2.9. Sample Size Calculation

Recruitment of 10 schools and 200 students per school would allow detection of a 19.5% significant difference in the purchase of ‘everyday’ items, assuming an Intraclass Correlation Coefficient (ICC) of 0.05, with 80% power, and 0.05 significance level. An effect of this size has been previously found in a study of 10 government schools where 5 were receiving a multi-component intervention to improve the healthiness of student lunch orders [35]. Similarly, the study was powered to detect a 13.5% significant decrease in the purchase of ‘caution’ items.