!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Chance news 2.13
(5 July 1993 to 24 July 1993)
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Prepared by J. Laurie Snell
as part of the CHANCE Course Project supported
by NECUSE and the National Science Foundation.
Please send suggestions to
dart.chance@dartmouth.edu
Current and previous issues of chance news and full text
of the newspaper articles can be found on the chance
gopher. Just point your gopher to chance.dartmouth.edu
===================================================
Vulcanology is the Cinderella science which only marches
forward on the ashes of catastrophe.
G.A. M. Taylor
===================================================
ARTICLES ABSTRACTED
>>>>>==========>> Lessons in reducing volcano risk. Nature, 22 July 1933, 354, p.277. Robert Tilling and Peter Lipman Tilling and Lipman review the major volcanic disasters throughout the world and the methods for predicting volcanoes. Volcano predictions have been based primarily on pattern recognition and, while there have had some successes, these methods have limited general application. The authors feel that the current studies being made of the source mechanisms that produce the precursory patterns observed will lead to more general successes in earthquake prediction. In another article in this issue of Nature, David Swinbanks notes that the recent devastating earthquake in the Northern coast of Japan highlighted gaps in Japan's seismic networks used for earthquake predictions and disaster prevention. <<<========<<
>>>>>==========>> Katherine Campbell provided the following contribution: Risk Analysis and Management. Scientific American July 1993, pp.32-41. M. Granger Morgan There's an interesting article about risk assessment and risk management by M. G. Morgan in the July issue of Scientific American. Among the points of interest to statisticians are the observations about factors other than expected number of deaths or injuries that people use to evaluate risks, and bases other than cost/benefit for rules to judge whether and how to deal with an issue (namely "rights based" and "technology based" rules.) Like Mary Gray's article (reviewed in the last issue of chance news) this discussion suggests that statistics is more marginal in policy making than we might wish. The most interesting part of the article to me was Granger's observations about risk communication, and about their ability to make good decisions when given balanced information and enough time to reflect on it, which is very compatible with recent work in educational "restructuring" movements. <<<========<<
>>>>>==========>> Bill Peterson wanted to further honor the solution of Fermat's Last Theorem by reminding us of the following urn problem: You have two urns A and B. Can you put 50 black and white balls in each urn in such a way that when you choose, with replacement, m > 2 balls from each urn the probability of getting all black balls from urn A equals the probability that all the balls drawn from urn B are the same color? Answer: Ask Andrew Wiles If you got interested in the Ann Landers party problem from mentioned in the last chance news you can read more about the problem in the July 17 issue of Science News. <<<========<<
>>>>>==========>> Flaw is discovered in crucial research on AIDS treatment. The New York Times, 22 July 1993, A,1 Lawrence K. Altman Last February, Boston scientists found what they suggested "may be the Achilles heel of H.I.V." The findings were more dramatic because they were made by a Harvard medical student Kang Chow working in the laboratory of Dr. Martin Hirsch. Chow's idea was that the HIV virus develops resistance to drugs like AZT by developing a mutation in the enzyme that helps it replicate. However, when it does this the enzyme becomes less effective. Chow theorized that a combination of different drugs would result in enough mutations to make the drug ineffective. Laboratory tests seemed to verify this and a study on humans to test this theory attractive a large number of volunteers because of the widespread publicity associated with the discovery. Now it appears there were some subtle but important errors were made in the interpretation of the laboratory data. The concept of combination of drugs is still thought to be a promising approach and experts do not believe that the finding of the error is any reason to stop the study though is has dimmed some of the optimism for the outcome. This is regarded as another example of the media getting carried away with a preliminary result and raising false hopes. For example, Chow was Peter Jennings "person of the week" and movie companies wanted to do the story of his life. <<<========<<
>>>>>==========>> Horoscopes for humanity? The New York Times, 14 July, 1993, A,19. Eric J. Lerner An op-ed that makes a blistering attack on Richard Gott's article in May 27 issue of Nature on how to find 95% confidence intervals for such things as the duration of the human species. (See the last two issues of chance news for more details of Gott's method for obtaining these confidence intervals. Lerner wonders how Nature could have accepted this paper. He remarks that "Dr. Gott's pseudo-statistics would not pass muster in a freshman math class." He states that Gott, by using only a sample of people alive today, cannot conclude anything about the whole system from such a non-random sample. <<<========<<
>>>>>==========>> The odds were 1-in-185 million, and he won. Boston Globe, 19 July, 1993, Science and Tech., p 25. David Leonhardt This article, based on conversations with Herman Chernoff and Hal Stern comments on Leslie Robbins winning the 111 million dollar Powerball lottery July 7. The Powerball is a 14 state lottery and to win players must choose the 5 numbers drawn from 45 white ping pong balls numbered from 1 to 45 and a sixth number drawn from a separate container of 45 red balls numbered from 1 to 45. This provides 55 million combinations. If the 68.4 million players chose numbers randomly, there would be about a 36 percent chance for a single winner. Herman Chernoff, who has written about lottery strategies (How to beat the Massachusetts Numbers Game," Mathematical Intelligencer, 3, 166-172) observes that people don't choose random numbers thus decreasing their chance to be a single winner. Hal Stern who wrote on lotteries with Tom Cover ("Maximum Entropy and the Lottery", Journal of the American Statistical Association, 84,980-985) says that people tend to avoid numbers over 40 and pick calendar dates and such things as special sports numbers. Robbins was clearly not trying to pick unpopular numbers since he chose popular low numbers and only one number that was not a day of the month. The 1 in 185 million chance of winning the 111 million dollars was obtained by multiplying the 1 in 55 million chance of choosing the right numbers by the 3 in 10 chance of being the only winner. The writer of this article comments "This is one of the basic tenets of probability -- that the chance of two events both occurring can be determined by multiplying the odds of one occurring by the odds of the other occurring" Tom Cover comments that the Powerball lottery is more unfair than other lotteries which typically just require that 6 numbers agree without order counting since in the Powerball lottery the last number chosen must be predicted. The articles provides a lesson in the use of the Poisson distribution in estimating the chance that there is a single winner. Other questions and references about lotteries can be found articles by Richard Paulson and James Hanley in the American Statistician Vol. 46, No. 3. Paulson's article is called "Using Lottery Games to Illustrate Statistical Concepts and Abuses", and is on pp. 202-204 and Hanley's is titled "Jumping to Coincidences: defying odds in the realm of the preposterous" pp. 197-201. <<<========<<
>>>>>==========>> Lancet has a column called Facts, figures, & fallacies. What's the denominator? Lancet, 10 July, 342, 97-98. Cesar G. Victoria The author reminds doctors that when they leave off the denominator in such statements as "most people with kidney stones drink less than six pints of liquid a day" they do not convey the relative risk that they intend. The author then summarizes the many different choices for the denominator and epidemiological terminology for describing disease occurrence and evaluating risk factors. Making comparisons. Lancet, 17 July, 342, 157-159 Jean Ann Grisso A concise summary of the various epidemiological study designs with their advantages and disadvantages. It is pointed out that there often is a natural evolution in the use of these studies. For example, when the AIDS epidemic began, the first published reports were small numbers of clinical observations (case studies) followed by epidemiological studies of the prevalence of the disease in different population groups (cross-sectional and ecological studies). Then came case-control and cohort studies to identify risk factors. Now clinical trials are being performed to evaluate treatment and preventive measures. <<<========<<
>>>>>==========>> Caring for Women's Health -- What is the problem? New England Journal of Medicine, 22 July, 329, 271-272 Marcia Angell The 22 July issue of the New England Journal of Medicine contains three articles about the health of women. One article reviews what is known about differences between men and women in relation to heart disease. A second article reviews the recent reversal of Food and Drug Administration policy on excluding women of reproductive age from participating in drug studies. A third article discusses issues raised by an proposed amendment to the Public Health Service Act that would require that research sponsored by NIH include adequate numbers of women and minorities to determine whether they respond differently from white men. This editorial and the articles themselves raise a host of problems about designing clinical trials to get information about subgroups of the population. <<<========<<
>>>>>==========>> Chance are... The Atlantic Journal and Constitution, 15 July, 1993, G, 1 Bo Emerson A discussion of "What are the odds?" inspired by recent books giving the odds for lot of everyday things. One author, Rorvik says that compiling the book changed his mind about some dangers. He notes that the flu kills about 70,000 Americans each year, AIDS about 30,000 and adverse reaction to prescription drugs about 125,000 Americans. The three books that are the source for this discussion are: What Are the Chances?: Risks & Odds in Everyday Life. James Ruke, Carol Pub Group, 1992, pap. $8.95 ISBN: 0-8065-1334-9 What the Odds Are: A to Z on Everything You Hoped or Feared Would Happen. Les Krantz, HarpC, 1992, pap. $14.00 ISBN: 0-06-271521-6 0-06-273060-6 What Are the Chances?: Risks, Odds, & Likelihood in Everyday Life Siskin, Bernard; Staller, Jerome; Rorvik, David NAL-Dutton, 93. pap. $4.50 SBN: 0-451-17488-7 <<<========<<
>>>>>==========>> Report suggests homosexuality is linked to genes. The New York Times, 16 July, 1993, A,1. Natalie Angier An article in the 16 July issue of Science reports on a study led by Dean Hamer which suggests that sexual preferences has a genetic component. The study began by looking at the family histories of 114 homosexual men. The researchers discovered a higher than expected number of gay men among the men's maternal uncles and male cousins who were sons of their mother's sisters. This suggested a gene or genes that pass through the maternal line and therefore through the X chromosome. This led them to study the genetic material from 40 pairs of brothers who were gay. They found that in 33 of the pairs the brothers displayed the same cluster of five markers bunched into a small region on the X chromosome. Since there should be a fifty percent chance that two brothers share the same allele by descent this finding is regarded to be significant. The authors are careful to say that this study will have to be replicated and, if correct, is surely only one part of the answer etc. A similar finding two years ago that suggesting anatomical difference between the brains of gay and straight men received a lot of publicity and let to heated debates, mixed emotions etc. <<<========<<
>>>>>==========>> Wilful public misunderstanding of genetics. Lancet, 22 July 1933, 364, p 281 John Maddox An editorial commenting on the way that the press greeted the news of the Hamer study on genetic markers for homosexuality. The Sunday Telegram filled a page with three articles with headlines "Born to be gay", "A lot of mothers are going to feel guilty" and "The gene genie comes out fighting" and featured another article on "an extraordinary alliance of gay rights activists and the Right to Life lobby". The editorial remarks that the many cautionary remarks in the research article did not seem to help very much. <<<========<<
>>>>>==========>> Sexual orientation and the X. Nature, 22 July 1933, 364, p 288. Mary-Clair King An evaluation of the Hamer study on genetic markers for homosexuality in terms of the statistical issues involved and the additional research necessary to clarify this issue. The Hamer study was intentionally biased to maximize the chance of detecting any X-linked genetic influence. Significant results in similar genetic linkage studies have sometimes later been shown to be the result of chance. The author comments that this study was carried out with unusual care to prevent this. However, this study says nothing about how frequent these markers are in the general male homosexual population etc. <<<========<<
>>>>>==========>> Scans 'link' with left handedness. Guardian, 16 July, 1993, Home Page, p9. Chris Mihill A study in the British Medical Journal suggests that ultra-sound scans used to assess the health of a baby in the womb may lead to an increase in left-handed children. The study followed children aged eight and nine who had received scans between 1979 and 1981 and compared them with children who had not had scans. They found no signs of neurological damage or difference in growth, but of 861 children given ultrasound, 19 percent were left-handed, as opposed to 15 percent in the 802 unscanned children. Hardly significant but to make it sound better it is reported that the scanned group was 30 percent more likely to be left handed than might have occurred by chance. <<<========<<
>>>>>==========>> Success is reported with prototype AIDS vaccine. The New York Times, 9 July, 1993, D,18. Warren E. Leary A report in a recent issue of Lancet provided human tests of a vaccine that in tests on chimpanzees had produced high levels of antibodies that protected the animals against infection after exposure to the HIV virus. The researchers gave the vaccine to 28 human volunteers in a series of three injections over 32 weeks. Ten were given 300-microgram doses, ten 100 micrograms, and eight were given placebo. In the high- dose group, 9 of the 10 developed antibodies that neutralized the AIDS virus in test tubes, compared to 5 in the low-dose group. This test used a strain of HIV virus that does not occur in the general population and the next step is to try a variation of the vaccine based on a strain of HIV that more closely resembles the type found in the United States and Europian HIV positive population. <<<========<<
>>>>>==========>> Study find less family breast cancer risk. Boston Globe, 21 July, 1993, National/Foreign, p1, Betsy A. Lehman A study reported in the 21 July issue of the Journal of the American Medical Association shows that daughters and sisters of women with breast cancer have less chance of getting breast cancer than previous studies had suggested. Researchers in this study followed 117,988 women in the Nurses' Health Study from 1976 to 1988. Of the 2,389 breast cancers detected in the women, only 2.5 percent of the cancers could be attributed to a family history of the disease. Women whose mothers developed cancer later in life faced less risk than those whose mothers were diagnosed at a young age. This study shows that the recent success in identifying genes that predispose women to breast cancer will not solve the general problem of breast cancer. The article reminds us that prospective studies like this (gather information from the subjects first and follow their medical histories) can be more accurate than retrospective studies (ask people who have the disease about their previous medical and family histories) because in retrospective studies people who have the illness are more apt to remember similar illnesses in their relatives. <<<========<<
>>>>>==========>> Study suggest way to greatly reduce heart bypass deaths. Washington Post, 21 July, 1993, A,3 Spencer Rich A study in the 21 July Journal of the American Medical Association that showed that differences in hospital death rates from bypass surgery are not the result of chance differences or how sick the patients are but rather the result of differences in the quality of hospital care. It is stated that the death rate from bypass operations could be reduced by 54 percent if patients went to the hospitals with lowest death rates. The researchers studied 115 California hospitals that had five or more bypass operations in any year from 1983 to 1989. It is argued that chance variation would be inconsistent with the consistent low death rates of certain of the hospitals and the consistent high death rates of others. Why not just a small variance? <<<========<<
>>>>>==========>> 'Bye to 'Frye': High court sets standards for admitting expert testimony. New York Law Journal, 8 July, 1993, Outside Councel, p1. P. Kevin Castel This is a discussion of the recent supreme court decision setting standards for admission of expert testimony. It remarks that if you read the Wall Street Journal report you would have thought it was on the side of admitting junk science but if you read the New York Times you would have felt it was a victory for those who wanted to limit the use of unproven scientific evidence. It is expected that this new Daubert rule as it is now called (Judge should decide if the scientific evidence is supported by reasonable scientific methods), will be refined by federal cases as judges try to determine how to decide if scientific evidence fits the rather subjective criteria set down by the Supreme Court. While the Frye criteria (evidence must be accepted by the scientific community) was rejected by the federal court it is expected to continue to be applied in the state courts that have followed it. <<<========<<
>>>>>==========>> DNA and police files. Lancet, 17 July, 1993, 342, p.130. Howard Cooke A commentary on the Royal Commission proposal to establish a national DNA database. To set up this data base, police would be allowed to take samples from suspects for a "serious arrestable offense" without consent even when DNA evidence is not relevant. If the suspect is convicted the data from the sample will be deposited in a national database for future use. If the suspect is found not guilty the information would be deposited anonymously, for statistical purposes, in a database supervised by an independent body. The author of this commentary is concerned about the resulting bias in the data base obtained from those convicted of a serious offense and recommends that a more well designed date base be established. <<<========<<
>>>>>==========>> Breast Cancer Incidence. Science, 16 July, 1993, 261, letters. Michael Swift A recent article in Science remarked that "about three quarters of the 4% annual rise in breast cancer incidence in the 1980's was due to the expanding use of mammography machines". The author of this letter states that this issue has been studied directly and the data tell a different story. He cites these studies and remarks that they claim that expanded use of mammography could have accounted for less than one-third of the increase seen in the 70's and 80's. <<<========<<
>>>>>==========>> Big health gap, tied to income, is found in U.S. The New York Times, 7 July, 1993, A,1. Robert Pear A study reported recently in the New England Journal of Medicine analyzed data from more than 13,000 death certificates to compare class differences in mortality rates in the U.S. in 1960 and 1980. The study showed that poor or poorly educated persons have higher death rates than the wealthy or better educated and that these differences increased between 1960 and 1980 and presumably are still increasing. The same has been observed in countries that have national health insurance suggesting that such insurance itself will not solve the problem. Blacks have a higher death rate than whites but another study reported that education seemed to be more important than race as a factor influencing death rate and that the difference can be explained in terms of socioeconomic status as measured by education and income. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Chance news 2.13 (5 July 1993 to 24 July 1993) !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Please send suggestions to: jlsnell@dartmouth.edu >>>==========>>|<<==========<<<
>>>==========>>|<<==========<<<