(configurations) were matched to find a consensus using the software Procrustes-PC V2.2 (OP and P, Utrecht, the Netherlands).
In order to evaluate whether the data from each subject were significantly fitted by the model the permutation test was performed (Dijksterhuis and Heiser, 1995).
This procedure rearranges at random each assessor’s score and produces new per-mutated data sets where the score no longer corresponds to the original products, thus allowing a distribution to be obtained which gives the probability that the
degree of fitness may have occurred by chance alone.
The first two axes obtained in the consensus configuration were interpreted using the value of the loadings (Thomson and McEwan, 1988).
In particular, attributes were considered to be important for the assessors when the corresponding values exceeded O-3
(absolute magnitude).