Scores from a major international test of 15-year-olds, released on Tuesday, showed a steep decline in United States’ teenagers’ reading abilities—and have already prompted education leaders to dissect potential causes and speculate on how to turn the trend around.
“What do we do now?” asked Matthew Soldner, the acting commissioner of the National Center for Education Statistics, part of the U.S. Department of Education, in a webinar on the results on Tuesday.
“Follow the evidence,” he said. “Follow the evidence of what works, for whom, and under what conditions.”
U.S. students’ reading scores on the Program for International Student Assessment, or PISA, fell 14 points, a drop that translates to six or seven months of learning. Science and math scores were statistically similar to those in 2022.
The United States still ranks above average in reading and science on the test, administered by the Organization for Economic Cooperation and Development. And the reading slide is far from a uniquely American phenomenon; other industrialized nations saw a similar sharp fall, on average.
But the United States did stand out in one respect: It has the largest gap between its highest- and lowest-performing students in reading, a gap that’s also been documented on national tests at other grade levels.
“The U.S. combines excellence and inequality,” said Andreas Schleicher, the director for education and skills for the OECD, during the webinar.
The country produces a large share of the highest-performing students globally across subjects, but its lowest-performing students fall below the OECD average across subjects.
Read on for three takeaways about the results from Tuesday’s webinar, co-hosted by NCES, OECD, and Vanderbilt University.
1. We don’t know why U.S. achievement gaps are growing—but theories abound
Multiple national and international tests showed the gaps between high and low performers in the United States narrowing until about 2012, when the trend reversed, said Sean Corcoran, a professor of public policy and education at Vanderbilt.
Understanding why should be a “major research and policy priority,” he said.
The answer likely lies at the intersection of several trends over time, rather than one sole factor, he continued. Research suggests that the federal accountability policies of the early 2000s raised student performance, especially for low-achieving students. Those policies loosened under President Barack Obama’s administration, which began issuing waivers for some federal accountability measures in the early 2010s. (The K-12 accountability law was rewritten in late 2015.)
The lingering effects of the 2008 recession could play a role, too, said Corcoran, as many years of depressed investments in education can have long-run effects.
Speaking on the webinar, U.S. Secretary of Education Linda McMahon said that increased flexibility for states and more options for school choice, both policy priorities of President Donald Trump’s administration, could help close the gap.
“In order to truly reform our education system, we cannot continue to box children in with one-size-fits-all mandates,” she said.
Research on the effects of school choice on student achievement is mixed, said Corcoran. It can improve outcomes, but is unlikely to reduce achievement gaps without accountability and standards for the schools from which families can select.
At least two states that have launched universal private school choice programs, Indiana and Iowa, require that participating students take the same state tests as public school students.
2. Declining reading scores are a global problem
Reading scores have fallen in many industrialized countries—not just the United States.
Teenagers are losing the most ground on tasks that require deep reading and analyzing of text, said Schleicher, skills such as remembering and integrating information, reflecting, and evaluating claims.
“In this world increasingly flooded with information, our students are less likely to think deeply and separate signal from noise,” he said.
It’s possible that social media use patterns and digital reading are contributing to this trend, he said.
Global patterns in PISA data show that limited or moderate use of classroom technology is related to better student reading outcomes, but more than three hours of instructional use per day is related to lower student performance, Schleicher said.
But the relationship looks different for leisure use of devices, such as smartphones, during the school day, he continued. There, the relationship is linear—more leisure use at school is linked with worse outcomes.
These findings should inform schools’ decisionmaking around screen time and the use of artificial intelligence in the classroom, said Jonathan Seiden, an assistant professor of early childhood policy at Vanderbilt.
“There’s a lot of hype around the idea that AI systems could be useful in solving this problem. The logic of this is really appealing,” he said. In theory, AI could target students’ individual needs, making it easier to close achievement gaps.
But it doesn’t always work that way in practice. Seiden cited a recent study from researchers at the University of Toronto that found middle school students provided with an AI tutor usually didn’t engage productively, regularly sending off-topic messages or asking for the AI tool to offer up the answers.
Educational technology is not a “magic bullet” that can solve big, complex problems, Seiden said.
3. Some of the most effective levers for raising scores may be low-tech
Across countries, two factors were consistently linked to higher student achievement, said Schleicher: the quality of relationships between students and teachers, and family support.
Students who felt their teachers respected them and cared about their well-being were more likely to score higher on PISA. The United States scored slightly above OECD average for student-teacher relationship quality.
Family support is another strong predictor of student success, Schleicher said, with students reporting their parents ask them questions about their schoolwork performing higher than students who say their parents don’t.
But these measures of family support have declined since 2022, both across the globe and in the United States, he said.
“Parents don’t need to know all the answers,” said Schleicher. “They need to show that the questions children are working on matter to them.”
