Learning About Culture: Overarching Evaluators' Report

Five cluster-randomised controlled trials, ~8,500 children in 400 English state schools; analysed samples 811-2,150 per trial, 5,255 in the pooled Key Stage 2 analysis

English state primary pupils. KS1 (ages 5-7) - First Thing Music and Speech Bubbles. KS2 (Year 5, ages 9-10) - The Craft of Writing, Power of Pictures, Young Journalist Academy. Over-sampled for pupil-premium eligibility.

The largest randomised test of arts-based education ever run, and structurally the strongest evidence in either the arts or the music cluster: five separate school-level randomised trials, independently evaluated by a team with no stake in the programmes, deliberately harmonised so that the two KS1 trials shared a primary outcome (Progress in Reading Assessment) and the three KS2 trials shared another (Writing Assessment Measure), with a common team building the logic models. All outcomes are externally published standardised instruments, not researcher-designed ones. The KS2 pooled analysis (n = 5,255) was not pre-specified and is flagged as exploratory by the evaluators. FIRST THING MUSIC is the entry that matters for music: a daily whole-class music programme delivered to 2,150 KS1 pupils, i.e. the single largest randomised test of classroom music on reading attainment that exists. The evaluators themselves cite Sala and Gobet (2020) in the report as the prior reason to doubt music-to-cognition transfer.

No statistically significant effect of any of the five arts programmes on any attainment measure. First Thing Music (whole-class daily music, n = 2,150): reading 0.07 [-0.02, 0.19], p = .13. Speech Bubbles (drama, n = 821): reading -0.05 [-0.20, 0.10], p = .93, and oral narrative -0.04 [-0.19, 0.11]. Craft of Writing: -0.03 [-0.19, 0.12]. Power of Pictures: 0.09 [-0.07, 0.24]. Young Journalist Academy: -0.13 [-0.32, 0.05]. Pooled KS2 writing: -0.04 [-0.14, 0.07], p = .49, n = 5,255. Secondary outcomes - creativity, ideation, writing self-efficacy, social skills - were also null, with one exception at the 10% level (Power of Pictures writing self-efficacy 0.11 [-0.00, 0.22], p = .06). The subgroup analysis of free-school-meal pupils found no differential effects anywhere. The most instructive number in the report is not an effect size: 97% of Craft of Writing teachers and 95% of Power of Pictures teachers perceived a positive impact on ideation, 85-97% perceived a positive impact on creativity, and 88-98% on engagement, against measured effects on those same constructs of essentially zero. The evaluators are careful to add that the programmes were not harmful, did not appear to crowd out other beneficial activity, and that "the evidence from these trials does not indicate that schools should not continue to pursue arts-based education."

Low. School-level randomisation means genotype cannot differ systematically between arms; this is precisely the design that removes the selection confound contaminating the observational arts-participation literature.

The null replicates five times within the programme, across two key stages, three art forms (music, drama, illustration/writing) and two different standardised outcome measures. Consistent with the Sala & Gobet meta-analyses; in tension with Román-Caballero et al. (2022) only insofar as First Thing Music is a classroom music programme rather than instrumental tuition, which is exactly the distinction that meta-analysis argues is load-bearing.