a. What happens if there is an outlier in one of the samples in a test of two unpaired samples? For each case (very large or very small outlier), say what the rank of the outlier would be.
b. Which statistical method (nonparametric or parametric) is more sensitive to an outlier? Why?