Mathematics skills in autism: A systematic review and meta-analysis

Based on stereotypes, autistic individuals are often expected to be mathematically gifted. However, research on mathematics skills in autism has yielded inconsistent findings, showing both superior and lower performance as compared to neurotypical people. To address this, we conducted a pre-registered systematic review and meta-analysis, following the PRISMA guidelines, comparing the mathematics performance of autistic and neurotypical individuals matched on age and intelligence. Nineteen publications (20 effect sizes) were included in our analyses, presenting results from 2029 participants (964 autistic and 1065 neurotypical). A random-effects model showed an autistic disadvantage in mathematics performance compared to neurotypical individuals, with a small-to-medium effect size (Hedges’ g = −0.33). Similar overall effect sizes were obtained when computation (Hedges’ g = −0.35) and word problem-solving skills (Hedges’ g = −0.25) were considered separately, with no significant difference between the effect sizes across the two domains. Although there was heterogeneity in the effect sizes, meta-regression analyses found that age and intelligence did not moderate the magnitude of group differences. Moreover, an additional meta-analysis showed greater variability in the mathematics performance of autistic participants in the included studies, compared to neurotypical participants. A publication bias analysis indicated that studies showing an autistic disadvantage in mathematics were less likely to be published. Even though there is evidence in the literature for significant inter-individual variability in mathematics skills in autism, with autistic people being represented at both the highest and lowest ends of the mathematics attainment continuum, overall, our findings are in contrast with the popular stereotype that autistic people, in general, have a special talent in mathematics.