Statistics:
Since the normality of the data was not known and normality tests have
low statistical power for a sample size of 24 (Ă–ztuna et al. ,
2006), non-parametric one-sample Wilcoxon signed rank test and one way
Kruskal-Wallis test were used for the analysis. Type I error probability
was kept at 5% by using Bonferroni-Holm method for correction of
multiple comparisons (stats.multitest.multipletests function from
Statsmodels module, Python, Holm method).