pointed out that we don’t have a “science of math” the way we have a “science of reading” — meaning there have been relatively few neuroscientific studies dedicated to understanding how kids learn math compared to how kids learn to read. Others feel there are things that are well understood about how kids learn math, but those findings are not yet instantiated well in curriculum or practice (commentary on this here, here, and here). Relative to reading, there are fewer assessments and interventions that have been highly tested and scaled in math. It is harder to find math teachers, and many teachers have steep learning curves in both content knowledge and understanding of how to teach the subject. It feels tougher to pull together a package of math reforms in math than in reading.

I worry that interest in “doing something that feels similar to what we did in reading” will lead policy makers to require schools to do things that don’t make sense. I worry generally about the burden that every requirement adds to schools. And, I worry especially about additional required assessments.

But more significantly, I worry that the policy debates I am watching seem to have forgotten that we know a lot about what to do to improve math outcomes. Between 1992 and 2019, 8th grade math results grew 20 scale score points (compared to 3 points in reading) with outsized progress for Black students, AAPI students, and multilingual learners. It is far from enough progress, but there are things we can learn from this growth at scale. Furthermore, at the unit of a school, every state can find schools that have seen dramatic improvements in math across student groups. There are local examples to lift up and learn from.

At Instruction Partners, especially during the last few years, our partners have seen higher growth than the state in math overall and for students experiencing poverty, multilingual learners, students with disabilities, and almost every group of students of color. Both partners that work with us directly and partners that work with regional service centers that we have coached have seen these outsized impacts.

Many factors may be shaping these outcomes, but here are some things that seem to be driving stronger instruction:

1. The school team (teachers and leaders together) benefits from a shared vision of what good math instruction looks like in action that moves beyond either/or thinking about procedural fluency and concepts and embraces the need for both.
2. Teachers can learn a lot about the content and how to teach it by studying good curriculum in well supported collaborative planning structures. It is easier to chunk and build this math content knowledge throughout the year while preparing for the next unit than to overview it all in one big dose over the summer.
3. Coaches and leaders (even those who don’t feel like they are math people) can also learn to give math teachers useful feedback when they use the curriculum as a guide. Leaders modeling a learning stance about math gives teachers, and students, permission to do the same.
4. School and district leadership teams benefit from short cycles of improvement focused on the question: “are the moves we are making working?” Short cycles support focus, follow through, and problem-solving. Improvement happens over time.

These are the kind of practical and disciplined leadership moves that I sense will be a big part of any reform at scale — and that I fear current policy debates are missing.

The theoretical probability that kids can learn math is a 6/6. I sense we will help them reach that potential faster if policy debates stay very close to instruction and build on what we already know can power growth.