Portal de Periódicos da CAPES

Letters/Editorial

2008; Oxford University Press; Volume: 5; Issue: 3 Linguagem: Italiano

10.1111/j.1740-9713.2008.00314.x

1740-9713

Homicide, Infanticide, and Child Abuse

SignificanceVolume 5, Issue 3 p. 139-140 Letters/EditorialFree Access Letters/Editorial First published: 28 August 2008 https://doi.org/10.1111/j.1740-9713.2008.00314.xAboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinked InRedditWechat editorial “Pierre glanced at the scene spread before his eyes and was spellbound at the beauty of it. The whole prospect swarmed with troops, smoke-clouds from the guns hung overhead and the slanting rays of the bright sun, rising slightly to the left behind him, filled the clear morning air with rosy golden light and long shadows. Over Borodino a mist had spread, melting, parting, shimmering with light in the brilliant sunshine, magically colouring and outlining everything. The smoke of guns mingled with this mist, and over the whole landscape, through the mist and smoke, sparkled the morning sun, gleaming on the water, on the dew, on the bayonets of the infantry congregated along the river banks and in Borodino. These puffs of smoke and the reports that accompanied them were, strange to say, what gave the chief beauty to the spectacle.” That is the Battle of Borodino, as described by Tolstoy in War and Peace. Minard's great graphic, reproduced on page 134 of this issue, shows the same battle, but simply as a thin red line getting suddenly rather narrower as 27 000 Frenchmen, and around double that number of Russians, died on August 26th, 1812, 75 miles outside Moscow. Does the novel or the graphic best convey the truth about Napoleon's march on Moscow? Tolstoy takes 1400 pages (in the Penguin edition) to draw his epic (the quote is from page 938). Minard, as Helen Joyce tells us, took just a single piece of paper. Tolstoy tells of the battle through the eyes of a single onlooker—Pierre Bezuhov; his individual, subjective, impressions are confused, fragmentary and unrepresentative. No individual is identifiable in Minard's graphic. It is, on the surface, about as objective and passionless as any record could be. Both, of course, are true. (That Tolstoy was writing fiction does not make that statement false: “The one thing necessary, in life as in art, is to tell the truth” he said.) Both accounts are necessary if we want to understand that campaign. Part of the genius of Minard is that his dry graph, his single crooked line snaking across the page and getting narrower and narrower as it does so, nevertheless conveys the horror of the French campaign; part of the genius of Tolstoy is that his subjective, partial, story nevertheless conveys the scale of the events that shook his single Russian family and changed their loves and their lives forever. Tolstoy never tells us how many men died at Borodino; that was not his business. Minard never tells us about the beauty of the mist above the battlefield; that, just as surely, was not his. If we want to understand history, and the roles of human beings within it, both the emotions and the numbers are needed. Michael Blastland, on page 127, says that storytellers deal with the individual, statisticians with the mass. But individuals, taken together, make up the mass; individual stories, taken together, make up the mass movements and world-shaking events of history. Novels can be written about the birth of a child; impossible to write a novel about the birth rate, says Blastland. The great statistical novel has yet to be written. Perhaps it never will be. Still, it is there, as a challenge, for a statistician, or a novelist, to attempt. What exactly would War and Peace have been like if Tolstoy had taken a course in elementary statistics? Julian Champkin Breast cancer risks When it is known that hormonal treatments increase the risks of female breast cancer, our most common cancer, it is surprising that Hannaford (Significance, March 2008) finds that contraceptive pills are beneficial in reducing cancer risks. The modern increase in breast cancer is apparent over age 50. But Hannaford has concentrated on a younger age range. Because of the high dropout, the study is weighted in favour of women-years of exposure at ages too young to investigate the increased risk from these pills. At ages younger than 45 the increase in incidence is small. Can he say what the results would be for cancers diagnosed over age 45? That would be very interesting. There is also potential bias both from left censoring and right censoring. The women, who started in the study with an average age of 29, will have included some who had already used hormonal contraceptives. These include upper-class women among whom there is a higher incidence of breast cancer. Drop out from the study has also led to right censoring: these women, a high proportion of whom are upper class, are lost to observation in respect of the cancers diagnosed over age 50 where the effects of all the reproductive risk factors are manifested. By ignoring the characteristics of the modern British breast cancer epidemic1, its age distribution and its social gradient, Hannaford has underestimated the risks of breast cancer so as to eclipse any aversion from other cancers with these pills. Patrick Carroll Pension and Population Research Institute, London Professor Hannaford replies Dr Carroll is correct in reminding us that, taken together, the global evidence indicates an increased risk of breast cancer being diagnosed among oral contraceptive users1. The increased risk occurs during the period when women are actively using oral contraceptives and for the first few years after stopping—by 10 years after stopping the risk has fallen back to that of never users. Most women, however, stop using oral contraceptives before they reach their late 30s, i.e. before the age when breast cancer is common. The extra number of breast cancer cases attributable to the pill, therefore, is likely to be very small. In longitudinal epidemiological research it is standard practice to calculate incidence rates for disease using only periods of observation during which the participants are known to be at risk of having the disease detected if it occurs. Furthermore, participants known to have the disease of interest prior to study entry should be excluded from the calculations. The Royal College of General Practitioners’ (RCGP) Oral Contraception Study used these standard practices of censoring when calculating its incidence rates. Although women of higher social class were more likely to have been lost to follow up than those in lower social classes, the proportion of such women in the ever and never user groups was similar. Comparisons between groups therefore should be unaffected by this differential loss to follow-up. In the RCGP study breast cancer represented roughly a third of all cancers in the main analyses. I do not believe that we have missed a substantial number of breast cancer cases, or that any errors have affected ever users more than never users. Even if this were the case, it would take a very substantial error to change our results materially. For example, if we make the large assumption that the “real” number of cases of breast cancer in ever users was 50% higher than that observed (i.e. that there were 1337 cases rather than 891), the overall relative risk of any cancer among ever users compared with never users changes from the observed 0.88 to 0.96. My interpretation of the published RCGP study results remains that oral contraception is not associated with an overall increased risk of cancer; indeed it may even produce a net public health gain. Professor Philip Hannaford Aberdeen Designing scatterplots In his article on scatterplots (Significance, March 2008) Allan Reese reprints the graph from Figure 7 of my paper1. He complains that “the graph does not tell the reader which value of s goes with each point”. The graph plots ED on the y-axis against EA on the x-axis. The important (and, to me, shocking) feature of the graph is that one value tends to go down as the other goes up, which is in contrast to the usual behaviour, shown in Figure 3 of the same paper, where they tend to increase together. Was I wrong to assume that the reader would look at Figure 3 (and absorb its message) before Figure 7? In my view, labelling the points of Figure 7 would obscure its most important message. It would also weaken the comparison with Figure 3, where there is no simple way of labelling the points. The reader who really wants to know which value of s goes with which point can refer to the short section immediately before Figure 7 is mentioned in the text. This section has eight sentences, of which the second and sixth show that the D criterion increases as s increases. R. A. Bailey Queen Mary College, University of London Fly–drive tips Many thanks for the article on “Statistics for the birds”, on how starlings keep their place in a flock by watching the birds on either side of them. It occurs to me that there is a parallel with what we are taught about driving. Don't just watch the car in front of you, but keep an eye on several cars in front as well as those either side and behind. Peter G. Nicholls University of Southampton A nice cup of † Stephen Senn concludes his excellent article on the t test (Significance, March 2008) by leaving it to the reader to decide whether Guinness or tea would be most appropriate for celebrating the centenary of said test. At the University of Canberra, Australia, we decided to go for tea, and held a “tea party for the t-test”. Around 30 academics joined in the fun, every single one of whom had used t tests in their research or teaching. They came from every corner of the campus, along with invited statisticians from the Australian National University. Decorations for the venue were supplied by two local Irish pubs, and a large and rich chocolate cake inscribed “Happy 100th Student's t” was consumed. The event turned out to be a great opportunity for researchers and teachers right around the campus to meet and discuss their projects. I hope that this event turns out to be one of many organised to celebrate the centenary of the t-test. Letters should be sent by e-mail to significance@rss.org.uk, or by post to: Significance Letters Page, Royal Statistical Society, 12 Errol Street, London, EC1Y 8LX. They should be short (preferably under 250 words), may be edited for length and should clearly indicate whether or not they are for publication. They must be received by November 1st 2008 in order to be considered for publication in the December issue. Alice Richardson University of Canberra, Australia References 1Carroll, P. (2007) The Breast Cancer Epidemic: Modelling and Forecasts Based on Abortion and Other Risk Factors. Journal of American Physicians and Surgeons, 12, 72– 78 (also available at http://www.jpands.org/vol12no3/carroll.pdf). Google Scholar 1 Collaborative Group on Hormonal Factors in Breast Cancer (1996) Breast cancer and hormonal contraceptives: collaborative reanalysis of individual data on 53,297 women with breast cancer and 100,239 women without breast cancer from 54 epidemiological studies. Lancet, 347, 1713– 1727. CrossrefPubMedWeb of Science®Google Scholar 1Bailey, R. A. (2007) Designs for two-colour microarray experiments. Applied Statistics, 56, 365– 394. Wiley Online LibraryWeb of Science®Google Scholar Volume5, Issue3September 2008Pages 139-140 ReferencesRelatedInformation