Recall that the variance of a binomial sample proportion, F, depends on the value of the population parameter,~. As a consequence, the variance of a sample percentage, (loop)%, also depends on p. Thus if you conduct an unpaired t-test (Section 7.1) to compare the means of two populations of percentages, you may be vlolatlng the assumption that a: = a;, upon which the t-test IS based. If the disparity in the variances is large, you w1l1 obtain more reliable test results umg the Wilcoxon rank sum test for independent samples.

In Exercise 7.18

(p. 361), we used a Student's t-test to compare the mean annual percentages of labor turnover between U.S. and Japanese manufacturers of air conditioners. The annual percentage turnover rates for five U.S. and five Japanese plants are shown in the table (p. 392). Do the data provide sufficient evidence to indicate that the mean annual percentage turnover for American plants exceeds the corresponding mean for Japanese plants? Test using the Wilcoxon rank sum test with a = .05. Do your test conclusions agree with those of the t-test in Exercise 7.182