*2.9. Statistical Analysis*

The normal distribution of the data was verified by means of the Shapiro–Wilk test and Q-Q plot representation. In addition, the equality of variance within among the variables was verified using the Levene's test. 3-factors ANOVA (treatment vs. pressure level vs. storage time) followed by Duncan's means post hoc comparison test was applied for the analysis of all quality attributes in the shelf-life study of yoghurt samples. The physicochemical, rheological and sensory data matrices were log transformed and subject to Principal Components Analysis (PCA) to explore the qualitative affinities between the different products developed. All statistical analyses were performed using Statistica® v.7 software (StatSoft Inc., Tulsa, OK, USA). Partial least squares regression (PLSR) using the leave-one-out validation method, as previously described in Kanta et al. [25], was used in order to explore the sensory modal drivers of degree of overall liking (DOL) of yoghurt samples. All PLS analyses were carried out using XLSTAT 2014 software (Addinsoft, UK).

### **3. Results and Discussion**
