What the Research Actually Says About Math Anxiety

"Math anxiety" is one of those phrases everyone uses and almost no one defines. Parents reach for it to explain a child who freezes at a worksheet. Tutors invoke it to explain a student who knows the method in practice but blanks in the exam. It gets used, loosely, to mean anything from mild dislike to genuine distress.

But math anxiety is a real, measured psychological construct with more than fifty years of research behind it — and the findings are more specific, more useful, and in places more surprising than the casual usage suggests. It isn't the same thing as being "bad at maths," and it isn't simply a confidence problem you can talk a child out of. The relationship between anxiety and achievement runs in a direction most people get exactly backwards half the time.

This piece tries to lay out what the academic literature actually shows, without the spin. It is written for parents trying to help an anxious child, for educators who want to think it through carefully, and for anyone who needs a reliable starting point rather than a motivational poster.

First, what math anxiety actually is — and isn't

The construct was formalised in 1972, when Richardson and Suinn published the Mathematics Anxiety Rating Scale (MARS) in the Journal of Counseling Psychology — a 98-item questionnaire asking people to rate their tension across everyday and academic maths situations. Their working definition has held up well: math anxiety is a feeling of tension and apprehension that interferes with the manipulation of numbers and the solving of mathematical problems, in both ordinary life and academic settings.

Two things in that definition matter. First, it names interference — anxiety that gets in the way of doing maths you might otherwise be able to do. Second, it spans ordinary life, not just exams. Shorter modern instruments exist — the 9-item Abbreviated Math Anxiety Scale (Hopko and colleagues, 2003) is widely used — and they consistently pick out the same two components: a cognitive one (worry, intrusive thoughts) and an affective-physiological one (dread, a racing heart, the wish to escape).

Crucially, math anxiety is statistically separable from general anxiety and from test anxiety, even though it correlates with both. A child can be calm about spelling tests and social situations and still experience real distress specifically around numbers. This is why "she's just an anxious kid" and "he just doesn't test well" are both incomplete explanations — math anxiety is its own thing.

One caveat up front: there is no agreed clinical cut-off for "high" math anxiety. It sits on a continuum, and every prevalence figure you will ever read depends entirely on where a given researcher drew the line. Keep that in mind for the next section.

How common is it?

Because there is no standard threshold, treat all prevalence numbers as illustrative rather than definitive. The most credible large-scale data comes from the OECD's PISA programme, which surveyed 15-year-olds across dozens of education systems.

In PISA 2012 (published by the OECD in 2013), averaged across OECD countries:

• 61% of students said they worry about getting poor marks in mathematics
• 59% often worry that maths classes will be difficult for them
• 33% get very tense doing maths homework
• 31% get very nervous doing maths problems
• 30% report feeling helpless when working a maths problem

Those are agreement rates on single survey items, not a diagnosis — but they establish that maths-related worry is common, not marginal. The same survey found that, within virtually every country, higher math anxiety was associated with lower maths performance — on the order of the difference a year of schooling makes. We'll come back to what that association does and doesn't prove.

The mechanism: why anxiety and maths genuinely interfere

This is the most important — and most counter-intuitive — part of the research, because it explains why "just relax and concentrate" is not merely unhelpful but slightly beside the point.

The dominant account is the working-memory explanation, developed principally by Mark Ashcraft and colleagues. Working memory is the mental scratchpad you use to hold and manipulate information in the moment — carrying a digit, tracking where you are in a multi-step problem, keeping a sub-total in mind. Ashcraft and Kirk's influential 2001 study in the Journal of Experimental Psychology: General showed that math anxiety is associated with reduced working-memory capacity — and, crucially, that the performance gap widens precisely on the problems that load working memory most, such as multi-digit addition that requires carrying. On easy, low-load problems, anxious and non-anxious people look similar. On demanding ones, the anxious mind is doing two things at once: the maths, and a running commentary of worry. The worry consumes the very resource the maths needs.

This dovetails with attentional control theory (Eysenck and colleagues, 2007), which holds that anxiety impairs the control functions of working memory — the ability to inhibit distractions and stay on task — even when raw ability is intact. The anxious student isn't less capable; they're running their processor with a background application draining it.

The counter-intuitive corollary comes from Beilock and Carr's 2005 study in Psychological Science on "choking under pressure." They found that, under exam-like pressure, it was the students with the highest working-memory capacity who choked most on demanding maths. The reason: high-working-memory people normally rely on sophisticated, memory-intensive strategies — and those are exactly the strategies pressure disrupts. Your most able child can be the one most undone by an anxious exam hall.

The neuroscience is consistent with all of this. Young, Wu and Menon (2012, Psychological Science) found that in seven-to-nine-year-olds, higher math anxiety was linked to heightened activity in the amygdala — the brain's threat-detection centre — and reduced engagement of problem-solving regions. Lyons and Beilock (2012, PLoS ONE) found something even more telling: in highly math-anxious adults, it was the anticipation of doing maths, not the maths itself, that activated brain regions associated with the experience of physical pain. The dread is doing much of the damage before a single sum is attempted.

Chicken or egg: does anxiety cause bad grades, or the reverse?

Here is where careful reading matters most, because the popular story ("anxiety wrecks performance") is at best half the picture.

There are three competing theories. The deficit theory says the arrow runs from poor performance to anxiety — you struggle with maths, so maths makes you anxious. The debilitating-anxiety theory says the arrow runs the other way — anxiety disrupts working memory and drags performance down. The reciprocal theory says both are true at once, in a self-reinforcing loop.

The most useful synthesis is Carey, Hill, Devine and Szűcs's 2016 review in Frontiers in Psychology, bluntly titled "The Chicken or the Egg?" Their conclusion, after weighing the evidence, is that the relationship is best described as bidirectional — and that the field has not cleanly established a single dominant direction. Longitudinal work in young children finds some evidence that early struggle predicts later anxiety (a deficit-direction signal); experimental work that manipulates pressure finds anxiety dragging down performance (a debilitating-direction signal). Both mechanisms appear to operate.

And the association itself, while real, is modest. The largest meta-analysis to date — Barroso and colleagues, 2021, in Psychological Bulletin — pooled hundreds of samples and found an overall correlation between math anxiety and achievement of roughly −0.28: a small-to-moderate negative relationship. It also found enormous heterogeneity across studies (much of the variation was not explained by the average), and a publication-bias signal, with published studies reporting stronger correlations than unpublished ones. In plain terms: anxiety and low achievement travel together, but loosely. Plenty of anxious students perform well; plenty of struggling students aren't especially anxious. Anyone who tells you math anxiety determines outcomes is overstating a modest correlation.

Math anxiety is partly learned — from parents and teachers

If there is one strand of this research every parent should know, it is this one — because it is both surprising and actionable.

Maloney, Ramirez, Gunderson, Levine and Beilock (2015, Psychological Science) tracked over four hundred first- and second-graders and their parents across a school year. Children of highly math-anxious parents learned less maths and grew more anxious over the year — but only when those parents frequently helped with maths homework. When math-anxious parents helped rarely, there was no such effect. The transmission wasn't in the genes or the household mood generally; it was in the homework interactions themselves — the sighs, the "I was never good at maths either," the tension around getting it right.

The classroom version is just as striking. Beilock, Gunderson, Ramirez and Levine (2010, in the Proceedings of the National Academy of Sciences) followed early-primary teachers and their pupils. The more math-anxious a (female) teacher was, the more the girls in her class — not the boys — came to endorse the stereotype that boys are better at maths, and the girls who absorbed that belief ended the year with lower maths achievement. The sample was small, and the finding is correlational, but the mechanism is plausible and sobering: adults transmit not just skills but attitudes, and children are watching for cues about whether maths is a threat.

In practice, that doesn't mean hiding your feelings perfectly. What seems to matter more is how the maths-anxious adult shows up around a child's maths — especially during help and homework — not whether the anxiety exists at all.

The gender question: what's robust versus overstated

This area is a minefield of half-truths, so it's worth stating carefully what the evidence supports.

Robust: on self-report questionnaires, girls and women consistently report higher math anxiety than boys and men. This shows up across large international datasets.

Overstated: the idea that this reflects a real gap in mathematical ability. It doesn't. Meta-analyses of maths performance — Hyde and colleagues over several decades, and Lindberg and colleagues' 2010 analysis of over a million people — find gender differences in achievement that are essentially zero (effect sizes around 0.05). Else-Quest, Hyde and Linn's 2010 meta-analysis in Psychological Bulletin, drawing on nearly half a million students across 69 nations, found near-identical achievement alongside less favourable maths attitudes among girls — and, tellingly, found that the size of the affective gap tracked each country's level of gender equity, not anything intrinsic.

Then there is a measurement twist that should make everyone cautious. Goetz and colleagues (2013, Psychological Science) asked whether girls really experience more maths anxiety, or merely report more when asked about it in general. Using experience-sampling — pinging students during actual maths lessons and tests to capture anxiety in the moment — they found the gender difference in trait (habitual, "in general") anxiety was real, but the difference in state (right-now) anxiety largely disappeared. Part of the reported gap seems to reflect girls' lower perceived competence despite equal grades, and a greater willingness to report anxiety, rather than more anxiety actually experienced.

The takeaway for a parent of a nervous daughter: the numbers do not support reading her anxiety as evidence she is less able. Her measured performance is very likely fine; her confidence may be the thing lagging her ability.

The Singapore paradox: top scores, and real anxiety

Singapore sits at or near the summit of international maths rankings. It would be easy to assume that means its students are relaxed about maths. The data says otherwise — but in a way that is widely misdescribed, so precision matters here.

In PISA 2012, the OECD explicitly named Singapore among a cluster of the highest-performing systems — alongside Hong Kong, Japan, Korea, Macao, Shanghai and Chinese Taipei — that "reported higher levels of anxiety than would have been expected given their performance." That phrase is doing careful work. The claim is not that Singapore students have the highest absolute anxiety in the world (they don't; the highest raw anxiety scores are found in a number of lower-performing systems). The claim is that, relative to how well they score, these East Asian systems are anxious outliers. A reanalysis of the same PISA data (Thien and Ong, 2015) placed Singapore's anxiety index right around the international average — even slightly below — while its performance and self-efficacy sat far above it.

Two cautions temper that claim. First, avoid the ecological fallacy: between countries, higher-anxiety systems actually tend to score lower on average, and East Asia is the exception above that downward trend, not proof that anxiety and achievement rise together. Second, within Singapore, as within almost everywhere, the individual-level relationship is still the familiar negative one — more anxious students tend to score somewhat lower.

What the pattern does suggest is that a high-performing, high-stakes environment can co-exist with, and perhaps quietly cultivate, real maths-related worry — even as scores stay excellent. That is consistent with the broader picture of academic pressure we discuss in our review of what the data actually says about tuition in Singapore, and with the wellbeing costs — like disrupted sleep — that a heavy academic load can carry. High marks are not proof that a child is fine.

What actually helps — the evidence on interventions

The encouraging news is that math anxiety is treatable. The catch is that the evidence base, while positive, is thinner and more short-term than the confident claims of any single programme would suggest.

The foundational synthesis is still Hembree's 1990 meta-analysis in the Journal for Research in Mathematics Education, which pooled 151 studies. Its findings are remarkably specific:

• Anxiety-reduction techniques worked, but selectively. Systematic desensitisation and cognitive-behavioural approaches produced large reductions in maths anxiety. Relaxation training on its own did not. Whole-class curricular tweaks and group discussion did not.
• Reducing the anxiety improved the maths. Across the effective treatments, performance rose — enough, on average, to bring formerly high-anxious students up to roughly the level of their low-anxious peers. That is a striking result: for some students, the maths ability was there all along, trapped behind the anxiety.

More recent work refines rather than overturns this. Two threads are worth knowing:

Address the fear directly. Ramirez and Beilock (2011, Science) had students spend about ten minutes writing about their exam worries immediately before a real classroom test. High-test-anxious students improved appreciably — on the order of a grade band — while calm students were unaffected. Jamieson and colleagues (2010) taught test-takers to reappraise their bodily arousal ("the racing heart is your body helping you focus, not a sign you're failing"), and saw improved performance on real, high-stakes GRE maths months later. Both are cheap, brief, and target the anxiety rather than the arithmetic.

Address the skill gap directly. Supekar and colleagues (2015, Journal of Neuroscience) put seven-to-nine-year-olds through eight weeks of one-to-one maths tutoring. High-anxious children became measurably less anxious, and their previously over-active amygdala responses normalised — evidence that building genuine competence can dissolve anxiety, not just paper over it. (The study had no untreated control group and a small sample, so treat it as suggestive rather than definitive.)

A 2023 meta-analysis by Codding and colleagues put both approaches side by side across 17 studies. Skill-based interventions produced the larger gains in achievement; therapeutic ones produced larger gains in anxiety — but once study quality was accounted for, the difference between the two types was no longer statistically reliable. The most defensible reading is that both levers work, the evidence base is quality-limited, and the best programmes probably do both: they lower the fear and close the specific gaps that justify it.

Two cautions on the fashionable options. "Growth mindset" workshops have weak and mixed evidence specifically for reducing math anxiety; the broad mindset literature shows small average effects on achievement (around 0.08) and is actively contested. And you will occasionally see excited coverage of brain-stimulation studies — these are small, single-session, mechanistic experiments, not treatments, and in at least one careful study the same stimulation that helped anxious participants hurt calm ones. Neither belongs in a plan for your child.

What the research does not establish

Intellectual honesty requires being clear about the limits.

• The causal direction is not settled. The relationship is reciprocal and most of the data is correlational. Claims that anxiety causes a specific grade drop (or that fixing anxiety will cause a specific rise) run ahead of the evidence.
• The correlation is modest, not deterministic. At roughly −0.28, math anxiety explains only a minority of the variation in achievement. It is one factor among many.
• Durability and long-run outcomes are largely untested. Most intervention studies are short, small, and often lack proper control groups. Whether a ten-minute writing exercise or an eight-week tutoring block changes a child's trajectory years later is essentially unstudied.
• Measurement is inconsistent. Dozens of scales, no clinical threshold — cross-study comparison is muddied by how anxiety was measured.
• Group patterns don't describe individuals. The average girl, the average anxious student, the average Singaporean — none of these describe your specific child, whose situation deserves specific attention rather than a demographic assumption.

Practical takeaways: what a careful reading suggests

This isn't a piece arguing that math anxiety is everything or nothing. A careful reading, though, surfaces a handful of things that are more defensible than most of what circulates:

1. Treat competence and calm as a pair, not a choice. The strongest evidence supports doing both — closing the specific skill gaps that make maths hard for a child, and reducing the fear that hijacks their working memory. A programme that only drills, or only reassures, is doing half the job.

2. Watch the homework interactions, not just the homework. If maths help at home reliably ends in tension, the research suggests that tension is teaching something. Sometimes the most useful change is who delivers the help, or lowering the emotional temperature of it — not more hours.

3. Do not read a nervous child, especially a girl, as a "non-maths person." Measured ability gaps by gender are near zero; confidence gaps are not. Anxiety is a poor proxy for aptitude.

4. Normalise the physical feelings. The single most portable finding is that reinterpreting a racing heart as readiness rather than danger — and getting worries onto paper before an exam — helps anxious students, at almost no cost.

5. Remember that high marks aren't proof of wellbeing. Singapore's own results show top performance and real anxiety can coexist. A child who scores well can still be quietly struggling with maths as an experience — and that is worth attending to on its own terms.

The one thing to remember: Math anxiety is real and it interferes with performance — but it is loosely, not tightly, linked to ability; it runs in both causal directions at once; and it responds to help that targets both the fear and the underlying skill. The useful question is never "is my child anxious or bad at maths?" It's "what specific gap is feeding the fear, and how do we lower the fear enough to close it?"

A note on sources

This review draws on peer-reviewed psychology and education research and on OECD survey data. Readers who want to go deeper will find the most accessible entry points to be:

• Dowker, Sarkar and Looi's 2016 review in Frontiers in Psychology, "Mathematics Anxiety: What Have We Learned in 60 Years?" — the best single overview
• The OECD's PISA 2012 results (Volume III) and the OECD brief "Does maths make you anxious?" for the international data
• Barroso and colleagues' 2021 meta-analysis in Psychological Bulletin for the current best estimate of the anxiety–achievement relationship
• The work of Sian Beilock and colleagues (Chicago) on working memory, choking, and the transmission of anxiety from parents and teachers

The uncertainty in this review is genuine, not a hedge — the causal story is unresolved and the intervention evidence is still maturing.

---

A note on our own position: We run a maths programme, so we have an obvious interest in the conclusion that good teaching helps. We've tried to write this honestly anyway — because a child's anxiety is not a marketing opportunity, and pretending the evidence is tidier than it is would be a disservice.

What the research above tells us about doing this work well: diagnose the specific skill gap rather than assuming a child is simply "anxious," keep feedback timely and psychologically safe, resist piling on volume for a child who is already frightened, and treat the fear and the gap together. That is how we try to teach students who have grown anxious about maths — and it is the same reason we've written a companion, more practical guide on math tuition for students with high anxiety. If you have a child who has come to dread maths and you'd like to talk through what would actually help, talk to our team.