Optimal design of national food waste monitoring for the food service sector

Abstract

The countries of the European Union have committed to monitoring and lowering food waste (FW) levels, including in the food service sector. In this work, the relative importance of factors such as food service type, outlet size, seasonal and weekday variation, and length of measurement period for monitoring originally edible and total FW is explored. We provide the first sample size estimates needed for national-level monitoring, and take the length of the observation period carefully into account and whether the food service type, outlet size, and regional variation are controlled for, and what size of a change in FW one wants to be able to observe. Two datasets collected as part of the Finnish national monitoring system were used. The first consisted of 192 two-week measurements from outlets of various sizes. The second consisted of three long-term (1.5–2 year) monitoring efforts. The food service type is by far the most important factor, explaining 66 % and 54 % of the variance in edible and total FW, respectively. The next factor is monthly variation (13 % and 10 %), municipal level variation (5 % and 9 %), and weekday-variation (6 % and 5 %), with the outlet size being relatively unimportant (2 % and 8 %). At the smallest, a sample size of n = 68 is needed to observe a change of 30 % in the total national FW levels. With two-week measurements n = 72 is sufficient, highlighting the fact that longer measurement periods are not resource efficient, because between-outlet variation tends to be greater than within-outlet variation. The results can be used to develop national data collection in other countries, to optimize the limited resources available for monitoring, and to improve the comparability of measurements between years.