!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Chance news 2.14
(25 July 1993 to 19 August 1993)
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Prepared by J. Laurie Snell as part
of the CHANCE Course Project supported by
NECUSE and the National Science Foundation.
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
===================================================
I have a great subject (statistics) to write upon, but
feel keenly my literary incapacity to make it easily
intelligible without sacrificing accuracy and
thoroughness.
Sir Francis Galton
===================================================
ARTICLES ABSTRACTED
>>>>>==========>> The Chances are good you're random. The New York Times, 27 July 1993, A12 J. Richard Gott: Letter to the editor This is a reply to the remarks of Eric J. Lerner (Op-Ed 14 July) on Gott's article in the May issue of Nature. Lerner suggested that a freshman in college could see the flaws in Gott's statistical arguments. I'll see if I can say what the freshman is worried about. Suppose that on a given day the manager of the local Inn wants to establish 95 percent confidence limits for the additional time that each guest will stay based on the time the guest has been in the hotel to date. The manager is willing to assume that the guests arrive independently and that the arrival times and lengths of stay are independent. Further, this process has gone on a long time. Then I think it follows that a guest is equally likely to be at any proportion of his or her stay in the hotel. In his Nature article Gott shows that this is the case when arrivals form a Poisson process and staying times are exponential. Gott observes that, from the uniform distribution for the proportion of time at the hotel, we can get a 95% confidence interval for the additional time the guest will be at the hotel by assuming that he or she is not in the first 2.5% nor in the last 2.5% of the total time in the Inn If T is the time the guest has spent in the hotel this confidence interval becomes the interval 1/39T to 39T. If the inn keeper makes predictions for all of the guests on a given day and watches how long they actually stay, these times should fall within the confidence interval about 95% of the time. Equivalently, if we choose a guest at random and calculate the 95% confidence interval for the remaining stay, we will be correct with probability .95. So this is a standard probability model where you can say something like what Gott says. Objections that have been made center on deviations from this setup. For example, if you are in the hotel and apply this to your own stay, you are not a randomly chosen guest. Lehmer suggests that his freshman would recognize this especially if we are in our own civilization applying Gott's method to find a confidence interval for how long we will be around. In addition Gott wants to go through life using his rule of 39 for everything in sight: how much longer this newsletter will be, how much longer Princeton University will survive, how much longer he will be at Princeton etc. Of course, Gott emphasizes that we may know more in certain situations that would allow us to do better (for example Gott is a full professor and so I assume he has tenure and this might be relevant). In this letter Gott remarks: "Popper's criterion for a good scientific hypothesis is that it makes predictions that can be falsified. My hypothesis can be falsified if we detect radio signals from civilizations around any of 1,000 nearby stars (there is a big NASA project testing this now), or if more than 2.7 trillion more humans are born. This is how science works: hypotheses are framed, predictions made and tested." Perhaps we should leave the subject on this lofty note. <<<========<<
>>>>>==========>> S.A.T. Scores improve for 2d consecutive year. The New York Times, 19 Aug 1993, A16 Karen de Witt The average S.A.T. verbal score was 424 for 1993, one point above last year's average, and the average math score was 478, two points higher than last year. It is remarked that "The number of students taking the test also rose (from 1,034,131 to 1,109,380), indicating that the higher scores are a real gain, not a fluke." S.A.T. began reporting ethnic breakdowns in 1976 and the minority groups have made significant gains while whites have remained about the same or, in the case of verbal scores, decreased. Women continue to do less well than men: verbal 420 compared to 428, math 457 compared to 502. This continues to be a puzzle given that women generally get better grades in high school and college in the same courses. <<<========<<
>>>>>==========>> Stories on cancer's causes are said to be misfocused; media overplay minor environmental threats to health, experts contend. Washington Post, 27 July, 1993, A6 Howard Kurtz The Center for Media and Public Affairs looked at 1,147 stories on cancer causation from 1972 through 1992 in leading popular journals and newspapers and on the three network evening news broadcasts. At the same time the Roper Center for Public Opinion surveyed 401 members of the American Association for Cancer Research concerning their opinion of media coverage in this area. The cancer experts ranked the top cancer threats as tobacco, sunlight, and diet, in contrast to the emphasis of the media on synthetic chemicals, food additives, pollution, radiation, and pesticides. Nicholos Wade, the New York Times's science editor is quoted as saying:"Often we're just doing our duty in following the activism of environmentalists, who make an issue of radon in houses or abandoned Superfund sites. Then it gets taken up in Congress and we have to cover it. We try to give the usual caveats and put things in perspective, but the public doesn't pay much attention to caveats." About 75 percent of the experts gave a high rating to the "New England Journal of Medicine" as a source of information on environmental causes of cancer, 22 percent had confidence in the New York Times, 9 percent in news magazines, and 6 percent in network news. <<<========<<
>>>>>==========>> Court ruling makes discrimination studies a hot new industry. Wall Street Journal, 13 August 1993, A1 Dorothy J. Gaiter In 1989 the supreme court handed down a decision that said that affirmative action programs, to set aside a portion of contracts to help minority businesses, were open to challenge unless the cities could show that there was a "significant statistical disparity" between the number of minority firms available to do government work and the number of minority firms actually hired to do so, and that this disparity was due at least partly to discrimination. This resulted in the creation of companies large and small to conduct studies to help cities decide if they could justify their affirmative action programs -- in fact, to help them show that they have been discriminating against minority industries! This article describes how these companies work and provides examples of the effect they have had on specific cities. It is stated that "the studies themselves are complex, generally consisting of two parts: a tedious statistical section, in which numbers are tracked down and crunched; and an anecdotal part, which consists of interviews with minorities, whites, men and women in business and the contracting officials they deal with." Apparently, it is often easier to "crunch the numbers" than to get minorities to jeopardize their future jobs by complaining about the past. <<<========<<
>>>>>==========>> You cannot exclude the explanation you have not considered. Lancet, 7 Aug 1993, pp 345-347 Facts figures & fallacies column Manula Datta Another excellent discussion in this continuing column on basic ideas in epidemiology. This time the topic is: what confounding means and how it is controlled for in epidemiology studies. Examples from actual studies are used as illustrations of the ideas presented. <<<========<<
>>>>>==========>> Toxic Terror; Phantom Risks. Science, 23 July 1993, p 407 Editorial by Philip H. Ableson This editorial lauds two recent books: "Toxic Terror" and "Phantom Risks". These two books address the issue of assessing risks of environmental hazards to human health in particular to cancer. "Toxic Terror" discusses the exaggerated claims of recent books and of the media, including Time, Life, Newsweek and the New York Times as well as of celebrities such as Jane Fonda. "Phantom Risk" explores two problems: (a) the disparity between the ease with which a controversy about a suspected hazard can begin and the difficulty in resolving the controversy and (b) the havoc in the courts that results from this disparity. "Toxic Terror", E. M. Whelan, Prometheus Books, Buffalo, NY, ed. 2, 1993 "Phantom Risks", edited by K. R. Foster , D. E. Bernstein and P. E. Huber, MIT Press Cambridge, MA. 1993 <<<========<<
>>>>>==========>> There is a much longer write-up of the opinions of Ableson on this topic in the July-August 1993 issue of "Public Perspective" published by the Roper Opinion Service. I was not aware of this journal and our library does not have it. However, here is a statement from the editor of the journal as to what it is all about. Editor's Note: When we launched Public Perspective four years ago, a primary objective was to explore the numbers we live by. The US collects an enormous amount of information bearing upon policy choices. Some of this is public opinion data, but many other types are gathered on a regular basis. Our experience with opinion surveys taught us how hard it sometimes is to measure what's on people's minds, even when skill and sophistication are brought to the task. Reality can be complex. Then, there are all manner of external contaminants -- such as those which enter the scene when an interest group decides to "do a poll" to show that "the people" stand squarely behind the group's position. We've become increasingly aware that many of the problems that affect polling afflict other types of measures, often in much the same way. Economic performance is complex, too, and various interests try to skew findings in this area to their own tastes. All across the arenas of social measurement, the press struggles to convey a balanced picture of findings that are quite intricate and often seem to contradict one another. Everywhere the ordinary citizen finds himself bombarded with competing claims about the effects of this or that. The pages of "Public Perspective" are open, then, to authors who can provide thoughtful guidance on how various statistics and other descriptions of social performance are properly assessed, which merit acceptance and which are highly misleading, etc. In this issue, Philip H. Abelson, who was long the editor of Science magazine and is still associated with that publication, reviews the arguments on the impact of various chemical substances on human health and explains the measurement problems risk assessments have encountered. Following that, Alan Reynolds, an economist who is director of economic research at the Hudson Institute, explores frailties of many widely-held descriptions of US economic performance. <<<========<<
>>>>>==========>> Study says some cancer may be linked to 'junk' DNA. Washington Post, 19 Aug 1993, A8 David Brown The Aug 19 issue of the "New England Journal of Medicine" reports a study by Krontiris and others identifying a risk factor for various forms of cancer in the form of a part of the DNA called a "minisatellite". This is a pattern of DNA "bases" that (with possible small variations) is repeated many times. The variability between people in the number of repeats in minisatellites is the basis of current DNA fingerprinting. These minisatellites by themselves have not been thought to have any specific task and are often said to be "junk" DNA. However they are increasingly becoming used as markers and possibly even causes for genetic diseases -- most recently for Huntington's disease. The results of this paper suggest that close to 10 percent of certain common forms of cancer (breast, colon, and bladder) arise from abnormalities in a particular minisatellite which the authors identify. Whether the satellite causes the diseases directly or indirectly, or is just a marker is not yet demonstrated. The authors had shown this association in an earlier case-control study but over half of the numerous studies after theirs were not able to replicate their results. Many of these studies were small and by combining them in a meta-study the authors found that the combined study did establish a significant genetic risk factor and suggest that some of the studies may have been too small to establish significance. They then combine their study with the other studies for their final analysis. An accompanying editorial in the same issue of the journal discusses the role of minisatellites in genetic diseases more generally. <<<========<<
>>>>>==========>> Australian study says AZT slows progression to full- blown AIDS The New York Times, 29 July, A20 Natalie Angier An AZT study in Australia is reported in the current "New England Journal of Medicine". This study looked at 993 patients infected with HIV. without symptoms of AIDS and with a CD4 count above 400. Those given AZT for three years were half as likely to progress toward AIDS as those given placebo. Experts point out that this does not contradict the recent Concorde study which indicated that the long range effect was not noticeable since the two trials used different ways to measure disease progression, different follow-up periods, and different end points on which to base their conclusions. These experts emphasize that all studies so far suggest a short time benefit for AZT but that it is just not likely to have the long term effects that they had hoped for. <<<========<<
>>>>>==========>> Cancer victims may sue over effects of radiation therapy. The Times, 7 August, 1993 Louise Hidalgo A group of more than 350 women have claimed crippling injuries such as loss of full use of hands and arms and even amputations as a result of radiation damage while being treated for breast cancer. Something like 200,000 patients a year are treated for breast cancer in England and most of these by radiation. Complications like these women have experienced are thought to be rare and if the group carries out a class action suit the degree of rarity will be tested in the courts. <<<========<<
>>>>>==========>> New study links liquor, cancer. Washington Post, 29 July, 1993, E2 A study conducted in Spain by a Harvard graduate student looked at 762 breast-cancer patients between the ages of 18 and 85. These patients were compared to 988 women of the same ages and regions who did not have cancer. The study found a 50% increase in the risk of breast cancer among women who drink at least one 8- ounce glass of wine each day and a 70% increase among those who had two glasses. The study is reported in the current issue of "Cancer Causes and Control" published by the Harvard School of Public Health. <<<========<<
>>>>>==========>> Restaurant employees' risky business. Washington Post, 3 Aug, 1993, Z7 Sandra G. Boodman A study reported in the current issue of the "Journal of the American Medical Association" leads to an estimate that workers in restaurants that allow smoking face a 50% greater risk of developing lung cancer than the general population. The author of the study Michael Siegel arrived at this conclusion by reviewing studies on indoor air quality in restaurants, bars, offices and homes with at least one smoker and six large epidemiological studies of mortality among food service workers. Siegal's results were consistent with a report of the Environmental Protection Agency on the dangers of secondhand smoke, but they were challenged by representatives of the Tobacco Institute. <<<========<<
>>>>>==========>> The search for sexual identity; false genetic markers. The New York Times, 2 Aug 1993, A15 Ruth Hubbard, Elijah Wald Ruth Hubbard is co-author with Elijah Wald of "Exploding the Gene Myth", and here she predicts that genetic causes for sexuality, like previous behavior traits, will not withstand further investigation. She suggests that the recent study that brought this back in the news (see the previous chance news) was not convincing. She points out that, of the relatively small number of siblings in the survey, almost a quarter did not have the markers and, in addition, the authors did not do the control of checking for these markers among heterosexual brothers. Hubbard traces the history of the treatment of homosexuality and concludes that showing that sexuality is inborn would not necessarily solve the problem of fair treatment of homosexuals. <<<========<<
>>>>>==========>> New doubts over AIDS infection as HIV test declared invalid. Sunday Times, 1 Aug 1993, Home news Neville Hodgkinson Australian researchers report in Bio/Technology on their concerns about the accuracy of the tests for the HIV virus. They claim that the "Aids test" is scientifically invalid and incapable of determining whether people are really infected with HIV. They assert that many people who appear to be infected can be suffering from such conditions as malaria and malnutrition that can produce a positive result in the test. The authors are concerned that the validity of the Elisa and Western blot tests are typically established on healthy populations, while their greatest unreliability is in populations where there is a lot of other ill-health such as in the third word countries -- the very countries where AIDS is the biggest threat. Of course the findings were welcomed by Peter Duesberg who said that helped to explain how "a false correlation" had been found between "HIV" antibodies and Aids. This article seems not yet to have been discussed in the American press. <<<========<<
>>>>>==========>> After a highly publicized death, second-guessing second opinions. The New York Times, 3 Aug 1993, C3 Dr. Lawrence K. Altman Jerome Kassirer, the editor-and-chief of the "New England Journal of Medicine" wrote an interesting editorial in the July 1 issue of the Journal on the problems of choosing between conflicting medical opinions. This editorial preceded the death of Reggie Lewis but was inspired by the conflicting opinions on how serious was his previous episode. When I read this I wondered if this topic was a reasonable chance topic, and I still have not decided if it is. However, this article gives a complete analysis of the Reggie Lewis case and discusses many of the same questions discussed by Kassirer on this topic but with the advantage of the later information. Kassirer raised some interesting technical chance questions, for example taking into account the likelihood of a false positive test related to the likelihood of the disease. I would recommend that if you do think this is a chance topic, in addition to reading this article you should go back to this editorial and to the article in the same issue of JAMA entitled "Sudden death in young athletes -- lessons from the Hank Gathers affair" by B. J. Maron. <<<========<<
>>>>>==========>> There were a number of articles coming out of the ASA meetings in San Francisco. Unfortunately, they were mostly baseball articles and the writeups did not make them sound very interesting. The following article did seem to be of some interest.
White male workers a shrinking minority. San Francisco Chronicle, 10 Aug 1993, A2 Ramon G. McLeod This is an account of a paper given by Jeffrey Passel showing that the numbers of white women, blacks, Hispanics and Asians will grow much faster than the number of white males during the next 20 years. McLeod estimates that by 2010 the white male work force will grow by about 3.3 million, white female workers by 7.3 million, blacks by 4.3 million, Hispanics by 8.8 million and Asian and Pacific Islanders by 5.1 million. Passel projects that the immigrant share of the U.S. labor force will double over the next two decades and discusses the effect of this on the ability of the U.S. economy to absorb a work force with the level of work skills and education of the various immigrant groups. <<<========<<
>>>>>==========>> Studies challenge idea that immigrants harm economy. Wall Street Journal, 13 Aug 1993, A1 Sam Enriquez This is a survey of the numerous studies on the effect of immigration on a regions economy. It is claimed that "a growing body of research challenges the popularly held view that immigrants are damaging California's much-battered economy" While studies find little evidence that immigrants take jobs from native-born workers, it is clear that the cost of social services fall most heavily on areas such as California where the immigration is high. This makes it possible to provide data on either side of the fence. However, economists claim that the solution is to change the federal policy of allocation of tax dollars so that California gets a bigger share of the taxes paid by immigrants than do areas with very little immigration. <<<========<<
>>>>>==========>> Florence Nightingale David. San Francisco Chronicle, 27 July, 1993, A19, Obituary Florence Nightingale David died at the age of 83. She was a well known statistician and helped found the Biostatistics department at the University of California Riverside. She wrote over a hundred articles on various aspects of statistics and nine books. Her book "Games, Gods, and Gambling" is a classic book on the history of probability theory. David was a native of England and named after another well known English statistician, Florence Nightingale. The obituary in the Daily Telegraph comments, "during the Second World War she studied air-raid bombing patterns to determine the locations of German launching silos. Since the Germans moved their silos from week to week, she had little success in this and described the war as having wasted six years of her life." David was a tireless advocate for women in mathematics. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Chance news 2.14 (25 July 1993 to 19 August 1993) !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Please send suggestions to: jlsnell@dartmouth.edu >>>==========>>|<<==========<<<
>>>==========>>|<<==========<<<