1. A 95% confidence interval for the mean of a sample of positive numbers can include negative values if the sample size is small and the data are highly skewed. Is it possible for the nonparametric confidence interval for the median based on a sample of positive values to include negative numbers in its range? Why or why not?
2. Does taking logarithms to remove skewness improve the sampling properties of a nonparametric interval for the median? Why or why not?