Anavex Life Sciences Corp (AVXL)

[Steady_T]

Good grief. I understand how to compute effect size etc. I also understand the consequences of high variability in scores resulting in a large standard deviation.

And there is a real possibility that a larger n in the placebo arm might have reduced the standard deviation and perhaps the mean.
Random selection has some issues especially when the n of the selected population is small. The smaller the n the likelihood that the selected population is less representative of the overall population. 30 is considered to be the smallest n that is normally used with the common statistical tools. An n less that 30 usually require the use of different tools.

There are both type 1 and type 2 errors.