Questions:  1) "Panel finds no major risk from 'yo-yo' dieting", 20 Oct 94,
                NY Times, p. A10.
            2) "Study on effects of yo-yo dieting is faulted", 26 Oct 94,
                NY Times, p. B12.

Journal Article:  "Weight cycling", National Task Force on the Prevention and
                   Treatment of Obesity, 19 Oct 94, JAMA, v. 272, no. 15,
                   pp. 1196-1202.

This study is a "meta-analysis," a review of previous studies on a 
particular topic.  The reason for performing a meta-analysis is because 
individual studies may come to different conclusions, but a large number of
studies may show a clear pattern.

1.  How many studies are included in the meta-analysis?  How did the
    researchers decide how many studies to include?

2.  Obviously, the meta-analysis is only as good as the original studies
    referred to in the report.  How is this issue addressed in the
    newspaper article and the report?

3a.  How come there are 43 studies analyzed but only 27 in Tables 1-2.  Plus,
     why not all of these 27 not in Table 3?
 b.  Use your knowledge of statistical confidence intervals and hypothesis
     tests to explain how this could occur.
 c.  True or false:  the studies that show "yes" effects in Table 3 tend to
     be in larger studies (with more subjects).  Note:  to answer this, you'll
     have to refer both to the references to the report and to Table 1.  If
     the large studies tend to show "yes" effects and the small studies tend
     to show "no," it could be that the effects are real but too small to show
     up in a small study.
 d.  Based on your answer to the previous question, do you think this is a 
     serious concern here?

4.  Do you think that the newspaper article adequately covers the strengths
    and weaknesses of the report?

5.  If you could add one sentence to the newspaper article, what would it be?

6.  If you had to remove two sentences from the newspaper article, what would
    they be and why?