CHANCE News 3.02
        (22 January to  12 February 1994)


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:

Current and previous issues of CHANCE
News can be found on our chance gopher.
Just point your gopher to:


"If all the statisticians in the world were laid head 
to toe, they wouldn't be able to reach a conclusion"

 Anon., after comment on economists by G.B.S.



    Two publications that raise interesting statistical questions are the "FairTest Examiner" published quarterly by the National Center for Fair & Open Testing, in Boston (617) 864-4810, and "Extra" published by the FAIR (Fairness & Accuracy in Reporting) with offices in New York (212) 633-6700 The current issue of the "FairTest Examiner" reviews a recent issue of the journal "Longterm View" published by the Massachusetts School of Law entitled "Are Standardized Tests Contributing to Social Stratification?" The articles in this issue are written by a variety of testing experts who raise some interesting statistical problems. For example, one writer poses the following question: ETS wants a last question for its Math SAT exam that about 16 percent will get correct. It has three questions that have been pre-tested to have this property. On one question women did better, on a second, men did better and on a third, men and women did about the same. Which question would ETS pick? In particular, might the fact that men do better on the math SAT bias the testers to choose the question on which men do better because of their desire to have the question correlated with the overall score on the exam? The current issue of "Extra" has an article that critiques the media coverage of the proposals for health care reform. It emphasizes that presenting both sides does not, in itself, prevent bias or tell the whole story, especially when there are more than two sides available. Another article discusses recent comments of news people at ABC such as that of executive producer Emily Rooney who said that she wanted to "tap more into a conservative point of view...rather than the traditional media liberal spin" and a similar comment by Peter Jennings. The article mentions research challenging the concept that the conservative view has been underrepresented in the media and refers to earlier issues where this question has been discussed in more detail. Discussion: Do you think the media show a liberal bias? What about NPR and PBS? Design a test to measure media bias. Is bias in the media a bad thing? <<<========<<

    >>>>>==========>> Fetal Test Focuses the Health-Care Debate The New York Times, 5 Feb. 1994, Section 1, Page 1 Esther B. Fein

    A pregnant women can now have an amniocentesis that will detect chromosomal disorders in her fetus. The test itself has a non-negligible probability of causing a miscarriage. The medical standard is that only women over 35 or with a history of genetic abnormalities should risk a miscarriage by having the test. Age 35 was chosen because this is the first age where the risk of giving birth to a baby with a severe disability, estimated to be 1 in 192, is greater than the probability of miscarriage due to the procedure, estimated to be 1 in 200. Insurance companies uniformly cover the cost of the test for women over this age and for younger women when a physician deems it necessary, but, in the latter case, the younger the patient the more often the insurance companies will challenge a doctor's decision. The article discusses the obvious difficulties, in terms of a women's value judgments, medical costs, etc., of setting such a decision procedure. The Dartmouth Ethics Institute has made a documentary on this issue, and I will review this next time. Discussion: Is this current medical standard reasonable? If not, suggest an alternative decision procedure. Whose business should it be to determine a standard for when an amnio is advisable? The insurance company? A government panel? The individual doctor and patient? How could one try to influence the decision of whether or not to have an amnio without simply saying, "This is what we will pay for?" <<<========<<

    >>>>>==========>> Mouse Study Raises Doubts on Sunscreens The New York Times, 25 Jan., 1994, C3 Gina Kolata

    The current issue of the Journal of the National Cancer Institute reports a study that shows that, while sunscreens protected mice from sunburn, they had no effect on melanoma, the deadliest of all skin cancers. Melanoma experts have known for some time that the relationship between sun exposure and melanoma is much more mysterious than it is for the less deadly skin cancers, the basal and squamous cell carcinomas. Studies have consistently found that the risk of these less dangerous cancers rise in direct proportion to the amount of time people spend in the sun. This has not been shown for melanoma. Epidemiology studies have shown a link between sun exposure and melanoma. People who live closer to the equator have a greater incidence of melanoma, and studies have shown that sun exposure during childhood may increase the risk of melanoma in adulthood. On the other hand, people who spend a lot of time in the sun, such as farmers, do not have a higher rate of melanoma, but do have higher rates of the less dangerous skin cancers. An accompanying editorial remarks that "a variety of studies over the years have yielded abundant circumstantial evidence that ultraviolet light 'causes' melanoma but there is little understanding of the precise mechanism of how this comes about." One indirect theory is that light suppresses the immune system, making it easier for melanomas to grow. Another theory is that light causes inflammation that triggers the cancer's growth. <<<========<<

    >>>>>==========>> Who's the DNA fingerprinting pointing at? New Scientist, 29 Jan. 1994, 51-52 David Pringle

    Pringle describes the successful appeal of a rape case where the primary evidence was DNA fingerprinting. In this case the statistician Peter Donnelly opened a new area of debate. He remarked that forensic evidence answers the question "What is the probability that the defendant's DNA profile matches that of the crime sample, assuming that the defendant is innocent?", while the jury must try to answer the question, "What is the probability that the defendant is innocent, assuming that the DNA profiles of the defendant and the crime sample match?" Apparently, Donnelly suggested to the Lord Chief Justice and his fellow judges that they imagine themselves playing a game of poker with the Archbishop of Canterbury. If the Archbishop were to deal himself a royal flush on the first hand, one might suspect him of cheating. Assuming that he is an honest card player (and shuffled eleven times) the chance of this happening is about 1 in 70,000. But if the judges were asked whether the Archbishop were honest, given that he had just dealt a royal flush, they would be likely to place the chance a bit higher than 1 in 70,000. The error in mixing up these two probabilities is called the "the prosecutor's fallacy", and it is suggested that newspapers regularly make this error. Apparently, Donnelly's testimony convinced the three judges that the case before them involved an example of this and they ordered a retrial. <<<========<<

    >>>>>==========>> Guns and Homicide in the Home New England Journal of Medicine, 3 Feb. 1994, 365-368 Correspondence

    Not too surprisingly, the study in the Oct 7 1993 issue of the "New England Journal of Medicine" suggesting that homicides occurred at a significantly higher rate in homes with guns, has brought a lot of letters. Some criticisms were related to the possible differences in the way that the information was obtained for the cases and the controls. The cases were homicide victims, and so proxies were needed to provide information about them. Proxies were also identified for the controls, but they participated in less than 50% of the interviews since it was obviously a lot easier to get the information from the controls. Presumably, information obtained from a control will be more accurate than that obtained by a proxy for the deceased. In addition, there were significant differences in the cases and the controls: Cases were more likely to live alone, more likely to have a loaded or unlocked gun and more likely to have a person in the home with a record of criminal arrest. Letters from the National Rifle Association complained about the lack of evaluation of the protective value of firearms in the study. The students of Dr. Mark Ferris's Mathematical Statistics 460 class had several objections, including the authors' generalization from a sample of people who have been murdered to a population of people who keep guns in their homes. The accompanying editorial also came in for its share of criticism for the sweeping conclusions that it drew from the study. In their reply, the authors answer some of the criticisms and remark that, while there are difficulties with the case-control studies, this was the research technique that was used to identify the link between cigarette smoking and lung cancer and as was true there, they fully expect their finding will be corroborated by subsequent studies. <<<========<<

    >>>>>==========>> Study Finds no Reason for not Cutting Cholesterol Reuters, 3 Feb. 1994

    This article discusses studies reported in the current British Medical Journal on the effect of lowering cholesterol levels and the prevention or heart attacks. The authors claim that by combining current studies with previous studies they have shown that the strength of the link between cholesterol and heart disease has been underestimated in many studies. They studied 21,000 men aged 35 to 64 who attended health clinics in London. They found a 10 percent reduction in cholesterol level caused a 30 percent reduction in heart disease at age 60, rather than the previously reported 20 percent. They claimed that a long-term reduction in cholesterol of 10 percent can be achieved by moderate dietary change and will lower the risk of heart disease by 50 percent at age 40, and decreasing to 20 percent at age 70. They did not find any evidence that low levels of cholesterol were associated with increased mortality from anything except strokes. They suggested that the previous association between low cholesterol and suicide was probably because depression is associated with suicide and lack of appetite. The latter could lower the cholesterol level. <<<========<<

    >>>>>==========>> Cholesterol Tests "No Help in Fighting Heart Disease" Press Association Newsfile, 1 Feb. 1994 Linda Jackson

    A study reported in Lancet concluded that cholesterol levels are too unreliable to be a satisfactory method of predicting heart problems. This is based on a 12- year study of 21,500 men in the UK. Over the period, 220 men died from heart disease. The 5% with the highest cholesterol measurements experienced only 12% of the deaths. The top 10% experienced only 20% of the deaths. The authors suggest that it is better to concentrate on getting people to quit smoking and improve their diet than it is to carry out a massive screening process. <<<========<<

    >>>>>==========>> Damage by Electromagnetic Fields Studied San Francisco Chronicle, 4 Feb. 1994, A4 David Perlman

    Researchers at Stanford University report in the current issue of Science that they have found that weak electromagnetic fields like those emitted by overhead power lines can disrupt synthetic membranes similar to those surrounding human cells. The leader of the research, Harden M. McConnell, believes that they have shown a possible mechanism for electric fields to disrupt normal cell behavior but is not himself convinced that fears of cancer or other cell damage from power line emissions are justified. However, he is quoted as saying, "If I were to make a bet on whether weak EMF has any effect on human cells, I would bet against it. But the nature of the problem, and the concerns raised over it, call for a hypothesis that must be tested." <<<========<<

    >>>>>==========>> Hidden Rules Often Distort Ideas of Risk The New York Times, 1 Feb. 1994, C1 Daniel Goleman

    Recent studies have suggested to psychologists that hidden psychological rules can help explain why some risks are exaggerated in people's minds while others are played down. That the average person has a different estimate of a risk than an expert has been born out in studies. One such study asked 40 members of the League of Women Voters to rank the risk of 30 activities and technologies. The rankings were compared with the rankings of 15 national experts. Striking differences were found. For example, the League members ranked nuclear power first and the experts ranked it 20th. The experts ranked X-rays as number 7 and the League members ranked it number 22. (Presumably these are some kind of consensus rankings.) The article gives the complete rankings of the two groups and they have obviously picked out two of the more extreme differences. Overall, the rankings don't seem all that different to me. Examples of these hidden rules are: risks that are imposed loom larger than those that are voluntary. We will accept the risk of skiing sooner than that of food preservatives. Risks that are unfairly shared seem more hazardous. Risks that outrage us exaggerate their danger. Natural risks are less threatening than man- made ones. The more a risk is discussed in the news the more we fear it. Of course, the government spends millions of dollars assessing risks, and then the public makes up its own mind as to what the risks are. One expert is quoted as saying, "It's not that there are 'real' versus perceived' risks. These are simply two sets of perceptions, both of which can be argued for as having some validity." Discussion: Do the differences between the League members' ranking and the experts' ranking really indicate different risk assessments, or could they be due to chance? What does it mean to say that `These are simply two sets of perceptions?' Some people claim that your chances of being bitten by a shark are far smaller than your chances of being hit by falling airplane parts. What does this mean? Should surfers carry shark repellent, or falling airplane parts repellent, or both? Does being hit by a whole airplane count as being hit by falling airplane parts? How much do the psychological factors identified in this study contribute to your assessment of the risk of being gunned down in a fast-food restaurant? Of being struck by lightning? Of being bitten by a shark? Of being hit by falling airplane parts? <<<========<<

    >>>>>==========>> Investment Professionals Scored a Decisive Victory but a Random Throw of Darts didn't do too Badly, Either. Wall Street Journal, 3 Feb. 1994, A1 Georgette Jasen

    I finally figured out that this contest has a new pair of groups starting every month, and they compete over a six month period. This is a report of the competition that started on Aug. 16 and ran through Jan. 31. The pros had an average gain of 27.8%, the darts a gain of 18.5%, and the Dow Jones a gain of 11.2% during this period. <<<========<<

    >>>>>==========>> Milk Helps to Combat Bone Loss From Coffee. The New York Times, 26 Jan. 1994, C12

    A study reported in the current issue of The Journal of the American Medical Association shows that women who drink as little as two cups a day of coffee containing caffeine experience a significant decline in bone density as they get older. This can lead to increased risk of fractures, and previous studies have linked coffee drinking to an increased risk of hip fractures. This study suggested that women who drink at least one glass of milk each day throughout their adult lives can largely counter the bone-thinning effects of a lifetime of coffee drinking. But they do suggest that it is not sufficient to start this during middle age. The study was done by asking 980 post-menopausal women to recall their coffee-and-milk-drinking habits throughout their adult lives. <<<========<<

    >>>>>==========>> Linemen Risk Highest Price for Bulking up to Play Los Angeles Times, 28 Jan. 1994, C6 Bill Plaschke

    A study by the National Institute for Occupational Safety and Health showed that NFL football players have an estimated average lifetime of at least 72 years, which is the national average. Previous claims have been made that their average lifetime was 53. However, the study showed that linemen have a 52% higher chance of dying of heart disease than the average person and face three times the risk of getting heart disease. As the article points out, no study has compared them to the 300-pound population. The study used data covering 3,420 players who played at least five pro seasons from 1972 to 1988. Discussion: You are introduced to two 300-pound 23-year-old males. One is an NFL lineman and the other is a couch potato. Which do you think is likely to weigh more at 43? Do you think linemen would be more or less prone to heart disease than randomly selected men matched to have the same weight as the linemen at age 23? At age 43? How could having been an NFL lineman increase the danger of dying of heart disease? <<<========<<

    >>>>>==========>> 'Sham' surgery of clinical trials divides scientists. Boston Globe, 4 Feb. 1994, National/foreign pg. 1 Judy Foreman

    Scientists and ethicists are divided over a clinical trial in Colorado to determine if fetal tissue injected into the brain can benefit patients with Parkinson's disease as suggested by previous research. The trial involves 40 patients with Parkinson's disease. Twenty will have a "sham" surgical procedure that involves cutting two holes in the skull, with no benefit from the surgery, while the other 20 will get real surgery that will allow fetal tissue to be injected into the brain. George Annas, director of the law, medicine and ethics program at the Boston University School of Medicine, stated that, "Not only is it ethical to do it this way, it's probably unethical to do it any other way." Other ethics experts feel that this is going too far. Some scientists are afraid that, if this well-publicized experiment is not successful, the government will back off of their support for research in this area and patients will lose faith in this kind of treatment. Others wondered who in the world would sign up for such an experiment. A spokesman for the study said that simply drilling two postage size holes is not very dangerous. The potential danger comes with the insertion of the needles into the brain which will not be done with the control patients. Great care will be taken to make sure that the patients will themselves not know if they are one of the controls. The controls will be offered the real thing a year later. <<<========<<

    >>>>>==========>> When the Dying Enroll in Studies: A Debate Over False Hopes. The New York Times, 29 Jan. 1994, Section 1 Page 7 Gina Kolata

    Evidently, a prime source of subjects for phase 1 clinical trials are terminally ill patients. One expert comments that the purpose of phase 1 studies is to determine the drug's toxicity and the best way to administer the drug. It does not focus on the beneficial effects, although consent forms do tell patients that they could benefit. It is remarked that terminally ill patients feel that they have nothing to lose and tend to be willing to try anything without consideration of possible harmful effects. In addition, it is pointed out that they are often not treated as patients but almost as white rats when they are bumped from the study because it is felt that they can no longer provide valuable data. There is a lot of disagreement among both medical researchers and ethicists about just what terminally ill patients should be told about participating in a phase 1 study. <<<========<<

    >>>>>==========>> Sundays and Saturdays are colder. The Daily Telegraph, 27 Jan. 1994, Pg 5 Adrian Berry

    Adrian Gordon writes, in a letter in the current issue of Nature, that he wanted to see if we are really getting a warmer world. So he studied satellite readings of global temperatures in the lower atmosphere on a daily basis for 726 weeks, from January 1979 to November 1992. He averaged two sets of figures, one for the northern hemisphere and another for the southern hemisphere, and graphed the results. His graph shows a steady rise from the coldest day Sunday to the warmest day Wednesday, then a sharp dip on Thursday decreasing to the second coldest day Saturday. Presumably, this is caused by industrial activity, which is less on the weekends. Gordon remarked that "the smallest of the average changes in temperature, four hundredths of a degree Fahrenheit between Wednesday and Sundays, shows how tiny is man's effect on the atmosphere." Two other scientists confirm that the average global warming over the same 14 years has been about .2 degree Fahrenheit which is a third of what theoretical models had predicted. Discussion: Could the reported difference in temperature between Wednesdays and Sundays be attributable entirely to chance? Does the steadiness of the rise and fall from Sunday to Wednesday back to Sunday indicate that there is more to the differences than chance fluctuations? Was 1900 a leap-year? Will the year 2000 be a leap-year? Is the 13th of the month equally likely to fall on any of the seven days of the week? Is the first of February equally likely to fall on any day of the week? In the absence of all industrial activity, which day of the week would you expect to have been the coldest over the long haul? Which the warmest? Which the luckiest? <<<========<<

    >>>>>==========>> A sweeter take on sugar. Boston Globe, 3 Feb. 1994, National/Foreign Pg 1 Judy Forman

    It has long been folk knowledge that sugar makes children hyperactive, restless, unable to learn new things, etc. This has not been born up in studies, and a new study, published in the current New England Journal of Medicine, was carefully designed to give sugar as good a chance as possible to show its harmful effects. Researchers studied two groups of students: 25 pre- schoolers with no psychiatric disorders, and 23 school- age children described by their parents as being sensitive to sugar. Three diets were created and given in three-week periods in a Latin square design. Within each three week period, the families had one of the three levels of sweetening: (a) high in sugar (sucrose), (b) low in sugar but high in aspartame (NutraSweet), or (c) low in sugar and high in saccharin, used as a placebo. Subjects were told that their diet would change every week, and indeed the changes in their diets were made each week, while the type of sweetening was kept constant through the three-week period. Families were asked to give their estimates of what diets they had (I assume as regards sweetening) each week. It is said that only one parent got them all correct. Urine samples were also checked to check compliance. For the supposedly sugar-sensitive children, there were no significant differences in any of the 39 behavior and cognitive variables measured. For the pre-schoolers, only four of the 31 variables measured were different on the various diets, and the differences associated with the ingestion of sucrose were more consistent with a slight calming effect rather than with hyperactivity. Experts do not expect this popular theory to die soon, pointing out that people look for a tangible cause outside themselves. <<<========<<

    >>>>>==========>> AIDS' other war; Science "bible" battles newspaper campaign to debunk HIV connection. The Ottawa Citizen, Weekend Observer, 12 Feb. 1994, Pg B3 Shelley Page

    For two years the British Sunday Times has campaigned to convince its readers that HIV may be irrelevant to AIDS, heterosexuals aren't at risk for the disease, and that AIDS scarcely exists in Africa. The newspaper has been right before, when it fought to tell the truth about thalidomide, but wrong before, when it tried to prove that the diaries of Hitler that it bought were not fake. The British journal Nature has declared a war of words with the Sunday Times, running angry editorials and critiques of the stories that the Times publishes. This article does an excellent job of reviewing the Times-Duesburg arguments and in discussing related issues on science reporting. <<<========<<

    >>>>>==========>> Coping with Computers Science Watch, 9 Feb. 1994 Pearce Wright

    The bulletin of the British Psychological Society reports a study suggesting that more than one person in four suffers from "computerphobia". They express concern that, as computers proliferate through the workplace, the job security of computerphobes will be threatened. They found some curious things in their studies such as: after six months, women found that they were more successful and enthusiastic about working with computers than men, but, after a year's experience, men liked working with computers more than women did. <<<========<<

    >>>>>==========>> How do you beat the SAT. Boston Globe, 3 Feb. 1994, Living Pg. 45 Patti Doten

    In 1987, three high school seniors wrote a humorous guide to improving your SAT scores called "Up Your Score". They wanted to find someone to update it. They thought that a Michael Colton might be their man, but required that he retake the test and try to improve his 1490 score to at least 1500, I suppose for advertising purposes. He did and, in fact, got one of the dozen or so 1600's last spring. This article described some of his improvements and the kind of tips that are given in the book. For example, Colton remarks that the SAT has a positive attitude towards artists and so will never have "Beethoven sucks" as an answer. Having found that students often spend as much as 2.3 seconds per circle leading to about 7 minutes and 4 seconds of their time filling in circles, the book includes a chapter on how to fill in circles. It is claimed that students who master this chapter will gain about 6 minutes to spend working on the test. <<<========<<

    >>>>>==========>> !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! CHANCE News (22 January to 12 February 1994) !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Please send suggestions to: jlsnell@dartmouth.edu >>>==========>>|<<==========<<< >>>==========>>|<<==========<<<