!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
CHANCE News 2.16
(16 Sep to 2 Oct 1993)
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Prepared by J. Laurie Snell as part of the CHANCE Course Project
supported by the National Science Foundation and the New England
Consortium for Undergraduate Science Education.
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
===================================================
Chance itself pours in at every avenue of sense:
it is of all things the most obtrusive.
C.S. Peirce 1893
===================================================
ARTICLES ABSTRACTED
>>>>>==========>> John Paulos mentioned two sources of statistical observations that might stimulate discussion in a Chance course: "Harper's Index, a compendium listing a variety of interesting, whimsical, and often tendentious numerical factoids that appears in each month's Harpers magazine, and the USA snapshot in the lower left-hand corner of USA." I was not aware of these but I must say the September Index does contain amusing information such as: average amount an American would be willing to pay to see live dinosaurs: $15. The Oct 1 USA Today factoid in the sports section tells us that the rate of success of NFL field goals 55 yards or longer since 1969 is 1 in every 23 games at Denver mile high stadium and 1 in every 184 games at all other stadiums. This should provide a fine discussion question! <<<========<<
>>>>>==========>> Ecological regression in voting rights cases. Chance, Summer 1993, p 38. Stephen P. Klein and David A. Freedman Assume that Blacks constitute 20% of a city's voters but, within the five city council districts, less than 50% of the voters in any one district are Black. Then, if it can be shown that Blacks would choose a different candidate for a district office than would whites, one can argue that this situation violates the 1965 Voting Rights Act. This has led to court cases where it is required to show that a particular minority group favors a different candidate than do other groups. The authors discuss their experience testifying in such cases. For example, they testified for the defense in the El Centro school board contest. In this election Hernandez was the only Hispanic candidate. It was necessary to determine if Hispanics as a group voted differently than non-Hispanics. The data available was the number of Hispanic voters , the number of non-Hispanic voters, and the number of votes for each candidate for each district. A simple regression estimated the number of votes for Hernandez to be .30 x number of Hispanics + .05 x number of non-Hispanics. A similar regression estimated the number of voters for Newton (a white candidate) to be .33 x number of Hispanics + .44 x number of non-Hispanics. Thus it did not seem that the Hispanics as a group preferred Hernandez. However, when the procedure was carried out again, this time breaking up the non-Hispanic group into Blacks and non-Blacks, the regression estimated that 27% of the Hispanics voted for Hernandez and 17% voted for Newton. The judge in the case was confused by this difference and in this article the authors explain the reason for the difference in terms of the "ecological regression fallacy" closely related to the familiar Simpson's paradox. <<<========<<
>>>>>==========>> Proportionality: the view of the special master. Chance, Summer 1993, p9 David Baldus and George Woodworth When a death sentence is to be imposed, the courts require a review to determine if the punishment is in line with the use of the death penalty in previous cases that were similar to the one under consideration. The celebrated Robert Marshall case in 1988 (subject of a movie and the best selling book "Blind Faith") led the New Jersey Supreme Court to appoint Baldus "special master" to analyze New Jersey's death sentences and to recommend to the court methods for carrying out proportionality review. This article describes two different ways for determining appropriateness of the penalty. The first, a "frequency " view, looks to see if a reasonable proportion of previous cases similar to the one under consideration led to the death penalty. The second, a precedent approach, looks for cases as similar as possible to the present case that can be considered precedent setting. Obviously, there are complicated statistical problems both in determining which cases are similar and determining the cases which are the "most" similar. These problems are discussed in this article. <<<========<<
>>>>>==========>> Proportionality: an alternative view. Chance, Summer 1993, p 18 Herbert I. Weisberg In the situation described in the above article the New Jersey Attorney General's office retained the author of this article to provide advice on statistical and other methodological issues related to the report of the special master (Baldus). Weisberg agrees with Baldus and Woodstock on the general role of statistics in proportionality review. However, he finds difficulties with both their method of choosing the samples needed for the frequency approach and their methods for determining the closeness of a particular case for the precedent method. He would try, rather, just to direct the court towards cases that are clearly relevant and let the court itself apply its own qualitative analysis. <<<========<<
>>>>>==========>> Can TQM improve athletic performance? Chance, Summer 1993, p 25. Harry V. Roberts The author provides a variety of ways that Total Quality Management (TQM) can help an athlete train, and he describes experiences of his students in applying these techniques to their training. For example, a student had a free-throw success corresponding to a coin tossing process with probability .75 for heads (was in statistical control). He then intervened by changing his aim point to the back of the rim and increasing the arch. He found that he was again in statistical control, but now with a coin tossing process having probability .82 for heads. More complicated examples involve the use of regression to sort out the sources of variation and the use of randomized procedures to test different strategies <<<========<<
>>>>>==========>> Guest Commentary. Chance, Summer 1993, p 44. Barbara Everitt Bryant Bryant was the immediate past director of the Bureau of Census. She reviews the problem of the undercount in the 1990 Census and the two major decisions made: the present director Robert Mosbacher's decision not to adjust the census and New York Judge McLaughlin's decision upholding Mosbacher's. She then gives her own recommendations for avoiding these problems in the census of 2000. Bryant recommends that small annual surveys of several percent of the population be made to use in decisions that should not wait for the decennial census. Then, a simplified complete census would form the decennial census to be taken in 2000. For this census she recommends a follow-up on a large sample of non-respondents combined with statistical estimation of those not contacted. Then, a single census count, based on the best of enumeration and estimation, should be delivered to the President. Bryant believes that this will avoid the legal suits that have plagued recent censuses. <<<========<<
>>>>>==========>> The best NFL field goal kickers: are they lucky or good? Chance, Summer 1993, p 30. Donald G. Morrison and Manohar Kalwani The magazine that made streaks in sports a household word now tries to answer the question: do the observed data on the success rate of NFL field goal kickers suggest significant skill differences between kickers? The authors of this article begin by showing, using simple correlation, that the percentages made in one year are a poor predictors for the percentages to be made in the next year. Next they try to decide which model better fits the data: (a) a kicker's field goal attempts act like a Bernoulli process with a probability p of success with the same value of p for all kickers or (b) individual kickers performance is a Bernoulli process but the p- value for different kickers vary significantly. (The probability of a field goal depends on the distance from the goal, but here the authors are taking a weighted average to compute p for a typical kick) They conclude that there is no evidence that the NFL kickers have significantly different probabilities for success so that (a) is the more appropriate model. <<<========<<
>>>>>==========>> Thoughts on Innumeracy: mathematics versus the world. The American Mathematical Monthly, October 1993, p 732 Peter L. Renz with a reply by John Allen Paulos. Renz criticizes the treatment of some of the examples in Paulos' best selling book "Innumeracy" (Hill and Wang 1988). He questions the use of examples like "what are the chances that you just inhaled a molecule which Caesar exhaled in his dying breath?" or "what is the probability of heterosexual transmission of AIDS from an infected to an uninfected person per act of intercourse?" As evidence that such examples are too complex for a book meant for the lay reader, Renz does his own analysis of Paulos' problems getting different answers and in some cases claiming mistakes in the calculations of Paulos. In his reply, Paulos defends his approaches to the problems challenged by Renz and says that he stressed in the book that many assumptions have to be made along the way in such problems and so it is not surprising that different people analyzing the same problem will get somewhat different estimates. <<<========<<
>>>>>==========>> Day care may cut risk of leukemia, study says. The New York Times, 24 Sept 1993, A3. Lawrence K. Altman A study in the current British Medical Journal reports that children who attended a day care center for at least three months before the age of 2 had a 70 percent lower risk for childhood leukemia than children who did not attend a day care center. The researchers suggest that, like polio, having an early exposure to the disease acts to prevent a serious attack later. The crowded conditions of a day care center contribute to this early infection. <<<========<<
>>>>>==========>> Studies show futility of rushing some heart victims to hospitals. The New York Times, 22 Sept 1993, A23. Warren E. Leary Two studies reported in the current Journal of the American Medical Association support previous studies indicating that there is little point to rushing a heart attack victim to a hospital who did not respond to advanced cardiac life support at the scene of the attack. Only about 1/2 of 1 percent can be revived at the hospital and most of these still have permanent cerebral disability. Because of the large expense involved and the danger to others in a high speed race to the hospital, the authors recommend that policies requiring this final attempt be discontinued. <<<========<<
>>>>>==========>> Baptist damnations stir cries of bigotry. Washington Times, 19 Sept 1993, A1 Joyce Price. Quite a stir has been created by a report published in the Sept 5 issue of the Birmingham News that found 46.1 percent of the population of Alabama are "lost" and will end up in hell unless saved. This study is prepared annually to help the church decide where to expend it efforts. Evidently, the study is based on a formula, not divulged, to determine the percentage of the membership of various denominations likely to be saved. According to the newspaper report, these estimates are based on how closely those groups' beliefs matched Southern Baptist doctrine. This puts Catholics at a disadvantage and apparently dooms Jews, Hindus, Buddhists and members of other non-Christian religions to say nothing of non- believers. <<<========<<
>>>>>==========>> Study undercuts beliefs on preserving species. The New York Times, 28 Sept 1993, C1. William K. Stevens It has been assumed by conservationists that if a geographic area has a diversity of certain kinds of species such as birds and mammals the same will be true with other kinds of species such as plants, insects and microorganisms. This has led conservationist to concentrate on the preservation of areas with visible diversity of certain species (hot spots) with the hope that at the same time they will be protecting other less visible species. Recent research in Britain has shown that, at least in Britain, this "hot spots" theory does not always work. In particular, they found that, a significant percentage (16%) of rare bird species were found in cold spots and not in hot spots. Experts suggest that these findings may apply to a region like Britain where human activity has divided natural areas into rather small regions but may not apply to the very large natural areas found in tropical forests though they admit to some concern. The results and commentary on them will appear in Nature. <<<========<<
>>>>>==========>> The cancer risk in delayed pregnancy. Boston Globe, 27 Sept 1993, Science and Technology p 27. Judy Foreman Apparently, less educated women tend to get pregnant early and more educated women tend to get pregnant late or never. This gives the former an advantage as far as avoiding breast cancer is concerned. A study by Nancy C. Lee compared 4000 women with breast cancer with 4000 without and showed that three reproductive factors - young age at first childbirth, longer breast feeding and multiple children-all independently reduced the risk for breast cancer. The article points out that these increased risk factors have been known for some time, but only recently have theories been developed to try to explain why they occur. <<<========<<
>>>>>==========>> How demand surged for prostate test. The New York Times, 29 Sept 1993, C3. Gina Kolata A recent poll, taken to see how many doctors were using a widely promoted screening test for prostate cancer for men over 50, suggested that about ninety percent were. The test, called the P.S.A. test, has not been approved by the Food and Drug Administration and not been shown yet to be effective on long-term outcomes. It has both a high false positive and false negative result (about 20% in each case). The P.S.A. test came into prominence in the 1980's when it was shown to be effective in detecting tumors early. A recent study showed that it was twice as effective for this as a physical examination. Since 1980, the drug companies have made major campaigns to promote P.S.A. and most doctors interviewed in this article felt that it was overly promoted. One commented that it has been promoted so much that a large controlled experiment to really settle its longterm effectiveness will never be possible. <<<========<<
>>>>>==========>> Study cautions female smokers. Toronto Star, 23 Sept 1993, A4. Michael Smith The current issue of the American Journal of Epidemiology reports on a study carried out in Ontario showing that smoking is a greater risk factor for lung cancer for women than for men. The researchers studied 442 women and 403 men with lung cancer matched with 772 healthy women and men of similar ages and living in the same sections of the city. It showed that women who have smoked a pack of cigarettes a day for 40 years have a risk factor 28 times greater than a non-smoker, while for men the risk factor is only 10 times greater. The researchers mention three possible explanations: (a) women might smoke more cigarettes then men (b) there might be more current smokers and fewer ex-smokers among women or (c) women might be more susceptible to the effects of smoke than men. In connection with (c), some experts speculate that the smaller average body mass of women might mean that a cigarette has a greater impact on a women than on a man. <<<========<<
>>>>>==========>> Zoo animals stun Wall Streeters. Arizona Republic, 20 Sept 1993, E1 Russ Wiles A report on the first three months of a one year study that has five groups picking stocks: a money manager, a business team, a group from the radio station KTAR sponsoring the contest, a group of animals from the zoo, and a sports team. In the first three months the money manager is in first place with a gain of 22.4 percent, the business team is in second place, up 6.8 percent, KTAR group is up 5.3%, zoo animals are up 1.7% and the sports team is down 2.5 percent. The article explains the rather imaginative ways in which the animals pick their stocks, which are hoped to be random choices. Stay tuned. <<<========<<
>>>>>==========>> It's the cusp of Jupiter, you say? Buy! International Herald Tribune, 25 Sept 1993, Finance. Philip Crawford The article states that an increasing number of financial advisors are seeking the advice of astrologers and psychics. It says that some of the councilors can prove that their perceptions of market movements are quite accurate. It mentions, in particular, Michael Harding the British psychologist and astrologer who, in 1986 in his newsletter "The Investment Cycles Report", used astrological forecasting to predict, in September 1986, that Oct 19, 1987 would be the apex for the long climb in the market to be followed shortly thereafter by a sharp falloff. Oct 19 was the date of the second great stock market crash of the 20th century. Another person regularly consulted is Ruth Berger of Evanston, Illinois who calls herself a Hunch Power authority. <<<========<<
>>>>>==========>> ESP results boggle skeptics' minds. Toronto Star, 19 Sept 1993, B6 Jay Ingram The article claims that, while experts have been able to debunk most of the traditional ESP experiments, recent "ganzfeld" experiments have been more convincing. In such an experiment, the "receiver" lies in a chair wearing headphones listening to white noise, eyes covered with half ping-pong balls. After about 15 minutes, the brain deprived of images, begins to supply its own in the form of vivid dream-like imagery. A few meters away a "sender" sits in a similar chamber and stares at one of four pictures (the choice made by a computer) trying to send the image to the receiver. The receiver is shown the same four pictures and asked to say which one was sent. The study discussed in this article was conducted by Dr. Marilyn Schlitz on Julliard School music students. Twenty students did one trial each. Ten of them identified the correct picture out of a group of four. <<<========<<
>>>>>==========>> Obesity affects economic, social status; women fare worse. Washington Post, 30 Sept 1993, A1. John Schwartz The current issue of the New England Journal of Medicine reports on a study to determine how much of a disadvantage it is to be significantly overweight. "Overweight women complete about half a year less schooling, are 20% less likely to get married and earn $6,710 less per year than their slimmer counterparts. Obese women also have rates of household poverty 10% higher than those of women who are not overweight". The study found that overweight men suffer few adverse economic consequences but are 11% less likely to be married. The study identified, in 1981, 10,039 young people between the ages of 16 and 24 and followed them over a seven-year period. For purposes of comparison, the researchers also followed other young adults with chronic conditions such as asthma, diabetes etc. These groups did not differ socially or economically from people who were not overweight during the survey period. The researchers say that there is evidence of discrimination against people who are overweight, and the article concludes with " our data suggest that the extension of [the Americans with Disabilities Act] to include overweight persons should be considered." <<<========<<
>>>>>==========>> New rheumatoid arthritis therapy shows promise. The New York Times, 24 Sept 1993, A24. Gian Kolata The current Science magazine reports a study showing that drinking small vials of protein (collagen) from the breastbones of chickens decreases the symptoms of rheumatoid arthritis. There is no known cure for arthritis and previous established treatments have serious side effects. For this study 28 people with severe arthritis drank a small vial of chicken collagen mixed with their daily morning orange juice, while 31 others drank a dummy substance. After three months, the patients who received the collegen showed an improvement that was significant when compared to the control group. Experts warned that the study is a small one, and three months is not much time in the length of time a person suffers with arthritis, but they still were excited by the discovery. Animal studies had previously shown that injecting laboratory rats with collagen could cause an arthritis-like disease, and giving them collagen in the liquid form by mouth could prevent or alleviate such an arthritis. There is no explanation for this rather strange behavior of collagen. <<<========<<
>>>>>==========>> Smoking and Alzheimer's disease: protective effect or differential survival bias? Lancet, 25 Sept 1993, p 793 Jack E. Riggs Several studies have been made recently that suggest that cigarette smoking protects against the development of Alzheimer's disease. The introduction of bias due to differential survival in these studies was considered minimal due to carefully matched case-controls. Here, the author suggests that age-matched and sex-matched case- controls should not be assumed to guarantee matched gene pools. He points out that non-smokers live longer than smokers, in particular, more than twice as many male non-smokers reach the age of 77 than do smokers. The capacity of cells to repair DNA declines with age and this decline is thought to be related to age related disorders such as Alzheimer's disease. But, by selective survival, the smokers that reach the age of the subjects in the studies could have significantly better DNA repair mechanism than the non-smokers and this could account for their apparent protection against Alzheimer's disease. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! CHANCE News 2.16 (16 Sep to 2 Oct 1993) !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Please send suggestions to: jlsnell@dartmouth.edu >>>==========>>|<<==========<<<
>>>==========>>|<<==========<<<