B4. For this exercise, you will be running a Mann-Whitney U test, comparing smokers and nonsmokers on an ordinallevel variable, drunk. This variable measures how frequently in the prior month the woman had consumed enough alcohol to get drunk (drunk), from never (code 1)

to 10 times (code 5). Begin by running a simple Crosstabs (following steps described previously) so that you can examine the distribution of values for drunk for smokers and nonsmokers. Instruct the computer to show observed frequencies and column percentages. Then run the Mann-Whitney U test via Analyze ➜ Nonparametric ➜

Two Independent Samples Test. Move drunk into the slot for Test Variable List, and smoker (the original smoking variable) into the Grouping Variable slot. Define the smoking groups as code 0 for Group 1 and code 1 for Group 2.

Select Mann-Whitney U as the Test Type. Then click OK and answer these questions:

(a) What percentage of smokers versus nonsmokers never got drunk in the prior month?

(b) For the overall sample, what percentage of women got drunk at least once in the month prior to the interview?

(c) What is the mean rank of “scores” for smokers versus nonsmokers?

(d) What is the value of the U statistic?

(e) Is this significant for a .05? At what level?

(f) What does the result mean? (g) Could we have analyzed these data using a chi-square test? Why or why not?