The meta-analysis found a small restriction-diet effect and small food-color effects that varied by informant, study quality, and publication-bias adjustment. Parent-reported food-color g was 0.18 and fell to 0.12 after adjustment; overall teacher/observer g was a nonsignificant 0.07. Effects were not reliable in studies confined to FDA-approved colors. The authors estimated the diet effect at roughly one-third of a summary medication effect of 0.9. Funding included ILSI North America and partial support from the National Confectioners Association.