1. (20 pts.) Select the most appropriate display for each of the following.

(a) Workers' income and whether they are black, white, Asian, or Hispanic:

(i) bar graph

(ii) histogram

(iii) pie chart

(iv) scatterplot

(v) back-to-back stemplot

(vi) side-by-side boxplots

(b) Rent charged and distance from the university for a sample of Oakland apart-

ments:

1. bar graph
2. histogram
3. pie chart
4. scatterplot
5. back-to-back stemplot
6. side-by-side boxplots

(c) Workers' marital status and whether they smoke or not:

(i) histogram

(ii) bar graph

(iii) scatterplot

(iv) back-to-back stemplot

(v) side-by-side boxplots

2. (10 pts.) Which one of the following is a paired study?

(a) Measure level of depression for a random sample of internet users and for a random sample of non-users

(b) Measure level of depression for a random sample of non-internet users; provide them with internet use for a year and then measure their level of depression.

3. (10 pts.) In general, which is more likely to contain the unknown population mean?

(a) a 90% confidence interval

(b) a 99% confidence interval

(c) both the same

4. (20 pts.) A biologist observes a sample of 81 cells to have mean dividing time 30.5

minutes and standard deviation 4.9 minutes.

(a) Construct a 95% confidence interval for population mean dividing time.

(b) Suppose a sample of only 11 cells had mean 30.5 and standard deviation 4.9.

Give three reasons why a 99% confidence interval forwould be wider than your

interval in (a).

5.(40 pts.) Researchers at the University of California Earthquake Center want to compare intensities of earthquakes at San Francisco Bay and at the Los Angeles Basin. The data, assumed to be independent random samples from the two areas, represent intensity on the Richter scale of tremors felt during several months in 1994.

(a) If researchers want to test if mean intensities differ between the two regions, state the appropriate null and alternative hypotheses.

(b) Use the two-sample t statistic in the output to characterize the p-value:

(i) extremely small

(ii) somewhat small

(iii) borderline

(iv) not small at all

(c) Is there statistically significant evidence of a difference?

(d) Would you expect a confidence interval for difference in population means to contain zero? (e) Are your conclusions still valid if one of the data sets is very skewed?

(f) Circle the part of the output that would help you decide if the Rule of Thumb for use of a pooled procedure is satisfied.

Output:

Two-sample T for SF vs LA N Mean StDev SE Mean SF 14 2.59 1.02 0.27 LA 16 2.13 1.09 0.27 Difference = mu SF - mu LA Estimate for difference: 0.454 %95% CI for difference: (-0.339, 1.248) T-Test of difference = 0 (vs not =): T-Value = 1.18