There’s a lot of buzz right now about the potential for the Institute of Education Sciences to finally get the resources and authority to support major breakthroughs in teaching and learning. Some of that is due to $30 million in funding included in the end-of-the-year spending package for fiscal year 2023 to support a stronger R & D infrastructure at IES. The big excitement, though, is around the NEED Act, a bill sponsored by Rep. Suzanne Bonamici (D-OR) and Rep. Brian Fitzpatrick (R-PA), and about to be reintroduced in the current Congress, which would create a National Center for Advanced Development in Education (NCADE)—or, as others are calling it, “ARPA-ED,” modeled after the Defense Department’s DARPA program. Here’s how IES director Mark Schneider describes its potential:
In the heart of every federal agency lies a dream of becoming the next DARPA—the folks that brought us advances like GPS and the internet. I want to bring you up to speed on our most recent efforts to create ARPA-ED and open a discussion with our stakeholders about where we will go from here as we pursue its creation.
While ARPA-ED won't create a new internet, there are many questions a DARPA-like unit might tackle: How will we identify the new literacy skills Americans need and develop them in learners across the lifespan? How do we even conceive of appropriate literacy and writing skills in a modern, AI-dominated world? How can we best harness innovations to create a personalized system of education so that instruction is tailored to the needs of learners? How can we use new technology to relieve the paperwork burden on teachers, freeing them to do what they do best—teach students?
I am proud to be a member of the Alliance for Learning Innovation, a coalition that advocates for ARPA-ED and similar initiatives, and I certainly hope that IES will receive a Congressional charter to do more to balance its impressive research agenda with an equally impressive development one. And I hope Director Schneider is right that an ARPA-ED could lead to a wave of instructional innovations. (Fordham and the Center for American Progress highlighted what some of those might look like through our Moonshot for Kids initiative in pre-pandemic times.)
But beyond all the razzle and dazzle, let me admit that I am desperate for something even more fundamental, starting with basic information about what the heck is going on in America’s classrooms. We arguably know less about the typical American school than about the dark side of the moon. Indeed, what we desperately need is a sort of Mars Rover for our classrooms.
More on that in a bit. But first, contemplate just how little information our policymakers and education leaders can tap about the impact of their decisions in the real world of teachers and students, desks, and books.
Take a simple example. Like many states, Ohio is on the verge of significantly strengthening its literacy policies to align them with the science of reading. The hope is to improve teaching and learning in the subject, in a way that will eventually be apparent via higher student achievement and greater student success.
How will we know if that new policy package works? Most likely, we won’t. Yes, policymakers might commission an evaluation, and we will be able to see which reading programs districts and charter schools say they have adopted, and we can gauge how well those programs are aligned with the science of reading. But we won’t have any trustworthy information about whether those adoptions are embraced and faithfully implemented by individual classroom teachers, nor whether those teachers get the support and ongoing professional development they need to shift their practice. (Perhaps we could survey teachers about this, but self-reports are famously unreliable.) We’ll know almost nothing about whether this rubber actually hits the road.
Someday, to be sure, we’ll be able to look at test scores, but we will have a hard time knowing if the policy reforms can explain any changes in student achievement. That is always a challenge, what with so many confounding factors in education, but the pandemic has made it more so. If reading scores for young Ohio students continue to slide over the next several years, will that be a sign that the reading policy failed? Or will it simply be because lots of today’s elementary school students missed out on preschool, as well as key developmental milestones during the pandemic? On the other hand, if test scores go up, will that mean that the new reading policy was a success? Or would that have happened anyway, as post-pandemic students enter the system?
What will be missing from a look back at Ohio’s reading reforms, as with any evaluation-type study in education, is the most important part. The middle stuff between policy enactment and its impacts. In other words, what actually changes (or not) in the classroom, in terms of teaching and learning.
In the case of Ohio, it would be great to know, ideally for every classroom, or at least for a representative sample: Which reading programs do teachers stop using and start using after enactment of the new reading law? Did those teachers thoroughly implement the new programs? Did they get skilled support in training? How much time did students spend with the new materials? What about interventions for students who were struggling? How frequent were those, and what was going on in them? Which programs were used for the interventions? Were they different than what teachers had used before? How adept were the interveners? Did teachers have students use any online programs to help support their reading? Which ones? How often and for how long? What else was happening in the classroom and around the school that might have impacted students’ achievement, experience, and engagement? For example, are schools spending more, less, or the same amount of time on English language arts than before the policy shift?
Those are critical questions, but are very difficult to answer because they call for nuanced and precise data collection. In the old days, researchers would hope for a multi-million-dollar federal grant to be able to send out graduate students to sit in the back of classrooms and collect such information by hand, systematically “coding” what they saw. Even then, it was exorbitantly expensive and extremely time-consuming, so inevitably you could only visit a relative handful of schools and classrooms.
Solving this evaluation challenge is a tough problem—precisely the type of problem that DARPA and its clones are supposed to tackle.
So enter my Mars Rover idea. (Be warned: This is going to sound far-fetched!) Imagine that we could design a small robot that could be sent into a classroom for a week or a month or even a whole school year to collect all of this information and more. It would be tricked out with high-tech gizmos that would collect data and beam it safely to the cloud. Then AI would make sense of those data, giving scholars, leaders, and policymakers an accurate, comprehensive picture of what was happening in our schools.
First and foremost, our rover would record everything going on in the classroom. As I’ve written before, video would be most powerful, but audio could work if that made people more comfortable in terms of student and teacher privacy.
Given that modern classrooms tend to have lots of movement and activity, the rover might have a sensor that could make sense of it all—for example, tracking the composition of student groups, and how much time the teacher spent with each one. Were those groups reshuffled frequently, as students made progress? Or were low-achieving students getting stuck in slow-moving instruction?
The rover should also have a little scanner built into it so every night teachers could insert any paperwork they had students complete, along with their own feedback and grades, so researchers would get some insight into what kind of work students were being asked to do, and the standards that teachers were holding that work to.
Since many assignments and grading (and even instruction) happens online now, our rover would also have the ability to track any digital activity of students in the classroom, seeing, for example, which programs they were using, how engaged they were, and the dosage.
To be sure, designing such a marvelous rover would be no small feat. Yet the toughest nut to crack would likely be political: How to get educators and families to buy in? Strong privacy protections would go a long way, but teachers especially would need to get something out of participating in a rover-assisted evaluation. Perhaps they could get real-time feedback about their instructional practices and/or grading standards. Or maybe they simply need to be paid for their involvement.
With the blizzard of information generated by our rover—tied to particular teachers, students, and schools, and then connected to other data from test scores, student surveys, and beyond, and analyzed by AI—scholars, leaders, and policymakers would actually have some insight into what’s going on in our schools and whether their policies and decisions were making things better.
This is, of course, how most modern organizations already operate, with a deluge of data, allowing them to fine-tune and adjust and continuously improve. As far as I can tell, flying blind is something that has been relegated just to our education system.
I’m just a policy wonk, and something of a dreamer, so I’m sure there are many reasons that make what I have described here technically challenging. Even with the use of artificial intelligence and high-tech sensors, it might be impossible for anyone or anything to make sense of what the heck is going on in an American classroom, given the cacophony that is so typical.
Then again, a lot of people were skeptical that we could land a rover on Mars.