*2.8. Statistical Analysis*

Physicochemical analyses data were statistically analysed using the generalised linear model procedures of SAS software (Version 9.4; SAS Institute Inc., Cary, NC, USA) in a completely randomised block design with the fat type as the main effect. Observations over time were combined in a split-plot analysis of variance with production stage (raw batter and finished product) as a sub-plot factor. For DSA data, judges (n = 12) were considered as block replicates for each sample (backfat type × replicate). A Shapiro–Wilk test was performed on the standardised residuals from the model to test for normality. In cases where there was significant deviation from normality, outliers were removed when the standardised residual for an observation deviated with more than three standard deviations from the model value. The data for consumer acceptance were analysed using mixed model repeated measures of ANOVA. Fisher's least significant difference was calculated at the 5% significance level to compare treatment means.


**Table 1.** Sensory attributes description, scale and reference standards used for the descriptive sensory analysis of two cabanossi treatments.

a Low, moderate and high intensity were made using 1, 2 and 3 drops of liquid smoke solution in 100 mL water; b Roasted pork/mutton loin was roasted to an internal temperature of 75 ◦C and surrounding fat was used; c Fried bacon, pan fried bacon; d Wood sawdust solution, a cup-full (250 mL) of pine wood saw dust was soaked in water overnight and filtered; e Low, moderate, high intensity salt solution was made using 0.2%, 0.5% and 1% salt solution; f Dry cabanossi, a batch of cabanossi produced with no additional fat.


**Table 2.** Demographic variation of the sample population according to gender and age group.

\* Proportion expressed as a percentage of the total population size.

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