State of the Student 2025
How We Got Here
“[System leaders] need better insights on what caused our depression [in education performance] and better ideas for reinvigorating instruction. We’ve all got a role to play.”
Tim Daly, Education Next
America’s math problem is complex, but its warning signs have been visible for years. Our review of the evidence suggests that at least four policy trends are at play in the math declines:
“Both sides of the argument sometimes act as though there is no room for the other one in the same classroom, but this is simply not true.”
Kristen Smith, math instructional coach, in Edutopia
The Math Wars
The math wars have been simmering for a century.
At the heart of the conflict are two visions of how math should be taught. On one side are “traditionalists” who emphasize math facts and fluency, and support teaching math procedures directly. In a lesson, teachers might show students how to add ½ and ¼ on the board, then give them a worksheet with 20 similar problems to practice.
On the other side are “progressives” who emphasize a deeper understanding of mathematical ideas and support teaching students through problem-solving. Here, teachers might give students measuring cups and recipes, then ask them to figure out how much rice they need for two recipes requiring ½ and ¼ cups of rice. The debate is still alive and well in American classrooms today: in a 2024 EdWeek poll, math teachers were almost evenly split.
The evidence shows that an either/or framing is misleading. Students need both conceptual understanding and procedural fluency, and—crucially—the two support each other. There’s no evidence to show that teaching either procedures or problem-solving is exclusively “best.” According to a summary of a recent peer-reviewed article:
“Children learn most effectively when instruction follows an evidence‑based cycle: grounding facts in conceptual understanding, using brief timed practice to make those facts automatic, then returning to discussion and reflection to deepen that knowledge.”
A National Mathematics Advisory Panel sponsored by the US Department of Education came to the same conclusion in 2008:
High-quality research does not support the contention that instruction should be either entirely “student-centered” or “teacher-directed.” Research indicates that some forms of particular instructional practices can have a positive impact under specified conditions.
Nevertheless, researchers on both sides elevate their own evidence and discredit the other side. “This is more like a religious disagreement,” policy expert Tom Loveless said. The polarization leaves little room for the balanced and specific guidance that teachers need to ensure that every student learns math.
Strong evidence for explicit instruction, in which teachers directly explain math concepts and procedures while also supporting deep thinking and student engagement, also exists. Explicit instruction is effective for all students, and it’s essential for struggling students, whose math scores have fallen the most. But explicit instruction also aligns with ideas from the traditionalist side of the math wars, making it controversial when it shouldn’t be. Advocates for progressive education worry that too much explicit instruction will stifle students’ curiosity and appetite for deeper math understanding, so they dismiss it as an approach that should only be a last resort to patch skill gaps. However, according to special education and math researcher Amelia Malone, “These are not ‘remedial’ practices—they are hallmarks of excellent instruction.”
The disagreement isn’t just among researchers, since many organizations that practitioners look to for guidance and curricula also lean toward one side or the other. Educators receive different guidance depending on which organizations they look to, which leads their instructional approaches to skew either traditionalist or progressive. No major math education organization is delivering unbiased guidance and helping educators navigate the confusion. In a recent example, the National Council of Teachers of Mathematics issued a joint statement with the Council for Exceptional Children that advised teachers on best practices for teaching math to students with disabilities, but left out high-evidence strategies like explicit teaching. The statement appears to have been influenced by progressive views on how to teach math, despite attempting to draw on all the evidence that exists.
The lack of unbiased guidance leaves teachers and school systems adrift, and has likely played a part in why math achievement gaps have grown since 2013. Although the Common Core State Standards call for a balanced approach to math instruction, they were often implemented (starting around 2012) with a bias toward conceptual understanding and more progressive teaching methods. While evaluating Common Core is nearly impossible due to varied implementation, some evidence suggests that it widened achievement gaps (see sidebar).
Moreover, widely used curriculum review tools like EdReports focus heavily on whether materials align with Common Core standards, but don’t evaluate the strength of the instructional strategies themselves. As a result, some widely used curricula—like Singapore Math—don’t pass muster for standards alignment despite having strong research backing, while others—like Illustrative Mathematics—are top-rated but still subject to concerns about inadequate practice time and supports for diverse learners. Meanwhile, according to an EdWeek Marketplace study, only 4% of administrators reported using the “science of math” (or the most substantial evidence about how to teach math) to guide their curriculum decisions. There is an urgent need for clarity among district decision-makers about what amounts to effective math instruction—standards-aligned or not.
How the math wars played out in Common Core implementation
Around 2010, many districts seized the opportunity to modernize math instruction through the Common Core State Standards. The standards themselves called for teaching a balance of conceptual understanding and procedural skill, and they emphasized accuracy and recall. But in their rollout, experts emphasized conceptual understanding far more, and at the expense of procedural practice and recall. Teachers reported that they were assigning far fewer practice problems and asking students to explain their answers in multiple ways instead of relying on standard procedures. Many teachers were still learning the new approaches themselves, while parents, confused by unfamiliar strategies and diagrams in children’s workbooks, were left unable to help.
For students already behind, such shifts may have generated more confusion and delivered less of the practice and explicit instruction they needed. Research has shown small positive effects from Common Core on student achievement overall, but not for economically disadvantaged students struggling in math or for students with disabilities. The result is that an ambitious effort to raise academic expectations may have inadvertently contributed to widening achievement gaps. Teachers needed—and still need—more explicit instructional guidance that prioritizes evidence, not ideology, on what’s best for students struggling in math.
The math wars echo the “reading wars” of past decades, but resolving them may be more difficult. The science of reading established that systematic phonics and phonemic awareness instruction were an essential component of effective reading instruction, and discredited ineffective methods like three-cueing. In math, progressives malign high-evidence strategies like explicit teaching—similar to phonics before the science of reading revolution—but unlike in reading, there’s no widely implemented “wrong” approach taking their place.
The question is not which side in the math wars is right and which is wrong, but what is the right combination of evidence-based instructional strategies—including explicit instruction—that will ensure every student will develop procedural skills and conceptual understanding. It’s as if the two sides are arguing over sun exposure, with one side only pointing to skin cancer risks and the other only highlighting the benefits of Vitamin D.
Weakened Accountability and Lowered Expectations
“Why would anyone want to lower standards? It’s like telling everyone that they’re a great swimmer when you know half of us really are drowning.”
Pete Shulman, former deputy commissioner at the New Jersey Department of Education, in Free Press (2025)
Over the past three decades, US education policy has swung from pursuing strong federal accountability to deference to states. The result is a post-pandemic era of lowered expectations, inflated grades, and obscured learning gaps.
The mid-90s saw the advent of the standards movement, in which most states—led by governors and business leaders—passed ambitious legislation to bring a greater focus on teacher quality and outcomes in public education. There were also efforts to increase school-level flexibility and supports—accompanied by heightened school-level accountability for results. At the same time, the charter school movement was promoting the idea of higher expectations and greater choice, especially for disadvantaged students.
The 2002 No Child Left Behind (NCLB) Act delivered a shock to public education through muscular federal accountability mandates. The law set the ambitious target of 100% student proficiency by 2013–14 and required schools to show “adequate yearly progress” for all students and subgroups. Research shows that NCLB drove gains in math, particularly for disadvantaged students and in states with previously weak systems. Around the same time, national foundations made significant investments in charter school expansion, especially in inner cities, and in alternative teacher preparation programs and policies.
The era of high expectations and accountability for results, however, had its fair share of unintended consequences and political backlash. Some states lowered standards to show better proficiency levels, and some schools concentrated support narrowly on “bubble students” near proficiency cutoffs. Teachers’ unions saw charter schools and teacher quality initiatives as an existential threat. Parents, especially in the suburbs, got tired of the emphasis on testing.
The Covid-19 pandemic was the nail in the coffin for accountability. Still, states had for years been walking away from commitment to related reforms, including Common Core State Standards, charter schools, and provisions to ensure teacher quality. The timeline of widening achievement gaps and declining math scores roughly corresponds to the ebbing commitment to high standards and accountability.
In the years since the start of the pandemic, states have been lowering the threshold for proficiency on exams and ditching tests required for high school graduation. Democrat-led New York and Republican-led Oklahoma both reported improved proficiency rates in 2024 after lowering proficiency levels in 2023. In 2013, 24 states had a graduation exam or exit test; by 2025, only six did. The Trump administration has shown little commitment to a muscular federal role in accountability, cutting back on oversight and signaling a desire to “return” power to states.
In the same climate of weakened state accountability, grade inflation has soared. In Washington State, students’ average math GPA jumped 0.34 points from 2019 to 2021—triple the increase of the prior eight years. In North Carolina, math proficiency dropped 11 points while A and B grades declined by only 3 points. A national study from 2021 to 2023 found that 57% of grades didn’t align with student knowledge as measured by tests, and two-thirds of those misaligned grades were inflated, most often for underserved groups. ACT data show rising GPAs, especially in math, despite falling test scores. By 2021, even students scoring in the 25th percentile were graduating with B averages or better.
Grade inflation has consequences. It masks fundamental skill gaps, misleads families, and can lower student motivation by suggesting that a decent grade is attainable without real effort. It also creates what one study called an “urgency gap.” If families think their children are doing fine, they may be less likely to seek help, contributing to low participation in academic recovery efforts. Schools may miss key intervention moments, misallocate resources, or act too slowly to address learning gaps. Educators, in turn, inherit students with inflated transcripts but real skill deficits.
Declining Access to Qualified Teachers
Teacher quality matters. But access to qualified math teachers has always been unequal, and it has worsened in recent years due to a shrinking pipeline of new teachers, especially those with a math teaching specialty.
Today, teacher preparation programs are producing only 77% of the overall graduates they did just over a decade ago. And fewer preservice teachers are specializing in math. As Figure 12 shows, the number of graduates prepared to teach math declined by 36% between 2013 and 2020, with a 45% decline in traditional preparation programs alone.
“There is a persistent misconception that anyone with a high school diploma has the requisite math knowledge to teach elementary math. In reality, teaching elementary math requires both a conceptual understanding of foundational mathematics and pedagogical knowledge of how to teach the concepts.”
NCTQ, April 2025
US schools also report significant challenges in filling math vacancies. A Gallup poll from late 2024 found that 82% of all K–12 leaders said it’s challenging to hire qualified math teachers, and half said it’s “very challenging.” Based on the NCES School Pulse Panel in August 2024, 28% of public schools had at least one math vacancy before the start of the school year; 15% had two. Of those, 61% reported difficulty finding a fully certified math teacher to fill the position. Vacancies are highest in schools serving mostly low-income students, enrolling mostly students of color, and in urban areas. This is likely one driver of the widening chasm between the highest math performers and the most struggling students.
Even when teaching positions are filled, teachers are not always fully credentialed to teach the subject. Roughly 11 to 12% of teachers are not fully certified for their assignments, which includes teachers with emergency credentials and those who are teaching “out of field” in a subject they’re not certified in. Over half of middle school math teachers lack a math or math education degree.
And even when teachers are credentialed, new research shows that they may not have been well-prepared to teach math. The National Council on Teacher Quality examined over 1,100 elementary teacher-prep programs and found that they do not allocate adequate time for math content learning. American teachers in lower grades also have limited mathematical content knowledge compared to international peer countries.
Teachers may also struggle with math anxiety, which affects many adults and especially women, who make up nearly 90% of elementary school teachers. Some reporting has even suggested that a lack of confidence in math is part of what leads some teachers to seek elementary teaching positions. Since students can “catch” math anxiety, its prevalence in early classrooms means many students have the deck stacked against them. Improving teachers’ math skills, however, could prevent them from developing math anxiety.
There is shockingly little effective state leadership for ensuring that all students can experience effective math instruction. In a recent report, NCTQ found that only one state—Alabama—has “Strong” policies for improving math teacher effectiveness, which include auditing teacher preparation programs for robust math instruction and requiring districts to support teachers to implement high-quality math curricula. Seven states received “Unacceptable” scores.
States must strengthen their policies for ensuring math teacher effectiveness. But even filling every school with qualified math teachers will not solve the problem. In a system where most teachers single-handedly instruct groups of 30 students, the growing variation in students’ skills and needs is too much for even the most qualified math teachers to address.
A Rigid and Outdated Delivery System
Most American classrooms adhere to some basic assumptions.
One teacher will lead a classroom. Students in each class should all learn the same material at the same time. And students who need additional support to succeed in class will receive it outside. In reality, this model works only when all or most students are entering the class with the prerequisite skills for the class and require little or no special attention from the teacher. This is rarely the case with math.
“The delivery system in mathematics education—the system that translates mathematical knowledge into value and ability for the next generation—is broken and must be fixed. This is not a conclusion about any single element of the system. It is about how the many parts do not now work together to achieve a result worthy of this country’s values and ambitions.”
—US Department of Education National Mathematics
Typical classrooms leave educators with an impossible task. NBC News reported on a South Carolina math specialist who was scrambling to support teachers who needed to create twice as many ability groups as usual to ensure that every student received instruction on their level: “Panic ensued,” she said, as teachers threw up their arms and declared, “There’s no possible way I’m going to be able to do this!” The trade-offs are daunting: either include a wider range of students in each ability group—meaning some students will not be receiving instruction targeted to their precise needs—or stretch the same teacher across twice as many groups, diluting their time, attention, and energy and possibly their effectiveness.
Instruction marches on, whether or not students have mastered it. A growing movement seeks to ensure that all students have access to well-designed curricula that deliver rigorous content at grade level. But these curricula are designed to ensure that all students receive demanding grade-level content and do not account for the reality that many students didn’t master the precursor material in prior grades. They write: “As a result, math educators have the massive challenge of both teaching grade-level material and addressing students’ individual needs.”
The assumptions behind how most schools deliver instruction were never well-suited to how students actually learn.
Typical classrooms leave educators with an impossible task.
NBC News reported on a South Carolina math specialist who was scrambling to support teachers who needed to create twice as many ability groups as usual to ensure that every student received instruction on their level: “Panic ensued,” she said, as teachers threw up their arms and declared, “There’s no possible way I’m going to be able to do this!” The trade-offs are daunting: either include a wider range of students in each ability group—meaning some students will not be receiving instruction targeted to their precise needs—or stretch the same teacher across twice as many groups, diluting their time, attention, and energy and possibly their effectiveness.
Instruction marches on, whether or not students have mastered it.
A growing movement seeks to ensure that all students have access to well-designed curricula that deliver rigorous content at grade level. But these curricula are designed to ensure that all students receive demanding grade-level content and do not account for the reality that many students didn’t master the precursor material in prior grades. They write: “As a result, math educators have the massive challenge of both teaching grade-level material and addressing students’ individual needs.”
Siloed interventions compete with core instruction.
Because the same teacher cannot possibly deliver multiple types of individual interventions and core whole-class instruction simultaneously, students who need tutoring, language support, or special education services often find themselves pulled out of their main classes to receive support services—thus reducing their access to core instruction. This zero-sum approach, in which more access to core instruction means less individualized intervention and vice versa, ultimately shortchanges students. The result: the most marginalized students secure fewer opportunities to learn grade-level standards compared to more advantaged peers.
Ability tracking reigns.
A common but crude response to widely varied student skills is tracking, in which schools evaluate students for their learning and ability, then group them in different classes with peers at similar levels. This practice restricts opportunities for some and reinforces inequality, but can also ease the logistical challenges of supporting diverse needs, since it reduces the amount of variation in each classroom. “Detracking,” meanwhile, removes the barriers students face in accessing courses, but can wind up pushing some students into courses they’re not ready for and delaying or preventing other students from moving ahead.
And now recent trends point to rapidly increasing learner variability in classrooms: In earlier sections, we highlighted growing gaps between the highest- and lowest-performing students, dating back to 2013. Shifting demographics in many regions and states are resulting in a growing number of multilingual learners. And the number of students identified with disabilities has been growing for decades. Diagnosis rates for attention deficit hyperactivity disorder (ADHD) and autism are both on the rise. The share of students who receive services under the Individuals with Disabilities Education Act (IDEA) has reached 15 percent, with further increases projected in future years.
This growing variability leaves education systems with a stark choice: either rethink the standard assumptions about how schools are organized to deliver instruction, or run the risk of predictable academic downspiraling for students who “can’t keep up” while adding to the strain on teachers.
1/5
of all students are behind grade level by grade five, and over a third by grade eight, according to Curriculum Associates’ i-Ready tests.
Triage or transformation?
Math teachers want to provide a transformative learning experience for students, helping them both master complex material and develop a love for numbers and computational thinking. However, most end up triaging instead, making choices on who they can best help in the classroom and whom they must funnel into remediation tracks or special education.
A recent TNTP report found that the most effective approach to addressing learning gaps in Algebra I was a system of individualized interventions that prioritized critical precursor skills and ensured that each student received the most challenging material they were ready to master.
That is a far cry from the current state of affairs, however. As the TNTP report concludes:
Unfortunately, this level of instructional coherence is rare. In many schools, Tier 2 materials only loosely connect to core classes, and assessments measure different things in each tier. Core teachers and specialists may not have time to meet or connect their lessons. As a result, the students who receive the most support typically have the most disjointed experiences at school.
The current delivery model is simply not designed to ensure that every student masters grade-level content. Teachers are, out of necessity, merely passing students on from grade to grade, whether or not they have mastered key concepts. Students may receive good grades in the class, leaving parents to think all is well. Then, in the absence of a coherent system, those students’ future teachers have to triage as best they can: help the students closest to grade level, give struggling students the simplest material just to keep them stable, and hope the rest can stay afloat with minimal intervention. Teachers care deeply about student success, but the current delivery system prevents them from stewarding each student’s math mastery.
This is a fundamentally broken approach—and it must stop.
Teachers should not be triaging in Algebra I. It’s an impossible task, and too many students are already too far behind at that point. Rather than triaging, schools should practice “preventative medicine” in grades K–7. If a student is failing to grasp a math concept or misses a week of instruction, they should receive immediate help. A substantial body of research suggests more effective instructional methods, such as:
- individualized tutoring integrated with core instruction;
- extended time for math instruction during the school day;
- math placement decisions based on student competencies, not rigid tracking;
- and identifying and addressing critical precursor skills in early grades before they compound.
In addition, emerging solutions such as AI-powered individualized learning and tutoring hold significant promise, though the evidence base is still small and the tools themselves are still developing.
An important question is whether more money is the answer. In some cases, it may be. However, tracking, uneven expectations, and lack of attention to true mastery exist in the highest-resourced schools as well as the lowest. The challenge is rethinking schools’ use of time, talent, data, expectations, and incentives so that no student ever falls behind.
Transformative math instruction is something every child deserves. In the next section, we propose how it can happen and what it will take.
Can AI revolutionize math education?
While AI platforms, in particular large language models like GPT-4, demonstrate strengths in algebra and arithmetic reasoning, they struggle with more complex and visual-spatial problems. Despite limitations, AI-powered instructional tools, such as Khan Academy’s Khanmigo and DreamBox, are already attempting to personalize math instruction, diagnose misconceptions, and significantly boost math outcomes, especially among disadvantaged students.
However, key considerations remain about AI’s effective integration into mathematics education. Educators have cautioned about the potential pitfalls of students using AI to shortcut learning rather than fostering deep conceptual understanding. And while the evidence base supporting AI’s efficacy in mathematics instruction is growing, many AI tools still require robust research, including randomized controlled trials, to confirm their educational impacts fully. For example, promising early pilots like Khanmigo and Saga AI Tutor have yet to be conclusively validated through rigorous studies.
Looking forward, rapid advancements in AI, including its use in systems that combine language reasoning power with precise computational capabilities and improved image recognition, promise more accurate and visually integrated math instruction. And AI’s potential extends beyond instruction and assessment.
For example, AI might empower parents to closely track their students’ progress and alert them to any missed foundational skills. It might allow educators to cut through ideology to bring a more evidence-based approach to helping struggling students. Advances in AI could enable students without access to advanced coursework to learn more independently. The possibilities are just beginning to take shape.
Unanswered questions about what’s driving math declines
In our research, we found several additional hypotheses that could be part of why math achievement has declined, but they lack sufficient evidence. Probing into these questions could help further illuminate what states and districts must change to achieve math success.
- Women and people of color have seen expanding professional opportunities in the past decades. Have schools lost out on a strong pipeline for math teachers in a “brain drain” to other professions with better pay and benefits?
- Charter school founder and policymaker Steven Wilson has argued that racial justice advocates who were critical of “no excuses” pedagogy led schools formerly seen as beacons of rising achievement to retreat from a culture of high expectations and adult accountability. How are schools striking the right balance of strategies to both affirm students’ identities and achieve math excellence?
- The science of learning is clear that well-being (e.g., basic needs, relationships, and relief from chronic stress) is essential for students’ readiness to learn. However, some research suggests that efforts to support student well-being post-pandemic have bordered on leniency. What kinds of student well-being practices in schools help or hurt math outcomes, and in which circumstances?
