Discussion:
Assuming data normally distributed:
Q1 Theoretically, does the nature of the observed data have an effect on the definition of what is called "mild" and extreme" outliers?
Q2 What percentage of values will be "mild" or "extreme" outliers for the observed data?