Biometrical Letters Vol. 53(1), 2016, pp. 1-20
|
ON PREFERMENTATIVE MACERATION TECHNIQUES: STATISTICAL ANALYSIS OF SENSORY DESCRIPTORS IN SANGIOVESE WINE Alessandro Magrini1, Ottorino L. Pantani2, Alessandra Biondi Bartolini3, Federico M. Stefanini1 1Department of Statistics, Computer Science, Applications, University of Florence, Italy, e-mail: stefanini@disia.unifi.it 2Department of Soil Vegetation and Forest Environment, University of Florence, Italy 3R&D wine and sensory consultant |
The analysis of wine sensory descriptors is a fundamental step in the improvement of wine making, because the procedures are judged just before bottled wine is ready for consumption. Despite several contributions in the literature, traditional analysis of variance methods are not adequate to analyse sensory descriptors, because they are defined on ordinal scales. In this paper, we exploit cumulative link mixed models in a three-way full factorial design to assess the effect of prefermentative maceration, temperature and saignée on wine sensory descriptors. Using cumulative link mixed models, the bias introduced by assessors' judgement and the ordinal scale of sensory descriptors are taken into account. The results were the following: the application of prefermentative maceration techniques did not lead to an improvement in the sensory profile of wines after a year from bottling; wines treated with saignée showed greater intensity in olfactive descriptors; and higher fermentation temperatures resulted in wines that were generally more concentrated.
cumulative link mixed models, proportional odds, assessor bias, cold soak prefermentative maceration, cryomaceration
