Methodology to analyse rank and carry-over effects

In the frame of a sensory evaluation in which assessors successively taste several products, this paper uses a model of analysis of variance including both the rank and the carry-over factors. The application of this model to a data set shows how important these factors can be. The analysis of the coefficients of the carry-over effects suggests either a contrast effect, especially for the attributes associated with the colour, or a more complicated effect, like for bitterness. These effects are often weak compared to the product effects, but a multidimensional approach reveals trends that are sensorially coherent. The value of balanced designs for the rank and carry-over effects is confirmed.