15.36 Rank-based correlation For data on two quantitative variables, x and y, an alternative correlation uses the rankings of the data. Let n denote the number of observations on the two variables. You rank the values of the x-variable from 1 to n according to their magnitudes, and you separately rank the values of the y-variable from 1 to n. The correlation computed between the two sets of ranks is called the Spearman rank correlation. Like the ordinary correlation, it falls between -1 and +1, with values farther from 0 representing stronger association.

a. The ordinary correlation can be strongly affected by a regression outlier. Is this true also for the Spearman rank correlation? Why or why not?

b. If you want to test the null hypothesis of no association, what value for the Spearman rank correlation would go in the null hypothesis?