Math anxiety has a small-to-moderate negative correlation with math achievement (r = −.28) in a meta-analysis of correlations from 223 studies¶

Claim · All claims
Evidence · 1 study · quant-synthesis r2 · q3 · i2 medium · n=332 independent samples (~385,441 participants) from 223 studies
Currency · newest study 2021 · checked against its sources 2026-10-20 (automated, 1 of 1 entry) · not reviewed by a person

Math anxiety — feelings of tension, apprehension, or dread when manipulating numbers or solving mathematical problems — goes with lower mathematical achievement. The one synthesis recorded here pools correlations, so it shows the association and its moderators, not that anxiety causes the lower scores: the relation may run both ways. The usual causal account, that anxiety consumes working memory needed for computation and reasoning and so impairs performance most under evaluative pressure or high working memory demand, is discussed below as a mechanism no entry here tests.

Subclaims¶

q3 i2 A meta-analysis of 223 studies (747 effect sizes, 332 independent samples, ~385,441 participants, 1992–2018) finds a significant small-to-moderate negative correlation between math anxiety and math achievement (r = −.28), significant across nearly every moderator subgroup examined. → Barroso et al. 2021

Evidence¶

Barroso et al. 2021¶

Barroso, C., Ganley, C. M., McGraw, A. L., Geer, E. A., Hart, S. A., & Daucourt, M. C. (2021). A meta-analysis of the relation between math anxiety and math achievement. Psychological Bulletin, 147(2), 134–168. doi:10.1037/bul0000307

q3 · meta-analysis of correlational studies · i2 · medium effect, r=-.28 · n=332 independent samples (~385,441 participants) from 223 studies · quant-synthesis · r2

Multilevel meta-analysis of 223 studies from research conducted 1992–2018 (published work and grey literature, plus 26 unpublished data sets and manuscripts), modeling 747 correlation coefficients from 332 independent samples (~385,441 participants) to account for non-independence among effect sizes from the same sample. The pooled correlation between math anxiety and math achievement was r = −.28 (small-to-moderate, negative, statistically significant), and the relation held for nearly every moderator subgroup tested (grade level, math-ability level, adolescent/adult anxiety scales, math topic of the anxiety scale, and assessment type), with the sole exception of anxiety and approximate-number-system tasks. Grade level, ability level, and scale/assessment type were significant moderators of the size of the relation, and effect sizes were somewhat weaker in grades 3–5 and postsecondary samples than at other ages.

Studies bearing on this claim¶

Do learners with higher anxiety about a subject or a test achieve less in it? This table gathers the studies behind this claim and the 13 narrower claims that bear on it: 11 distinct studies (3 review, 2 quant-synthesis, 2 associational, 2 causal, 1 qualitative, 1 theoretical), q1–q4, 3 of 11 with an effect size coded. One row per study, codes as each entry records them; no effect is averaged and no verdict is drawn. Built by scripts/synthesize_claim.py from eval/synthesis/; a row that looks wrong is fixed on the claim it comes from.

Discussion¶

Mechanism. The dominant account is processing-efficiency: anxious thoughts (worry, rumination) act as a secondary task that occupies the same working memory resources mathematics demands, so performance drops most on problems with high working memory demands. This places math anxiety in the same family as other load-inducing impairments — see Cognitive overload degrades learning and Chunking reduces working memory load. Within Cognitive load theory, anxiety functions as an additional source of extraneous load rather than a separate affective phenomenon.

Boundary conditions. The degradation is typically strongest under timed, high-stakes, or evaluative conditions, and weakest when pressure is removed or when learners can offload computation (e.g., scratch work, calculators for routine steps). Highly anxious individuals are often not low-ability: performance gaps between high- and low-anxious learners widen with pressure, consistent with anxiety suppressing rather than reflecting competence. This makes the effect context-dependent rather than uniform — a [~]-direction pattern in which the impairment emerges only when situational demands collide with anxiety-generated load.

Directionality. The relationship is likely reciprocal — poor performance breeds anxiety, and anxiety degrades subsequent performance — so this claim should be read as one loop in a cycle, not a simple one-way cause. This also means interventions can target either side of the loop: reducing evaluative pressure or building fluency to remove sources of failure.

Design implications. If the working-memory account holds (no entry here tests it), the impairment is load-driven rather than ability-driven, so instructional designs that reduce working memory demands — worked examples, externalized computation, untimed practice — should disproportionately benefit highly anxious learners. Timed testing and public performance of mathematics are the conditions most likely to widen anxiety-based gaps; designers using Test-enhanced learning should weigh retrieval-practice benefits against the evaluative pressure that can trigger impairment in anxious learners. Designers working within Cognitive load management should treat anxiety as an additional source of extraneous load, not merely an affective side issue, and appraisal-focused approaches such as Belonging interventions improve outcomes address the evaluative concerns that drive the anxiety in the first place.

Open questions. Effect sizes vary widely across age groups and task types; whether anxiety specifically harms arithmetic retrieval, procedural execution, or novel problem-solving to different degrees remains contested. The field also lacks strong causal identification for the working-memory account in classroom settings, where most evidence is correlational.

Learner Variables¶

• Academic Anxiety — predictor: learners who differ on it differ in outcomes

Related Claims¶

• Test anxiety is negatively associated with exam performance, and no moderation by proctoring was detected — the general claim this finding bears on, with the studies behind it

• Cognitive overload degrades learning — anxiety imposes extraneous load that competes with the same working memory resources

• Chunking reduces working memory load — reducing intrinsic load can buffer against anxiety-induced impairment
• Cognitive load management — practical strategies for the load reduction that buffers anxiety effects
• Cognitive load theory — the theoretical framework underlying the working-memory account of math anxiety
• Belonging interventions improve outcomes — social-belonging and appraisal-reduction interventions address the evaluative concerns that drive anxiety
• Test-enhanced learning — testing benefits must be weighed against the evaluative pressure that can trigger anxiety-based impairment
• Brief expressive writing before an exam can raise test-anxious students' scores, but across the wider literature its effect on test performance is negligible — related
• Synthesis and Evaluation subtests measure general mental ability rather than knowledge for lower grades or unfamiliar content — related
• Drop-out from adult language classes is slightly higher than from other courses, with anxiety and feelings of inadequacy a contributing reason — related
• Writing anxiety can constrain or paralyze adult writers, with recognizable signs — related
• Guiding analytics for learners can improve mathematics achievement, retention, learning strategies, and reduce math anxiety — related
• The anxiety–achievement correlation varies across educational levels, target languages, and anxiety types — related
• L2 anxiety is negatively correlated with L2 achievement, with a mean correlation of r = −.36 across 216 effect sizes — a narrower finding that bears on this claim
• Interest correlates moderately with academic achievement (mean correlation r = .31, meta-analytic) — related
• Early identification and treatment of math anxiety is important because early anxieties can snowball into avoidance of mathematics — related
• Intrinsic math motivation is positively correlated with arithmetic skills and negatively correlated with explicitly measured math anxiety in sixth graders — a narrower finding that bears on this claim
• Math anxiety correlates strongly and negatively with mathematics achievement among third graders — a narrower finding that bears on this claim
• Math anxiety compromises working memory only when the anxiety is aroused by computation-based tasks, not on verbal span assessments — related
• Math anxiety's negative impact on math achievement is limited to more difficult math, emerging around late elementary school content — a narrower finding that bears on this claim
• Math anxiety correlates negatively with math achievement, grades, enjoyment, motivation, and amount of math taken — possibly the same claim (merge candidate)
• Lower-than-average math ability, low working memory capacity, susceptibility to public embarrassment, and a nonsupportive teacher are proposed risk factors for developing math anxiety — related
• Mathematics anxiety does not moderate the effect of mathematics test practice on subsequent performance — related
• Enjoyment correlates positively (r = .31), and boredom, mathematics anxiety and test anxiety negatively (r = −.23 to −.28), with mathematics achievement — possibly the same claim (merge candidate)
• No significant impact of frustration or boredom (or anxiousness, discouragement, distractedness) on practice performance was found — related
• The anxiety–achievement correlation varies across types of language achievement measures — related
• Anxiety and toxic stress from stereotypes, bullying, and exclusionary practices impede learning by preoccupying the brain with worry and fear — related
• At semester end, intrinsic motivation was positively correlated and programming anxiety negatively correlated with programming performance — a narrower finding that bears on this claim
• Math anxiety explained variance in posttest monitoring judgments beyond trait anxiety, supporting its distinctness from general anxiety — related
• The goal–internalizing relations are moderated by goal terminology, goal setting, education level, goal model, and cultural context — related
• Trait test anxiety predicts performance-based goals, and in a path model state anxiety and worry, but not emotionality, link performance-avoidance goals to lower exam performance — related
• Age group significantly but weakly moderates the math anxiety–performance link (children r = −.31 vs adults r = −.25) — a narrower finding that bears on this claim
• Math anxiety correlates negatively with math performance at a pooled r = −.30 across 169 independent samples — related
• Type of math task moderates the math anxiety–performance link: basic numerical knowledge shows the weakest correlation (r = −.22) and applied math among the strongest (r = −.31) — a narrower finding that bears on this claim
• Test anxiety correlates negatively with math achievement at a pooled r = −.23, weaker than math anxiety's correlation — related
• Academic buoyancy moderates the relation between anxiety and mathematics test performance, with performance highest when anxiety is low and buoyancy high — related