Early last year, DOGE axed 90 percent of the staff and $900 million worth of contracts at the U.S. Department of Education’s Institute of Education Sciences. The draconian cuts raised a lot of eyebrows and some legitimate concerns but also offered a once-in-a-generation opportunity to rethink the federal agency charged with investing in education research and collecting vital education statistics.

Few people have spent more time rethinking IES’ role over the past year than Amber Northern, the senior vice president for research at the Thomas B. Fordham Institute. U.S. Secretary of Education Linda McMahon tapped Northern to serve as a special adviser tasked with reimagining IES.

This spring, the Department of Education released Northern’s report, “Reimagining the Institute of Education Sciences,” which sketches a path forward for the agency. Last week, in an AEI webinar, I had the chance to get her take on the state of IES and the future of federal education research. She outlined a number of her suggestions for the agency. While her report focuses on the extensive changes that can be made right now, she also got into discussing some specific reforms she’d like to see Congress make to the law that governs IES.

For readers who can spare an hour to watch the full conversation, I’d encourage you to do so. For those who can’t, it’s worth highlighting a few of Northern’s key points.

On what IES has been doing well, Northern aptly credits founding director Russ Whitehurst for battling to bring rigor into education research. It’s a well-deserved hat tip. Whitehurst took a slew of bullets as he fought to establish some beachheads of serious research in a field long dominated by dreck. But Northern is critical of what she sees as a lack of focus at the agency, saying, “We’ve had 15 priorities at IES at any one time. It’s been all over the place.” She describes an agency that is spread thin across a sprawling assortment of studies and data collections and that isn’t doing enough to provide useful findings to teachers, parents, and policymakers. She thinks IES has been “glossing over the importance of conveying information to the people it’s supposed to serve.”

Asked what a leaner, more effective IES entails, Northern sketches six proposed shifts, starting with a much tighter research focus. IES would concentrate on a handful of pressing challenges states are wrestling with, such as early literacy. The emphasis would shift to multistate grants designed to generate large-scale studies. She’d also like IES to rethink how it communicates findings—more tools, graphics, and practical resources and fewer unread academic papers.

Northern argues that federal data collection is critical but needs to be revamped and made timelier. “If we’re asking 28 questions but only using two, why are we still asking the other 26?” She would streamline and focus data collection on core functions like the National Assessment of Educational Progress. She also calls for overhauling the What Works Clearinghouse. The WWC isn’t even machine-readable—meaning it’s walled off from the online sources teachers rely on. Northern argues its findings need to be made accessible through the digital tools educators use.

All these recommendations can be pursued within existing law. Given that, I asked whether there are other changes worth pursuing if Congress were to take up the Education Sciences Reform Act, which created IES back in 2002. One such change, she says, would be to designate IES a stand-alone “microagency” with its own budget appropriation and independent hiring authority. Otherwise, she worries about threats to IES autonomy. For example, she sees a risk that a downsized IES could be “swallowed up” by a larger federal agency like the Department of Labor through an interagency agreement. As she puts it, “IES is not a program. It’s a statistical agency with authority to deliver the best evidence we can to support teaching and learning.”

Asked whether we still need a federal research agency given the Trump administration’s emphasis on returning education to the states, Northern answers yes. Her reasoning is twofold. Many states, she says, lack the capacity to conduct rigorous research on their own. Moreover, she’s skeptical that shipping research responsibilities out of Washington would depoliticize the work or yield more reliable data. “Let’s not kid ourselves,” she says. “The state level is no less political than the federal level.”

On the suggestion that IES has funded too many low-quality studies or been too tolerant of a field with inconsistent rigor and reliability, Northern pushes back. She argues that most IES-funded work consists of randomized controlled trials and quasi-experimental designs. The real problem, she says, isn’t the quality of the work but IES’ failure to communicate its value. She invokes the motto of the TED talks—"Ideas Change Everything"—as a model for forging a public identity around a body of knowledge. Regarding concerns about ideological tilt in ed. research, she puts it bluntly: “IES’s job is not to offset the bias of the field.”

I also asked Northern for her thoughts on IES’ role surrounding AI and ed tech. She argues that IES should avoid becoming an arbiter of which classroom tools are “approved” or “safe,” warning that such a role would inevitably invite lobbying and politicized disputes. Instead, she believes the agency should focus on creating clear benchmarks and evaluation standards for AI-powered tools—and let educators and the market make their own judgments.

Northern has thought deeply about the future of IES and offers a vision for where the agency goes from here. Some of what she has to say I find compelling, some less so. Agree or not, her suggestions deserve a thoughtful hearing. Those planning the future of IES would do well to read her report.