On a proposed method for analysing experiments on food choice