Answers:  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.

1.  43 studies were included in the meta-analysis.  The researchers only
    chose studies that were published from 1966 to 1994, written in
    English, evaluated the effects of weight change or weight cycling on
    humans or animals in depth, and emphasized human subjects.  (p.1196-
    "Methods")

2.  The newspaper articles and the report were very explicit in stating
    the limitations of the studies used.  They indicated that these studies
    failed to show aspects such as the effects due to intentional and 
    nointentional weight loss and the differences in weight loss effects
    in obese people as opposed to those of normal weight.  The report
    even called for more study on the psychological aspects of weight loss,
    while the article suggests for further research in the long-term health
    effects of yo-yo dieting.  (p.1199-3rd pgrh, p.1201-"Conclusions",
    10/26 article-7th and 10th pgrhs)

3a. Not all studies are in Tables 1 and 2 because there are several studies 
    that did not define what weight cycling is and because some of the 
    studies may not have concentrated on metabolic or psychologic effects
    or on mortality as an outcome.  Not all the studies in Tables 1-2 are
    in Table 3 because of the previous reason (not focus on metabolic
    effects on weight cycling in humans).  (Tables 1-3)
 b. If the studies are small, then one would more likely not find any
    significant effects because the confidence interval would be wide.  Large
    studies, however, would more likely show yes effects because the
    confidence intervals are small to test the hypotheses.
 c. False.  There are only two studies that could be considered large 
    (classified as those with over 100 subjects) with
    "yes" effects, compared with around 4 smaller studies that are known.
    On the other hand, the smaller studies did tend to show "no" effects.
    (Table 1, Table 2, and References)
 d. Having many smaller studies in this meta-analysis do give a misleading
    picture of the situation because the sample sizes are very small to get
    results that are more likely on the average and are not outliers.

4.  The newspaper articles are very good at explaining the limitations on
    such kind of analysis but are not explicit in the kinds of studies
    that were used (no description of the studies themselves in terms of
    sample sizes used, etc.)

5.

6.