Project 1. Reporting a medical study. Many of the major articles in medical journals concern statistically designed studies and report the results of inference, usually either P-values or 95% confidence intervals. You can find summaries of current articles on the websites of the Journal of the American Medical Association (jama.ama-assn.org) and the New England Journal of Medicine (www.nejm.org). A full copy may require paying a fee or visiting the library. Choose an article that describes a medical experiment on a topic that is understandable to those of us who lack medical training—anger and heart attacks and fiber in the diet to reduce cholesterol are two examples used in Chapters 21 and 22. Write a two-paragraph news article explaining the results of the study.

Then write a brief discussion of how you decided what to put in the news article and what to leave out. For example, if you omitted details of statistical significance or of confidence intervals, explain why. What did you say about the design of the study, and why? News writers must regularly make decisions like these.

Project 2. Use and abuse of inference. Few accounts of really complex statistical methods are readable without extensive training. One that is, and that is also an excellent essay on the abuse of statistical inference, is “The Real Error of Cyril Burt,’’ a chapter in Stephen Jay Gould’s The Mismeasure of Man (W. W. Norton, 1981). We met Cyril Burt under suspicious circumstances in Exercise 9.27 (page 201). Gould’s long chapter shows that Burt and others engaged in discovering dubious patterns by using complex statistics. Read it, and write a brief explanation of why “factor analysis’’ failed to give a firm picture of the structure of mental ability.

Project 3. Roll your own statistical study. Collect your own data on two categorical variables whose relationship seems interesting. A simple example is the gender of a student and his or her political party preference. A more elaborate example is the year in school of a college undergraduate and his or her plans following graduation (immediate employment, further study, take some time off, …). We won’t insist on a proper SRS.

Collect your data and make a two-way table. Do an analysis that includes comparing percentages to describe the relationship between your two variables and using the chi-square statistic to assess its significance. Write a description of your study and its findings. Was your sample so small that lack of significance may not be surprising?

Project 4. Car colors. We have heard that more white cars are sold in the United States than any other color. What percentage of the cars driven by students at your school are white? Answer this question by collecting data and giving a confidence interval for the proportion of white cars. You might collect data by questioning a sample of students or by looking at cars in student parking areas. In your discussion, explain how you attempted to get data that are close to an SRS of student cars.