On cognitive overload, emotional burden, and finding your footing in a data-heavy world

If I had to choose a word of the year for education, it would be "data."

Not because data is new to schools. We’ve been living with test scores for decades — the kind that follow students from grade to grade, influence school funding, and have quietly shaped teacher careers in ways that were never fully transparent. We’ve spent years now navigating the absenteeism numbers that surged post-pandemic and still haven’t recovered.

But something has shifted. The data burden isn’t just bigger. It’s heavier in a way that’s different in kind, not just degree.

Heavy as in overload. Dr. Carla Evans of the Center for Assessment has a phrase for what teachers are experiencing: “caught in a snowstorm of data masquerading as information” (Forefront Education, 2025) Data-rich, information-poor. More numbers flowing through schools than at any previous point in history, and less clarity about what any of it actually means for Monday morning.

Heavy as in emotional weight. Because data in schools isn’t neutral. It arrives with stakes. It is tied to funding, to evaluations, to public perception of whether a school or a teacher is succeeding or failing. Research has found that using a single year of test score data to evaluate teachers carries a statistical error rate of 35 percent (Rethinking Schools, 2024). That means more than one in three teachers could be misclassified by the very instrument used to judge them. When teachers know their livelihoods depend on tools that unreliable, data stops feeling like a resource and starts feeling like a threat.

And now: Artificial Intelligence (AI).

We are now expected to protect students from data harms we were never trained to identify. To understand how AI systems leverage student data–not just to personalize learning, but to train commercial models, build proprietary datasets, and deepen partnerships with school districts that were often designed, at least in part, to acquire more data. To navigate all of this while still being accountable for the classroom data we collect, analyze, and present in meetings.Test data, attendance data, classroom data, AI data — each wave arriving before the last one was processed. Each carrying its own stakes, its own terminology, its own expectations.

The heaviness you feel is real. It is documented. And it is not your fault.

– The Shift

The problem isn't that teachers have too much data. It’s that most of the data in schools was never designed to help teachers actually teach. Too often, data is used to "push" teachers rather than support them. It is used to evaluate rather than to help us learn.

We aren't suggesting you ignore data. Instead, we want to change your relationship with it. You can move from being a passive recipient to an active questioner. Genuine data literacy isn't about being good at spreadsheets. It's about the habit of pausing before you accept a data claim.

Researchers Jim Knight and Michael Faggella-Luby found that when teachers choose their own questions—when they set their own inquiry—they commit to the work rather than just complying with it.

– The Approach

These three questions should follow you into every data meeting, every new AI tool, and every admin dashboard. They’re not a checklist. They’re a habit of mind.

Question 1: What is this data actually measuring — and what is it missing?

Standardized tests are the clearest example, because we’ve had decades to study them. The National Education Association states directly that standardized tests are “inaccurate, inequitable, and often ineffective at gauging what students actually know.” They measure a narrow slice of a narrow set of subjects at a single moment in time. They return results months after the school year ends, which is too late to change anything for the students who took them. Classroom surveys consistently show that most teachers don’t find them useful for informing instruction.

But this question applies equally to the exit ticket you ran on Friday, to the AI-generated “learning profile” built from twelve student interactions, and to the attendance dashboard showing who’s missing school. Every data source has a design. That design shapes what it can see and creates blind spots for everything it can’t.

Question 2: Who benefits from collecting this data?

This question feels uncomfortable. It shouldn’t. It’s the most clarifying question you can ask.

Some data collection serves students and teachers directly. Formative assessment — exit tickets, quick checks, student work samples — collected and used by teachers to adjust what happens tomorrow. The benefit flows to learning.

Other data collection serves more complicated interests. High-stakes test results are used to rank schools, allocate funding, and in many states evaluate teachers. AI sharpens this question most urgently. For example, some AI companies seek student data specifically to train their models or have more data points on their users to support the design of tools that keep them coming back throughout their lives (in other words, serving their bottom line). When a teacher pastes a student’s writing sample into a free AI tool, that sample may become training data. The company grows its model. The student receives–maybe–some useful feedback.

Asking “who benefits?” doesn’t mean assuming bad intent. It means being clear-eyed: your data and your students’ data have enormous value to systems beyond your classroom. Not everyone collecting it is primarily motivated by student learning.

Question 3: What does this data tell me to do — and is that the right response?

Data always arrives with implied instructions. Test scores imply: teach differently. Sometimes those implied instructions are exactly right. A clear pattern in exit ticket data tells you what to reteach and the response is obvious.

But sometimes the implied instruction is wrong. Low test scores may imply a teaching failure when the more accurate diagnosis is a test design problem, a socioeconomic factor the test can’t see, or an attendance pattern with nothing to do with instruction. An AI tool’s personalization suggestion may imply that more screen time with that product is the answer, when what the student actually needs is a conversation.

Data literacy means being willing to interrogate the implied instruction — to ask: is this what the data actually shows, or is this what the system presenting the data wants me to do with it?

– Where This Could Go Wrong

Individual data literacy won't fix structural problems. Federal laws like FERPA are outdated, and data breaches at major education companies have affected millions of students. The December 2024 PowerSchool breach compromised student information systems potentially affecting millions of students. The AI assistant deployed by the Los Angeles Unified School District shut down abruptly after its developer went into financial trouble, leaving parents with no answers about where the student data went. A 2025 Student Privacy Compass analysis of state-level AI guidance found that the overwhelming majority is superficial–states saying little more than “prioritize student privacy” without specifying how.

However, asking these questions gives you a more accurate map of the terrain. It gives you the language to participate in school-level conversations with more weight. The goal of this series isn't to make the data landscape less complicated. It's to make you feel less alone inside it.

– Try It Out

Pick one	One data encounter from this week — a meeting, a dashboard, a tool, a report.
Ask	What is this actually measuring, and what is it missing? Who benefits from collecting it? What is it asking me to do — and is that the right response?
Write it down	Not for anyone else. For yourself. You’re building a habit of mind, not completing an assignment.

What’s your word of the year? And what’s the data encounter that’s weighing on you most right now? Drop a comment below. Let's name what's hard so we can figure out what to do.