# Question

In Exercise 13.24 you compared credit card debts for college men and women using the Mann-Whitney test to compare medians. We'll use the same data again, but this time you will apply a randomization test to determine whether men and women in college have different median credit card debts. There were 19 men and 18 women in this data set. As part of the randomization test, the credit card debts were randomly assigned the label "man" or "woman," and the difference of the medians (men minus women) was calculated and recorded. This was repeated 1000 times. Figure A shows a histogram of the resulting differences of medians, and Figure B shows the resampling output.

a. Why are medians a better choice than means for comparing typical debts of men and women? Refer to the histograms given in Exercise 13.24.

b. The sample medians were $1250 for men and $725 for women, so the difference of men minus women was $525. This value is indicated with a red vertical line in the histogram. Is the observed difference unusually large? Explain.

c. The proportion of differences in the histogram that were greater than or equal to $525 is 0.245, as shown in Figure B. Our alternative hypothesis, that the medians are different, requires a two-sided test. However, with a significance level of 0.05, and knowing that the two sided p-value must be bigger than the one-sided value of 0.245, will we reject the null hypothesis? Explain.

a. Why are medians a better choice than means for comparing typical debts of men and women? Refer to the histograms given in Exercise 13.24.

b. The sample medians were $1250 for men and $725 for women, so the difference of men minus women was $525. This value is indicated with a red vertical line in the histogram. Is the observed difference unusually large? Explain.

c. The proportion of differences in the histogram that were greater than or equal to $525 is 0.245, as shown in Figure B. Our alternative hypothesis, that the medians are different, requires a two-sided test. However, with a significance level of 0.05, and knowing that the two sided p-value must be bigger than the one-sided value of 0.245, will we reject the null hypothesis? Explain.

## Answer to relevant Questions

Soda Does soda constitute a larger part of the diet for women than it does for men? A StatCrunch survey asked people to report the percentage of their liquid intake that is soda. The sample mean for the 169 females was ...Random samples of 30 professors of ethics and 30 professors in other disciplines (not ethics) were asked how many days it had been since they contacted their mothers; this included phone calls and face-to-face meetings. ...Looking at the data about contacting mom, for which group (ethicists or other) could you perform a log transform, and for which group could you not? Explain. Do older college baseball players tend to weigh more than younger baseball players? Our data are ages and weights for a random sample of 24 college baseball players. A scatterplot (not shown here) shows that the association ...Figure A shows information about a random sample of students' math SAT scores and GPAs at an unidentified four-Year College. a. Use the formula on the graph to predict the GPA for a person with a math SAT score of 600. b. ...Post your question

0