The Arts and Academic Achievement: What the Evidence Shows

Ten meta-analyses over a comprehensive 1950-1999 search; 188 reports retained yielding 275 effect size correlations

Studies of arts education in children and adolescents, published and unpublished, in English

REAP (Reviewing Education and the Arts Project, Harvard Project Zero, Ellen Winner PI) is the founding survey of arts-transfer claims and it still frames the debate a quarter of a century later. Ten separate meta-analyses were run rather than one pooled model, on the correct grounds that "arts education" is not one intervention. Graded C: the pooled designs are mixed quality and the published summary reports size descriptors (small/medium/large against the authors' own benchmarks of r = .10, .24, .37) rather than per-meta-analysis numeric estimates. Two related documents exist and are often conflated - the Journal of Aesthetic Education double issue 34(3-4) titled "The Arts and Academic IMPROVEMENT" and this Arts Education Policy Review summary titled "ACHIEVEMENT". Widely quoted figures such as r = .32 for drama and r = .25 for music-to-spatial come from secondary sources and are not verified against the primary here.

Three of ten links showed a reliable causal relationship and seven did not. FOUND: classroom drama enacting texts to verbal skills (80 reports, 107 effect sizes; mostly medium effects on oral and written story recall, reading readiness, reading achievement and writing, and - the important part - the gains transferred to new, non-enacted texts); learning to play music to spatial reasoning (19 reports, 29 effect sizes, large, larger still when notation was taught); listening to music to spatial-temporal reasoning (26 reports, medium but temporary and inconsistently found). NOT FOUND: arts-rich education to verbal and mathematics scores or grades (31 reports, 66 effect sizes - a small-to-medium CORRELATION exists but "no evidence was found that studying the arts causes academic indicators to improve"); arts-rich education to creative thinking; music to mathematics (small, inconsistent); music to reading (small, not generalisable); visual arts to reading (very small, driven entirely by reading-READINESS measures, which are themselves visual, and absent for reading achievement); visual arts integrated with reading instruction (medium but not generalisable); dance to reading. The authors' own explanation for the correlation is institutional confounding: schools that take the arts seriously "may become more inquiry-oriented, more project-based, more demanding of high standards."

Not modelled, and the authors reach the same destination by a different route: the arts-rich-education correlation is attributed to school and family selection rather than to a teachable transfer mechanism. Nothing in REAP is genetically informative.

Never formally replicated, but the central negative has held for twenty-five years and has since been confirmed by designs REAP did not have: the Houston school-randomised trial (null on reading and mathematics), the five EEF Learning About Culture randomised trials (null on reading and writing) and the AIR ESSA review (g = 0.11, mathematics null). The one positive REAP identified - drama to verbal skills - has a mixed subsequent record: Lee et al. (2015) reproduce a small positive, while the EEF Speech Bubbles trial (n = 821) is null on both reading and oral narrative recall.