A combined scoring and ranking approach for determining overall food quality

The problem of obtaining the consensus evaluation of food samples is commonly considered by researchers in food science. Typically, trained panellists provide scores describing the overall quality of the samples. However, due to the fact that trained panellists are limited in number and very expensive, it is common to recruit untrained panellists to provide rankings of the samples. This paper describes a method allowing to combine scores and rankings with the aim of obtaining a consensus score and, thus, improving the quality of the assessment of a sample. This method is based on a combination of the median of scores and the Kemeny median of rankings. An experiment on raw Atlantic salmon (Salmo salar) was carried out to illustrate the applicability of this method. The samples were evaluated by trained (scoring) and untrained (ranking) panellists, and the consensus score of each sample was determined. Finally, the influence of the rankings on the assigned consensus scores was validated on the basis of clustering analysis.