Anavex Life Sciences Corp (AVXL)

[Doc328]

This is not an issue. MWU and t-test are similar. If data is non-parametric and unpaired then the MWU is the appropriate test. If results are distributed in a more gaussian manner, then the t-test is the appropriate test. In my last publication, I used the Shapiro Wilk test on several result sets to determine whether the data deviated enough from the normal curve to require non-parametric statistical sets. If non-parametric, I used the Wilcoxan rank test (similar to the MWU for comparison of means instead of the t-test) or Spearman rank test (for correlation instead of Pearson).