← Back to Library

How can teachers find clarity in a sea of confusing data?

Natalie Wexler cuts through the noise of modern education reform by identifying a paradox: teachers are drowning in data but starving for clarity. She argues that the solution isn't just more research, but a smarter way to filter it, leveraging artificial intelligence not as a student crutch but as a rigorous research assistant for educators.

The Information Deluge

Wexler opens by diagnosing a systemic failure in the classroom: the "chaotic information environment" that leaves even well-intentioned teachers confused. She cites education researcher David Griffith, who notes that an older teacher might offer one piece of advice, a new curriculum another, and a think tank article a third. "Teachers want to do the right thing," Griffith is quoted as saying, "but there's a lack of clarity about points that really should be clear." Wexler uses this to pivot to a broader issue than just reading instruction; the entire profession is paralyzed by conflicting signals.

How can teachers find clarity in a sea of confusing data?

This framing is effective because it shifts the blame from individual teacher competence to the structural disarray of the profession. However, it also highlights a deep vulnerability in the system: reliance on pre-service training that often contradicts current cognitive science. Wexler points out that despite clear evidence, nearly 60 percent of teachers still view reading comprehension as a set of generalizable skills rather than a product of background knowledge. This disconnect is not accidental; it is the result of decades of educational orthodoxy clashing with how the brain actually learns.

"If we could somehow improve the quality of pre-service preparation, we would really be making progress, because it is hard to change the practices of teachers who have been teaching for 15 to 20 years."

The author suggests that fixing this requires a two-pronged approach: better tools for current teachers and better training for future ones. She introduces "The Evidence Checker," a new AI tool developed by researchers Nidhi Sachdeva and Jim Hewitt. Unlike generic chatbots that might hallucinate or offer vague platitudes, this tool uses a sophisticated prompt engineering strategy to force the AI into the role of a critical research assistant. The prompt explicitly instructs the model to avoid jargon, think critically, and provide a summary of the evidence state before offering a conclusion.

Wexler tests the tool herself, asking whether reading instruction should focus on knowledge or skills. The AI correctly identifies the question as a false dichotomy, a nuance that generic models often miss. Yet, Wexler remains skeptical about the tool's transformative power on its own. She notes that even without the specialized prompt, standard models like Perplexity and ChatGPT produced similar, albeit less detailed, answers. This raises a critical counterpoint: if the underlying data is the same, does the prompt really change the outcome, or does it just make the output more readable? The answer likely lies in the volume of obscure or controversial questions where a generic model might fail, but for common topics, the marginal gain is debatable.

The Limits of Individual Agency

The commentary takes a sharper turn when Wexler addresses the "implementation gap." Even with perfect information, individual teachers struggle to change. She references the revised edition of The Science of Learning by Deans for Impact, a document designed to translate cognitive science into practical classroom guidance. The new version includes "Pitfalls to be wary of," such as the misconception that memorization is merely rote regurgitation. In reality, as the document explains, "when students have more knowledge stored in long-term memory, they have more capacity for complex thinking."

This historical context is vital. It echoes the principles of cognitive load theory, which suggests that working memory is limited and must be supported by long-term memory structures. Wexler argues that while these resources are excellent, they are insufficient to move the needle significantly. The problem is not just a lack of knowledge; it is a lack of systemic support. Teachers are often given instructional materials or administrative directives that contradict the science they are trying to apply.

"Most professional development, including that focused on cognitive science, is aimed at changing teachers' beliefs in the hope that it will change their practice. But the evidence indicates that the dynamic more often works the other way: If teachers see their students clearly benefiting from an approach they're asked to try, they're likely to embrace it."

This inversion of the standard change model is a powerful insight. Wexler suggests that we should stop trying to convince teachers to believe in the science before they use it. Instead, we should provide them with coherent, evidence-based curricula and support structures that allow them to see results immediately. This aligns with findings in implementation science, where the success of a new practice often depends less on the user's initial attitude and more on the fidelity and support of the system implementing it.

Critics might argue that this approach risks treating teachers as mere technicians rather than professionals, stripping them of their autonomy to adapt instruction. However, Wexler counters that without a coherent curriculum, teachers are left to invent their own solutions in a vacuum, which often leads to the very confusion she described at the start. The goal is not to remove teacher agency but to provide a scaffold that makes effective practice possible.

Bottom Line

Wexler's most compelling argument is that the path to better education lies not in more teacher training seminars, but in systemic changes to curriculum and assessment. While tools like The Evidence Checker are promising, they cannot fix a broken infrastructure. The strongest part of her piece is the realization that practice must precede belief; teachers will adopt science-based methods only when they see them working in their classrooms. The biggest vulnerability remains the political will required to overhaul curricula and testing regimes that are deeply entrenched in outdated assumptions.

Deep Dives

Explore these related deep dives:

Sources

How can teachers find clarity in a sea of confusing data?

by Natalie Wexler · Natalie Wexler · Read full article

What’s the main reason we don’t see more evidence-based instruction in schools? According to education researcher David Griffith, as quoted in The 74, it’s the “chaotic information environment that the typical teacher is subject to.”

“An older teacher tells you one thing,” Griffith said. “Your curriculum tells you something else. You read an article online written by some think tank and it tells you a third thing. Teachers want to do the right thing, … [but there’s a] lack of clarity … about points that really should be clear.”

Griffith was discussing the “science of reading”—and more specifically, a recent report he co-authored finding that nearly a third of early elementary teachers still use methods that conflict with well-established evidence. But his comments could apply to teaching in general.

There’s an enormous amount of research out there on education, but findings can conflict, and it’s not always clear which ones are reliable. Even if you know a study is trustworthy, it can be a slog to read and make sense of the highly technical language used to convey the results. (Ask me how I know.)

But there’s a new AI tool that can help. Yes, AI. I’ve written about the serious risks of allowing students unfettered access to generative AI, but AI can be hugely helpful—especially when used by teachers rather than students.1

The new tool, called The Evidence Checker, was devised by two science-of-learning researchers at the University of Toronto, Nidhi Sachdeva and Jim Hewitt, who publish an excellent Substack newsletter called, appropriately, The Science of Learning. Their goal is to communicate research findings in accessible prose and connect them to practical recommendations for classroom practice.

In a recent post introducing The Evidence Checker, Hewitt and Sachdeva echo Griffith’s observations about the barrage of conflicting information to which teachers are subjected. They offer examples of questions teachers may be seeking clear, reliable answers to, like:

What does the evidence say about inquiry-based learning?

Will regular math quizzes in my grade 5 class increase student anxiety?

Of course, teachers could just pop those questions into ChatGPT or one of the other readily available AI models. What The Evidence Checker provides, essentially, is a well-designed prompt that tells the model in no uncertain terms what it is supposed to do.

“You Are a Research Assistant”.

First, it tells the model who it is: “an educational research assistant” who is supposed to ...