highly significantP < 0.01
The non-parametric approach was preferred if the Shapiro/Wilk statistics 16 indicated a highly significant ( P < 0.01) deviation from normal distribution which was a conservative procedure in favour of the normality assumption.