Music Lessons Enhance IQ

144 six-year-olds randomised (36 per arm); 132 completed post-testing (attrition 8.3% — keyboard 6, voice 4, drama 2, no-lessons 0)

Toronto six-year-olds recruited by newspaper advertisement offering free weekly arts lessons; families screened for a keyboard at home and willingness to accept any assignment

The single most important positive study in the music-transfer literature and the only one that combines individual random assignment, a genuine active control, blinded testers and a standardised outcome. Four arms: standard keyboard, Kodály voice, drama, and no lessons, taught for 36 weeks in groups of six at the Royal Conservatory of Music by certified professionals, two instructors per arm. Outcomes were WISC-III full-scale IQ and index scores, the Kaufman Test of Educational Achievement, and the parent-rated Behavioral Assessment System for Children. Test administrators were blind to condition. Grade B rather than A: single site, unreplicated, n = 132 analysed, and the effect is a between-groups difference of 2.7 IQ points sitting on top of a 4.3-point rise that every arm showed.

All four groups gained IQ (all p < .005), which the author attributes to starting school. The music groups gained 7.0 points (SD 8.6) against the control groups' 4.3 (SD 7.3) - a 2.7-point difference, t(130) = 1.99, p < .05, d = 0.35. Index scores gave F(1,130) = 6.88, p < .01, d = 0.46, and the music advantage was similar across all four indices; on 10 of 12 subtests music gained more. On the standardised ACHIEVEMENT test there were NO reliable group differences, though music was numerically ahead on all five subtests (sign test p < .05). The result nobody quotes is the drama arm: adaptive social behaviour improved substantially in the drama group, t(33) = 4.00, p < .0005, and not in the music or no-lessons groups, with drama versus the rest at t(129) = 2.88, p < .005, d = 0.57 - a LARGER effect than the music-IQ effect this paper is famous for. Schellenberg's own proposed mechanism is not music-specific: "Music lessons, taught individually or in small groups, may provide additional boosts in IQ because they are like school but still enjoyable... extracurricular activities (other than drama lessons) with similar properties (e.g., chess lessons, programs in science or reading) may confer similar benefits."

Low. Individual random assignment with verified pre-test equivalence on IQ, age and family income. This is the cleanest genetic control in the music-training literature.

Never directly replicated. Mehr et al. (2013) ran two randomised preschool trials with a visual-arts active control and a no-treatment control and found no consistent cognitive benefit. Sala & Gobet (2017) note this is the ONLY study in their corpus using both randomisation and an active control for IQ, where it gives d = 0.16 - less than half the headline. Román-Caballero et al. (2022) point out that Schellenberg's randomly assigned children practised only 10-15 minutes a week, far less than children who choose music, which cuts the other way. Schellenberg himself has become the finding's most careful critic; his 2024 Annual Review with Lima concludes the evidence that music training causes non-musical benefits is "weak or nonexistent."