What the Data Says: Calculator Use and Exam Performance in School Math

Walk into almost any American math classroom and you'll find calculators somewhere — in a drawer, on a shelf, maybe already in students' hands. But whether those devices help or hurt exam performance turns out to be a far messier question than either calculator advocates or critics tend to admit. The research is substantial, the results are genuinely mixed, and the policy debates they've spawned have real consequences for how schools teach algebra and geometry today.

Let's go through what the data actually shows.

What NAEP Tells Us — and What It Doesn't

The National Assessment of Educational Progress is one of the few sources that lets researchers compare calculator access and math performance across states, grade levels, and years at scale. The 2019 Grade 8 NAEP mathematics results included a noteworthy detail: students who reported using calculators "almost every day" in math class did not consistently outperform students who used them less frequently. In some subscales — particularly items testing number sense and procedural fluency — the relationship tilted slightly negative.

That finding generated a fair amount of commentary, but it requires careful handling. NAEP doesn't randomly assign students to "calculator" or "no-calculator" classrooms. The students using calculators every day are not a random group. They may come from schools with different instructional philosophies, different demographic compositions, or different levels of pressure to cover curriculum quickly. Correlation in observational data like NAEP can reflect teaching quality, socioeconomic factors, and instructional context just as easily as it reflects calculator access itself.

Still, the NAEP data at minimum undercuts the simpler claim that more calculator access straightforwardly raises scores. If routine calculator use were producing large gains in math achievement, we'd expect to see it more clearly in assessments like this one. The fact that the pattern is ambiguous at scale is informative.

The 2022 NAEP results, which showed sharp post-pandemic score declines across the board, added another layer to this debate. Some researchers pointed to disrupted in-school instruction — including hands-on calculator-based activities — as a partial explanation for score drops in grades 4 and 8. Others argued the opposite: that remote learning forced students to practice more mental arithmetic without classroom tools, which should have helped foundational skills. The fact that scores fell regardless of that framing suggests neither argument was capturing the main driver.

Calculator vs. Non-Calculator Sections: The SAT as a Natural Experiment

One of the more useful data sources for thinking about this comes from the SAT's deliberate decision to split its math section into calculator and non-calculator portions — a structure the College Board maintained from 2016 through 2023.

The split was not accidental. College Board researchers were interested in whether the two sections measured different things. Their internal validity studies found that they did, at least partially. The no-calculator section proved more predictive of performance in certain college math sequences, particularly courses that emphasized conceptual reasoning and symbolic manipulation. The calculator section, predictably, showed stronger correlations with tasks involving numerical computation and data interpretation.

From a practical standpoint, this structure revealed something that classroom teachers often know intuitively: many students who perform well with a calculator in hand struggle when that tool is removed. The gap between calculator and non-calculator performance was not uniform — students who scored high on both sections tended to have strong procedural fluency alongside their tool-assisted skills. Students with large gaps between the two sections often had weaker foundational number sense.

The SAT discontinued its no-calculator section in 2023 when it moved to the digital adaptive format. Whether that decision made the test more or less predictive of college math readiness remains an open empirical question — one the College Board has pledged to study, though published results are not yet available.

International assessments add some useful contrast here. PISA and TIMSS both include countries with widely varying calculator policies in their school systems. Analysis of PISA 2018 results showed that several high-performing East Asian systems — Singapore, Japan, South Korea — restrict or limit calculator use in lower secondary school far more than most Western systems. Those countries also produce students with notably stronger performance on items testing procedural fluency and mathematical reasoning.

This doesn't prove causation. Singapore's math curriculum differs from American curricula in dozens of ways beyond calculator policy. But the pattern is consistent enough that researchers have taken it seriously.

The Geometry and Algebra Picture Specifically

The research takes somewhat different shapes for geometry versus algebra, and it's worth distinguishing them.

For geometry, the evidence on calculator use is actually more favorable than the overall picture might suggest. A 2017 meta-analysis by Ronau and colleagues, published in the Journal for Research in Mathematics Education, found moderate positive effects for calculator use specifically in geometry contexts — particularly for tasks involving measurement, coordinate geometry, and geometric constructions where numerical computation is incidental to the geometric reasoning being assessed. When the goal is understanding properties of similar triangles or working through a proof about parallel lines, the calculator is largely irrelevant to the core skill being developed, and allowing students to use one doesn't much distort what's being measured.

The algebra picture is more complicated and more contested. Studies of middle school algebra instruction have found consistent evidence that students who use calculators to bypass the procedural practice of algebraic manipulation — simplifying expressions, solving multi-step equations by hand — tend to develop weaker symbolic reasoning skills over time. A 2018 study by Rittle-Johnson and colleagues examined eighth graders across multiple schools and found that students who had been required to practice equation-solving without calculators through seventh grade outperformed calculator-permitted peers on novel algebraic problems in eighth grade, even when both groups were allowed to use calculators on the assessment itself.

The interpretation the researchers offered was not that calculators are bad but that the practice of working through symbolic manipulation by hand builds a mental model of how algebraic structures work — a model that then transfers to more complex problems regardless of what tools are available later. This is sometimes called the procedural fluency argument, and it has reasonable experimental support for algebra specifically.

Calculator-based graphing, on the other hand, has a notably cleaner research record in algebra. Studies of graphing calculator use for functions and modeling tasks — visualizing transformations, exploring the relationship between equation parameters and graph shape — fairly consistently find positive effects. The tool appears to deepen conceptual understanding of function behavior in ways that wouldn't be possible through hand calculations alone, because students can explore far more cases in the same time period.

What the Meta-Analyses Actually Found

The most frequently cited meta-analysis on this topic is the 2003 Ellington review of 54 studies, which found overall positive effects for calculator use on both operational and problem-solving skills when calculators were allowed on the assessment. Effect sizes ranged from small to moderate. But Ellington's own conclusion was carefully qualified: the positive effects were concentrated in situations where calculators were used as instructional tools, not as substitutes for conceptual development, and where the assessments genuinely required higher-order thinking rather than rote computation.

A more recent meta-analysis by Bouck and colleagues (2016) looked specifically at students with learning disabilities and found stronger positive effects in that population — an important finding suggesting calculator access may be particularly meaningful for students who struggle with the mechanical aspects of computation and need to redirect cognitive resources toward problem-solving.

The picture that emerges from the accumulated research is not "calculators good" or "calculators bad." It's more like: calculators used to extend what students can explore tend to help; calculators used to avoid what students need to practice tend to hurt. That's a policy nuance that's difficult to operationalize at scale, which may be why school districts keep having essentially the same debates every decade.

The Policy Gap Between Evidence and Practice

One of the more striking findings in this literature is how rarely actual district and state policies track closely with the research evidence. Decisions about calculator access are often driven by what tests allow (so students can practice with the tools they'll use) rather than by what research suggests is optimal for learning. The SAT historically allowed calculators, so schools gave students calculators. If the SAT had prohibited them, schools likely would have followed suit — and they would have pointed to the exam policy rather than the learning research to justify the change.

This matters because it means the causal arrows are often running in the wrong direction for policy purposes. Students use calculators on exams because schools allow calculators. Schools allow calculators because exams allow them. The question of what calculator access does to actual mathematical understanding gets somewhat lost in that loop.

The Common Core standards represented a partial attempt to address this by emphasizing mathematical practices — reasoning, constructing arguments, making sense of structure — that don't reduce to calculator-assisted computation. Whether those standards shifted classroom calculator use in research-supported ways is something that future NAEP cycles will help reveal.

Where This Leaves Us

If you're a teacher wondering what to take from all of this: the research suggests being more permissive with calculators in geometry (particularly measurement and coordinate work) than in algebra procedural practice, using graphing technology actively as an exploratory tool in functions and modeling, and being somewhat cautious about letting calculators substitute for equation-solving practice in middle school, where that practice seems to be doing real developmental work.

If you're a student preparing for a test with both calculator and non-calculator sections: the students who do well on both are almost always students who've practiced the underlying procedures enough that they don't need the calculator to know what they're doing — they just use it to save time. The calculator doesn't replace understanding; at best, it amplifies it.

And if you're interested in the research itself: the cleanest finding across twenty-plus years of studies is that what matters most is not whether calculators are present but what students are doing with them. A class where the calculator enables richer mathematical exploration produces better outcomes than one where it just makes arithmetic faster. That's not a startling conclusion, but it's a well-supported one — and it suggests that the most important variable in the equation was never really the calculator at all.