CHANCE News 2.02
              (6 Jan to 21 Jan 1993)

Prepared by Laurie Snell

 Please send suggestions for  articles or format to 



>>>>>==========>> From David Kratz: Tenneco chief discloses brain cancer. The New York Times, 21 Jan 1993, D1 Thomas C. Hayes Michael H. Walsh, chair & CEO of Tenneco, age 50, has inoperable brain cancer; experts said that similar illnesses can be cured but that the odds for recovery are not good. "Mr. Walsh said that he expected to beat the odds, and that he was confident of recovery and a long tenure at Tenneco. 'My approach to life has been to deal head-on with the problem and not to pay any attention to statistical probability,' he said." Dave comments "While I certainly wish him good luck, I wish he and other CEO's would recognize unpredictability." <<<========<<

>>>>>==========>> Coin-tossing computers found to show subtle bias. The New York Times, 12 Jan 1993, C1 Malcolm W. Browne Report of a recent article in Physical Review Letters showing that five of the standard methods for generating random numbers can give incorrect estimates if pushed too by the power of parallel processing computers. The investigators were studying the Ising model for phase transition. Monte Carlo methods are routinely used to study this model. In the study reported the simulation was carried out in some special situations where the exact solutions are known and found to be significantly off. The article includes comments from mathematicians Ron Graham, Persi Diaconis and James Reeds. The latter quotes von Neumann to the effect that "anyone who believed a computer could produce truly random sequences of numbers was living in a state of sin." (Unpleasant results from standard random number generators were also found by David Griffeath and Bob Fisch in developing their probability demonstration GASP when they tried to simulate random walks. ) <<<========<<

>>>>>==========>> Calculators are found to improve performance on SAT math test. Wall Street Journal, 11 Jan 1993, page B8 William M. Bulkeley Preliminary testing had suggested that the Educational Testing Service had made their new tests in such a way that the use of the calculator would not have a significant effect on the scores achieved. A field test of 200,000 students last spring showed that they were wrong and that student who use the calculator on average do 10 to 20 points better than those who did not. As a result ETS will lower the scores "so that there are no difference from year to year due to calculator use." The article discusses the various concerns about possible inequities due to a the introduction of the use of calculators on SAT exams. <<<========<<

>>>>>==========>> U.S. Ties Secondhand Smoke to Cancer. The New York Times, 8 Jan 1993, p A14. Warren E. Leary An Environmental Protection Agency report states that secondhand tobacco smoke has been "conclusively shown" to increase the risk of lung cancer in nonsmokers. An estimated 3000 nonsmokers a year die of lung cancer from second hand smoke. 20% of all lung cancers caused by factors not attributable to direct inhalation of tobacco smoke are said to be attributable to indirect, environmental smoke. The EPA now classifies environmental smoke as a Class A carcinogen, the same category as asbestos and benzene. 140,000 smokers per year die of lung cancer caused or aggravated by lung cancer, putting a smoker's risk of developing cancer at between 1-in-10 and 1-in-20. Non-smokers face a 1-in-1000 risk of lung cancer from environmental smoke. Higher risks are incurred by those exposed to second-hand smoke in enclosed homes, small rooms and automobiles. Spouses of people who smoke at home have a 2-in-1000 risk. <<<========<<

>>>>>==========>> Pollution-Weary Minorities Try Civil Rights Tack. The New York Times, 11 Jan 1993, pA1 Roberto Suro Describes a growing body of evidence (leading up to a June 1992 Environmental Protection Agency report) indicating that racial and ethnic minorities suffer most from pollution. Race, even more than economic factors, is seen as the most important common characteristic of communities exposed to environmental hazards. For example, lead is one of the most widespread hazards in minority communities. Federal research has shown that lead from plumbing, paint and soil contamination reaches poor children of all races. But a 1988 study concluded that black children, regardless of family income, were more likely than white children to have unacceptable lead levels in their blood. Other studies have found similar patterns in location of hazardous wastes and exposure to air pollutants: large minority population is the most "statistically significant factor" that exposed communities share. "The possibility that these patterns resulted by chance is virtually impossible." (Includes bar graph displays). <<<========<<

>>>>>==========>> Deal yourself a riffle for that sure shuffle. Sunday Telegraph 17 Jan 1993 Robert Matthews A discussion of how Persi Diaconis work on "seven shuffle suffices" has effected shuffling in Casinos in England. <<<========<<

>>>>>==========>> Mammography campaigns draw in the young and healthy. The New York Times, 10 Jan 1993, Section 4, page 6 Gina Kolata Televisions ads for mammograms and programs on breast cancer seem to be encouraging the wrong women to get mammograms, namely those in their 20's and 30's rather than in their 60's or 70's. The article discusses ways to make older women more aware of the problem and calm the fears of younger women. In the latter case one doctor is quoted as saying that she tells a younger woman that "a 70-year man has a risk of breast cancer that is four times that of a 20-year-old woman. A 70-year old man does not worry about breast cancer so neither should a 20-year old women." <<<========<<

>>>>>==========>> Safety of genetic evidence is put to test. The Independent, 9 Jan 1993, HN 8 Steve Connor Another account of the debate about the use of DNA fingerprinting in the courts that grew out of the article in Science by Lewontin and Hartl raising concerns about it's use. British scientists do not understand what the fuss is all about. It is stated that Hartl now accepts the conclusions of the recent National Academy of Sciences report and endorses DNA fingerprinting but Lewontin has not changed his position. <<<========<<

>>>>>==========>> New definition for AIDS arrives, bring new concerns. NYT, 6 Jan 1993, B3 Mireya Navarro An expanded definition of AIDS went into effect January 1. In this new definition three illnesses and a test of a person's immune function (CD4 count of less than 200) were added as indicators of AIDS. The New York Health Department estimates that this will double the number of new AIDS cases reported to about 15,000. In New York under the current system hospitals, clinics and private doctors report AIDS cases to health departments, which then report them to the CDC. It is claimed that this misses about 17 percent of the cases and the health department would like to develop a method which can include laboratory reporting. The article discusses the problems involved in developing a reporting method that gives an accurate count of those with AIDS and does not violate the right to confidentiality of the patient. <<<========<<

>>>>>==========>> Old men's depression tied to low cholesterol. Washington Times, 9 Jan 1993, A1 Report of a study in Lancet in which studied cholesterol levels in 1,020 men aged 50 to 89 and 1,200 women. For men over 70 the study found that significantly more men with low cholesterol suffered from depression than men with more moderate cholesterol. No significant difference was found for younger men or for the women in the study. Interesting comments about what if anything such a study shows. <<<========<<

>>>>>==========>> U. S. students advance in math, for a change. The New Yoyk Times, 13 Jan 1993, B6 Karen De Witt A preliminary report of the National Assessment of Education Progress suggests a modest improvement in mathematics achievement between 1990 and 1992 although there was less good news about children in urban areas. Albert Shanker argues that reporting these results now was a political ploy and that the government should have waited for the final report to better understand what is really going on. <<<========<<

>>>>>==========>> Healthtrust, Apple put pros ahead of darts. Wall Street Journal 14 Jan 1993, p C1 John R. Dorfman The experts beat the dart throwers in the past six month period of this ongoing contest. The four experts collectively made 14.1% while the dart throwers only made 1.1%. In the series of 31 overlapping six-month contests the pros have won 17 times and the dart throwers 14 times. The average six month gain for the pros is 9.2% and for the dart throwers 5.5%. Doesn't look significant. <<<========<<

>>>>>==========>> Almanac: The Other Hand. The New York Times, 11 Jan 1993, pA14. "When Bill Clinton raises his hand to take the oath of office, he will become only the fifth left-handed president. The others were George Bush, James Garfield, Harry Truman and Gerald Ford." [The left-handed incumbent rule--that no left-hander wins a second term--has now worked for Garfield, Ford and Bush. Any bets on 1996?] <<<========<<

>>>>>==========>> Science Watch: Quake Risk Update. The New York Times, 12 Jan 1993, pC2. New geological data show that over the last 500 years big quakes have occurred on the southern San Andreas fault about every 100 years, not 132 years as previously thought. By studying how gravel and peat layers have been disrupted, major quakes are seen to have occurred in 1857, 1812, 1700, 1610 and 1470. It has now been 135 years since the last major quake, leading some scientists to predict that a quake of disastrous proportions will occur "within the lifetime of most people living in Southern CA. The article references a report by USGS geologist Tom Fumal in the current issue of the journal Science. <<<========<<

>>>>>==========>> Landers Quake's Long Reach is Shaking Up Seismologists. Science, 1 January 1993, p29. The aftermath of the Landers quake (in California's Mojave Desert east of LA) last June apparently triggered aftershocks up to 1250 kilometers away. This contradicts conventional thinking, which held that the effects die out a distance of a couple times the length of the ruptured fault. In this case, effects were recorded as far away as Yellowstone Park. According to Paul Reasenberg of the US Geologic Survey: "If I showed my mother" some of the data "she'd have not trouble seeing the connections." Also discussed is a computer program developed at the Institute of Earthquake Prediction Theory and Math. Geophysics in Moscow, which predicted the Loma Prieta quake. That prediction turns out to have depended heavily on seismic data from hundreds of kilometers away (including some of the same areas sensitive to Landers). The algorithms predictions may point the way to a new physical theory. <<<========<<

>>>>>==========>> A 100-Year Average Recurrence Interval for the San Andreas Fault at Wrightwood, California. by Thomas E. Fumal, et. al. Science 259 (8 January 1993), p199-203. This is the research article cited in the NYTimes article above. Includes maps of regions, diagrams of soil layer analysis, and data for carbon-14 dating of the peat layers from which earthquakes dates were inferred. Mean and standard deviation figures are provided for the radiocarbon age estimates, and graphics show associated probability density functions. <<<========<<

>>>>>==========>> Bob Norman pointed out that there was an interesting series of Snoopy cartoons this week that involved revolved around getting a ten point true false exam all correct by chance. He also observed that in the January 21 New York Times page D10 there is a comparison on how the two stock market indexes had done so far this year with the observation that they had differed (one went down and the other up) six out of the eleven trading days so far. How strange is this? <<<========<<

>>>>>==========>> The current issue of "The American Statistician" February 1993 has two articles that bear on issues that come up in a Chance course. The first is an article by Hahn and Meeker "Assumptions for statistical inference" In this article the authors discuss the difference between enumerative and analytic studies. They call a study enumerative if it deals with sampling from a completely known population (polling for example) and analytic if it is trying to predict a future process (the effect of a drug on future patients). They compare the inference problems for these two types of studies. The second is a commentary by Velleman and Wilkinson "Nominal, Ordinal, Interval, and Ratio Typologies are Misleading". The authors argue that while this typology may be useful in psychology it has been used unwisely in statistics and is apt to be so used more so in the future if statistical packages try to decide for the experimenter which tests to use based upon this typology. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! CHANCE News 2.02 (6 Jan to 21 Jan 1993) !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Please send suggestions to: jlsnell@dartmouth.edu >>>==========>>|<<==========<<<