Can changing the meal sequence in school canteens reduce vegetable food waste? A cluster randomized control trial

• We investigate the impact of changing the dishes sequence in schools on food waste. • We apply a cluster RCT to the largest and the most diverse sample of schools so far. • 26 Italian schools are randomly selected and assigned to control and treatment. • Serving the side dish first changes vegetable waste by minus to plus one third. • Children’s waste of the side dish depends on the type of vegetables served. School canteens are crucial drivers for stimulating sustainable eating habits of children in line with global sustainability objectives. Using a participatory approach, we co-design a choice-architecture intervention to assess whether inverting the sequence of Mediterranean lunch dishes served in school canteens is effective in reducing food waste. We implement a randomized control trial using 26 Italian schools, creating one of the largest and the most diverse samples achieved so far for such an experiment. Hence, our analysis yields evidence of internal and external validity far beyond existing interventions which consider only a few education institutions without random selection. A food mass flow analysis of the weighed prepared food, serving waste and plate waste per meal course suggests that in primary school canteens 29% of the lunch prepared gets wasted, i.e., 22,000 tons per year in Italy. Big catering providers are found to produce 23% more food waste. Prioritizing the vegetable side dish as first course has heterogeneous effects across school clusters changing the share of plate waste in total vegetable servings between −26 pp to + 32 pp. Also, the side dish plate waste depends on the type of vegetables served. Differences in lunch implementation and food and eating environments across schools appear to dominate the effectiveness of the intervention, hence, no treatment effect is found for the full sample. Upscaling of the dish reordering as a binding policy strategy for the entire regional territory is therefore not recommended. We conclude that policymakers and researchers should more often use citizen science to maximize benefits for society.