Distributions of serving sizes and plate waste in school lunches. Implications for measurement.

Distributions of school lunch serving size and plate waste data are discussed in items of implications for plate waste measurement. Mean serving size was a representative measure when based on 10 rather than the 4 or 5 samples typically weighed. Distributions of percent plate waste supported previous findings of L- or U-shaped distributions. Percent waste of a given food was better represented by the proportion of individuals who consumed all or almost all, and none or almost none, of a serving, than by mean percentage of waste.