What the Data Is Telling Us

And why learning communities share assessment data early

Photo by Nathan Dunlap on Unsplash

Ms. Jan’s* Answer

I worked at an international school with an amazingly robust support system—five classroom teachers, a learning support (LS) specialist, and an English as an Additional Language (EAL) teacher, all working on one team, sharing one collaborative office. Unfortunately, the LS and EAL weren’t fully in. Their shared perspective was: students should be grouped by ability and I’m going to work with only the students on my caseload, subject (or ability) be damned. If a student was on their list, that is who they were teaching—regardless of what the student actually needed in a particular subject area. 

We were in our 2nd or 3rd math unit of the year, and the homeroom teachers on the team were trying desperately to shift these patterns. We knew students needed mixed groupings and different teachers, but we also knew that baby steps were best. So when we were grouping mathematicians, Jan stepped forward quickly and said, “I’m going to teach this group.” The group she pointed to was the one that the learning support teacher had been clutching for months. The students weren’t being challenged, the data showed they weren’t progressing, and most of us knew they needed something different. Jan was the one who said it out loud first. 

As the story goes, Jan walked in after the students were already seated—into a classroom they didn't expect to see her in. One of the students looked up and said, “Miss Jan, I think you’re in the wrong classroom."

"What makes you say that?" she asked.

"We're not the smart kids," the student said. "You teach the smart kids."

Jan didn't miss a beat. "That's actually why I'm here," she said. "I heard you're some of the smartest mathematicians in our community. You think about things so differently. I want to learn from you. And then I was hoping you could teach your strategies to others in the community that need your help."

That was day one. Within four weeks, most in the group showed marked improvement—enough that no one on the team could keep defending the caseload model, at least not in math. In the next grouping, the classroom teachers (who were already on board) were able to push for alternative groupings. For two weeks, mathematicians were grouped by how they understood and communicated ideas instead: some who could write out a problem, some who could interpret it, some who could draw it. Different entry points into the same thinking. Something we hadn't needed to formalize in any other subject—because in every other subject, we hadn't let a child's ability quietly calcify into an identity.

What the Data Is Telling Us: Sharing Assessment Early

Jan's answer is one response to a complex challenge many learning communities face early on. Once co-planning starts working the way it's supposed to, data is flowing constantly. More data points on this student, that group, this profile of learner. That's a good problem to have.

But it comes with two real risks. The first is time—there's suddenly so much to consider that a planning meeting can disappear into it. The second is more dangerous: grouping by skill level can quietly turn into streaming. And streaming is not what a learning community is for.

The Blue Bird Problem

You know the group. Every elementary teacher does. The one that started as a temporary cluster of kids who needed a bit of remediation before a unit really took off—and somehow, by week four, week eight, or week eighteen, is still the same eight kids, doing the same simplified version of the work. The label stuck before anyone noticed it was happening.

Groups should move in a silo classroom. And they definitely need to move in a learning community. A group of twenty who needed extra scaffolding in week one should not all still be together in week three. As foundations strengthen, kids should shift—toward a stronger thinking partner, toward someone they trust enough to be vulnerable with, toward a completely different kind of challenge, towards a teacher who has another way of teaching them or new insights to observe.

This is where co-planning and co-teaching conversations matter most, because it's in that room that someone has to be willing to say: this is feeling very blue-birdish. Or: let's look at this grouping; is it stagnating?

There's no universal script for this moment—but there should be one for your team. It's worth creating one at the vision-crafting stage, before trust is tested by an actual difficult conversation. Decide now what your team will say to each other when grouping has calcified. Agree on it while it's still hypothetical.

Left unnamed, the pattern doesn't stay small. What used to be one child labeled in one teacher's silo becomes a child labeled by five adults across an entire learning community—the exact opposite of what the model is supposed to protect against.

Why This Shows Up Most in Math

In my experience, this happens far more in math than in any other subject. Literacy differentiation tends to feel more natural—most elementary teachers can sense a reader's or writer's level in the moment, through conferring, almost instinctively. Math is different. It's more visible, more comparable, more instantly right-or-wrong.

There's research behind that instinct. Math is one of the only elementary subjects with a genuinely dominant competition culture—Math Kangaroo, MOEMS, math olympiads turn problem-solving into contest sport for kids as young as six or seven. And part of the reason for this competition is structural: answers are binary, gradeable at scale, with none of the human judgment that essays or historical arguments require. Math also has what's sometimes called an "infinite ceiling," meaning you can make a problem harder without changing the language needed to understand it (just substitute bigger numbers). This is different from literacy comprehension, which eventually requires emotional maturity or abstract thinking that a young child hasn't developed yet.

The cost of that visibility is real. A study tracking students ages 7 to 11 found that children placed in lower-ability math groups were significantly less likely to ever enjoy the subject again—even after controlling for their actual ability. The self-concept damage outlasted the grouping itself (ScienceDirect). And across large-scale studies, the academic benefit of streaming averages out to essentially nothing—while the students placed in lower tracks are the ones who consistently lose ground (Education Endowment Foundation Toolkit; Kulik & Kulik, 1988/1992).

Stale Data Is Its Own Kind of Sabotage

One more thing worth saying out loud is that sometimes the barrier to good grouping isn't a person, it's the data itself. Using "official" reading scores from September to group students in March tells you (almost) nothing. Anecdotal documentation, the kind that happens naturally within a team with real trust, is often more reliable, because it's current.

The same goes for bottlenecks. If a team can't regroup students until one specific person has finished scoring a pre-assessment, that delay quietly does the same damage as the resistance I mapped in the Knoster model last week. It's not sabotage in the dramatic sense. It's just friction that keeps kids in groups they've already outgrown.

The Real Takeaway: Rotate the Eyes, Not Just the Groups

If I could encourage your team to do one thing to give it a go this week—it’s this: as students change, the data changes, and the groups need to change with it. That's not a disruption to the learning community model. That is the model.

There is no reason I should be Jason's math teacher for an entire year. In a learning community, I might teach Jason math for part of the year, guide his reading at another point, support his writing later, walk alongside a personal inquiry after that. My colleague—the EAL specialist—might also spend time with Jason, not because he needs EAL support, but because she's another set of eyes on him, the same as every other adult in the community.

That rotation is the point. If your data tells you one teacher has seen a student far more than anyone else on the team, that's not neutral information. That's a signal.

Mix it up.

*names changed to protect anonymity

Note: This post was edited with AI as an assistant to refine structure and readability. My ideas, voice, and words remain intact.