Normal and random distribution are required for
optimality in all multivariate statistical analysis. The
Shapiro–Wilk (Shapiro and Wilk, 1965) test was performed
on each dataset to determine normal distribution. Except
for Na+, K+, HCO3
and PO4 3 for Ankwaso, and Na+,
and K+ for Dominase and Prestea, which were initially not
normally distributed due to the presence of obvious
outliers in the data, the rest of the parameters are normally
distributed at po0.05. When the outliers were replaced by
mean values, data for the above named parameters
assumed normal distribution and were well suited for
optimal multivariate analysis.