2.7. Statistical analysis
All of the results, elaborated by IBM-SPSS statistics 19 (Armonk,
NY), are reported as the mean of duplicate data from independent
experiments. In all of the parametric tests, the variance homogeneity
was verified by the F-test. All of the hypothesis have been
tested at the significance level of a ¼ 0.05.
Data obtained from physicochemical analyses were elaborated
by paired T-tests in order to evaluate the significance of the differences
between pasteurized smoothies and uninoculated reference
samples.
Microbial count data (Log cfu/mL) from the 10 working suspensions
were elaborated by unpaired T-tests to determine the
significance of differences between selective and non-selective
media, while the significance of differences between the
threshold value and the mean value of the sub-lethal injury percentage
was calculated by one sample T-test.
To verify the possible effect on the microbial viability solely due
to temperature changes, count data (Log cfu/mL) from unpressurized
control samples were elaborated by paired T-tests.
Finally, the influence of various combinations of CCD parameters
was evaluated by modelling data from the CCD runs (Log cfu/mL) as
the dependent variable, by multiple regression procedures and the
step-wise method (backward procedure) to select the most significant
model. The resulting models were shown as surface
response curves.