Statistics Articles

8/7/2019 Statistics Articles.

1/14

SIR RICHARD DOLL1912: Born in Hampton, England, on 28 October1937: Graduated from St Thomas's Hospital Medical School in London1939-45: Served in the Royal Army Medical Corps1946: Started work at the Medical Research Council

1951: Co-authored a paper suggesting smoking causes lung cancer1954: Co-authored a paper confirming the link between smoking and lungcancer1956: Awarded an OBE1961: Appointed director of the MRCStatistical Research Unit1969: Appointed Regius Professor of Medicine at Oxford University1970-71: Served as vice-president of the Royal Society1971: Received a knighthood1996: Made a Companion of Honour for services of national importanceJuly 2005: Dies after a short illness aged 92

A 50-year study has provided the most comprehensive picture yet of the perils of

smoking.

For half a century eminent scientist Sir Richard Doll has followed smokers to assessjust what impact their habit is having on their health.

BBC News Online profiles the man who first confirmed the link between smoking andlung cancer.

Fifty years ago, doctors at the UK's Medical Research Council published a scientific paper that was trulyground-breaking. They revealed that smoking can cause lung cancer. It was the first time the link hadbeen confirmed. The findings were to change the minds and lives of millions of people around the world.In 1954, 80% of British adults smoked. Today, that figure is 26%. Sir Richard Doll was one of the menbehind that pioneering study. He was 41 at the time and had been working in the MRC's StatisticalResearch Unit since the end of World War II. The study was the culmination of years of work, all aimedat trying to find out why so many people were dying from lung cancer. "Mortality from lung cancer was

increasing every year in the first few decades of the last century," said Sir Richard. "People didn't payany attention to these mortality rates during the war. "But in the years that followed, they started tobecome concerned." Today, few people dispute that smoking causes cancer. In post-war Britain it was avery different story. Some scientists had suggested that rising rates of lung cancer may be due tosmoking. But tests on animals appeared to rule out a link. Many researchers, including Sir Richard,started to investigate other potential suspects. "I personally thought it was tarring of the roads. Weknew that there were carcinogens in tar." Sir Richard and his colleagues interviewed 700 lung cancerpatients to try to identify a possible link. "We asked them every question we could think of," he said. "Itwasn't long before it became clear that cigarette smoking may be to blame. I gave up smoking two-thirds of the way through that study." The findings were published in 1951. However, it wasn't until the1954 paper was published that people started to take notice. "Nobody believed us," said Sir Richard."They thought there may be other explanations."

Historic study The MRC researchers continued with their work. This time they enrolled every doctor inthe UK in their study. In 1951, they asked 40,000 doctors if they smoked. Over the course of the nextthree years, they compared those answers with information about doctors who went on to develop lungcancer. They found a direct link. The findings prompted the then UK health minister Iain Macleod to call anews conference. Chain-smoking throughout, he said: "It must be regarded as established that there is arelationship between smoking and cancer of the lung." The study has provided the foundation for allother research into the impact of smoking cigarettes on health. It has arguably helped to save millions oflives.

Sir Richard's work has been recognised throughout the world. He has received honorary degrees from 13universities. He has won countless awards, including the United Nations Award for Cancer Research in


2/14

1962 and the gold medal of the European Cancer Society in 2000. His achievements have beenrecognised by the Queen. He was knighted in 1971 and made a Companion of Honour in 1996 forservices of national importance.

But among his peers, Sir Richard is known for much more than just his 1954 paper. Over the course ofthe past five decades, he has published hundreds of papers on topics as varied as oral contraception,peptic ulcers and electrical power lines. He has shown that all radiation is potentially harmful, whichwasn't always thought to be the case, and that aspirin can protect against heart disease. He hasuncovered evidence to suggest that drinking alcohol increases the risk of breast cancer and that

electrical power lines do not cause cancer.

Hitting the headlines His findings have sometimes sparked controversy. So too has the man. In 2001,he riled the anti-smoking lobby after appearing to downplay the risks from second-hand smoke. In aninterview on BBC Radio 4's Desert Island Discs, he said: "The effects of other people smoking in mypresence is so small it doesn't worry me." In February 2004, he hit the headlines after saying he wouldbe willing to go to prison because of new rules on medical research. At 91, Sir Richard remained as busyand as sharp as ever. In March that year, he took part in the topping-out ceremony for the new RichardDoll Building at Oxford University. Fittingly, the building will house some of the country's top cancerresearchers. In June 2004, he published further findings from the study he started in 1951. Some 67years after graduating from medical school, he was only then for the first time considering retirement.

http://www.economist.com/science/displayStory.cfm?story_id=2724226&no_na_tran=1#top
http://www.economist.com/science/displayStory.cfm?story_id=2724226&no_na_tran=1#tophttp://www.economist.com/science/displayStory.cfm?story_id=2724226&no_na_tran=1#tophttp://www.economist.com/science/displayStory.cfm?story_id=2724226&no_na_tran=1#tophttp://www.economist.com/science/displayStory.cfm?story_id=2724226&no_na_tran=1#top


3/14

Jun 3rd 2004 From The Economistprint edition

Far too many scientists have only a shaky grasp of the statistical techniques they are using. What is published in scientific journals may not be as true as it should be

SCIENTIFIC and medical journals, with their august panels of peer reviewers and fact checkers, are not

the sort of places many mistakes are to be expected. Yet Emili Garca-Berthou and Carles Alcaraz, tworesearchers at the University of Girona in Spain, have found that 38% of a sample of papers in Nature,and a quarter of those sampled in the British Medical Journal(BMJ)two of the world's most respectedjournalscontained one or more statistical errors. Not all of these errors led to erroneous conclusions,but the authors of the study, which has just been published in BMC Medical Research Methodology,another journal, reckon that 4% of the errors may have caused non-significant findings to bemisrepresented as being significant.

Dr Garca-Berthou and Dr Alcaraz investigated 32 papers from editions ofNature published in 2001, and12 from the BMJin the same year. They examined the numbers within each, to see whether the datapresented actually led to the statistical conclusion the authors drew, and also whether there wasanything fishy about the numbers themselves. Appropriately, they used a statistical technique to do theirchecking. If a set of data are unedited, the last digits in the numbers recorded will tend to have the

values 0-9 at random, since these digits represent small values, and are thus the ones that are hardestto measure. If those numbers are rounded carelessly, however, 4s and 9s (which tend to get rounded upto the nearest half or whole number) will be rarer than they should be. The two researchers dulydiscovered that 4s and 9s were, indeed, rarer than chance would predict in many of the papers underscrutiny.

False data, false results. Though it was difficult to show whether, in any given case, this falsity led to aresult being proclaimed statistically significant when it was not, it was possible to estimate how mucherror there was likely to be. In one case, however, there was no doubt. A number supposed to bestatistically significant was explicitly mis-stated, and a false inference drawn in the paper's conclusion.

Of course, mistakes will creep through from time to time in the best-run organisations, and there is no

suggestion that any of the errors observed was a deliberate fraud. But there do seem to have beenrather a lot of them. However, as Kamran Abbasi, deputy editor of the BMJ, laments, although the worldat large looks at scientific peer-reviewthe system journals use to keep their authors accurate andhonestas a sacred process, it is in fact imperfect. We certainly do not spend our time recalculating allthese numbers, and our whole review process would likely grind to a halt if we tried to do so.

Maxine Clarke, publishing executive editor ofNature, says her journal will be examining the papers citedby Dr Garca-Berthou and Dr Alcaraz before deciding what action, if any, needs to be taken. At firstsight, some awareness-raising about statistical accuracy among manuscript editors, peer-reviewers andproof-readers seems necessary, but we have changed our workflows considerably since the periodstudied, says Ms Clarke.

One cure might be for researchers to publish raw data as well as statistical analysis and conclusions.That way, anyone who really cares can check the sums. For some years, Nature has offeredsupplementary information online to accompany its papers. This information is peer-reviewed, but MsClarke believes it is too specialised for people outside the field to find interesting. We do not explicitlyask authors, as routine, for the raw data underlying their reported statistical results, she says. Thissuggestion is now on the agenda for our next editorial meeting on editorial practices and criteria.

The real answer, however, surely lies with the researchers themselves. Far too many scientists have onlya shaky grasp of the statistical techniques they are using. They employ them as an amateur chefemploys a cook book, believing the recipes will work without understanding why. A more cordon bleuattitude to the maths involved might lead to fewer statistical souffls failing to rise.


4/14

Autism rises despite MMR ban in JapanNew Scientist 05/03/05http://www.autisticsociety.org/autism-article817.html

By Andy CoghlanParents need have no more fears about the triple vaccine against measles, mumps and rubella. A study

of more than 30,000 children in Japan should put the final nail in the coffin of the claim that the MMRvaccine is responsible for the apparent rise in autism in recent years.

The study shows that in the city of Yokohama the number of children with autism continued to rise afterthe MMR vaccine was replaced with single vaccines. "The findings... are resoundingly negative," saysHideo Honda of the Yokohama Rehabilitation Center.In the UK, parents panicked and vaccination rates plummeted after gastroenterologist Andrew Wakefieldclaimed in a 1998 study that MMR might trigger autism, although the study was based on just 12children and later retracted by most of its co-authors. Soon the vaccine was being blamed for theapparent rise in autism, with Wakefield citing data from California (see Graph). In some parts of the UK,the proportion of children receiving both doses of the MMR vaccine has dropped to 60 per cent. This hasled to a rise in measles outbreaks and fears of an epidemic.

Not one epidemiological study has revealed a link between the vaccine and autism. But until now theyhave all concentrated on what happened after MMR vaccination for children was introduced. Honda's isthe first to look at the autism rate after the MMR vaccine has been withdrawn. Japan withdrew it in April1993 following reports that the anti-mumps component was causing meningitis (it plans to introduceanother version).

With his colleagues Yasuo Shimizu and Michael Rutter of the Institute of Psychiatry in London, Hondalooked at the records of 31,426 children born in one district of Yokohama between 1988 and 1996. Theteam counted children diagnosed as autistic by the age of 7. They found the cases continued to multiplyafter the vaccine withdrawal, ranging from 48 to 86 cases per 10,000 children before withdrawal to 97 to161 per 10,000 afterwards. The same pattern was seen with a particular form of autism in which childrenappear to develop normally and then suddenly regress - the form linked to MMR by Wakefield.

The study cannot rule out the possibility that MMR triggers autism in a tiny number of children, as someclaim, but it does show there is no large-scale effect. The vaccine "cannot have caused autism in themany children with autism spectrum disorders in Japan who were born and grew up in the era whenMMR was not available", Honda concludes. His team's findings appear in the Journal of Child Psychologyand Psychiatry (DOI: 10.1111.j.1469-7610.2005.01425.x).

So if the vaccine is not responsible for the rising rates of autism, what is? "Clearly some environmentalfactors are causing the increases," says Irva Hertz-Picciotto of the University of California at Davis. Otherexperts disagree, saying the apparent rise could be the result of changing diagnostic criteria and therising profile of the disorder (New Scientist, 17 February 2001, p 17).

MMR and the incidence of autism recorded by GPs:

a time trend analysisBritish Medical JournalPaper by James A Kaye, epidemiologist, Maria del Mar Melero-Montes,epidemiologist, Hershel Jick, associate professor of medicine.BMJ ( 24 February 2001)

IntroductionThe possibility that the mumps, measles, and rubella (MMR) vaccine may be causally related to the riskof autism is currently causing substantial concern. This proposition originated primarily from apublication by Wakefield et al in 1998 that described 12 case reports of children who were diagnosedwith ileal-lymphoid-nodular hyperplasia followed by behaviour disorders that were clinically diagnosed asrepresenting autism. In eight of 12 children the behaviour disorder was "linked" in time with MMRvaccination by the parents or the child's physician.


5/14

In June 1999 Taylor et al published in the Lancet the results of a study in which they identified childrendiagnosed as having autism in the North East Thames region for birth cohorts from 1979 to 1992. Theyreported that the incidence of autism started to increase in children born in the late 1980s and increaseddramatically in those born from 1989 to 1992. They also provided estimates of the coverage(prevalence) of MMR vaccination from 1987 to 1995, which rose to over 90% by 1988-9.

They found no temporal association between MMR vaccination and the incidence of autism within one totwo years of vaccination, and there was no "clustering" of cases in the two to four months after

vaccination.

In a subsequent letter to the Lancet's editor Wakefield described the study by Taylor et al as containing a"fundamental flaw" and cited data from the United Kingdom (north west London) and the United States(California) based on the time trend of autism occurrence by birth cohort in relation to the introduction ofthe MMR vaccine.3 In both areas a dramatic increase in the incidence of autism was reported in temporalassociation with the rapid introduction of the vaccine.

Objective: To estimate changes in the risk of autismand assess the relation of autism to the mumps,measles, and rubella(MMR) vaccine.Design: Time trend analysis of data from the UK generalpractice research database(GPRD).Setting: General practices in the United Kingdom.Subjects: Children aged 12 years or younger diagnosedwith autism 1988-99, with further analysis ofboys aged 2 to 5years born 1988-93.Main outcome measures: Annual and age specific incidence for firstrecorded diagnoses of autism (thatis, when the diagnosis of autismwas first recorded) in the children aged 12 years or younger;annual,birth cohort specific risk of autism diagnosed in the2 to 5 year old boys; coverage (prevalence) ofMMRvaccinationin the same birthcohorts.Results: The incidence of newly diagnosed autism increasedsevenfold, from 0.3 per 10 000 personyears in 1988 to 2.1 per10 000 person years in 1999. The peak incidence was among 3 and4 year olds,and 83% (254/305) of cases were boys. In an annualbirth cohort analysis of 114 boys born in 1988-93, the risk ofautism in 2 to 5 year old boys increased nearly fourfold overtime, from 8 (95% confidenceinterval 4 to 14) per 10 000 forboys born in 1988 to 29 (20 to 43) per 10 000 for boys born in1993. Forthe same annual birth cohorts the prevalence ofMMRvaccination was over 95%.Conclusions: Because the incidence of autism among 2 to5 year olds increased markedly among boys

born in each year separately

from 1988 to 1993 while MMRvaccine coverage was over 95% forsuccessive annual birth cohorts, the data provide evidence thatno correlation exists between theprevalence ofMMRvaccinationand the rapid increase in the risk of autism over time. The explanationforthe marked increase in risk of the diagnosis of autism inthe past decade remainsuncertain.

http://www.autisticsociety.org/modules.php?name=News&file=article&sid=125&mode=&order=0&thold=0


6/14

helpingto quantify evidence in criminal cases

The importance of forensic statistics

Forensic statistics is the application of statistics to forensic science and the law. Broadly speaking,forensic science is the analysis of traces of evidence (such as body fluids, glass fragments, footprints anddrugs) left at the scene of a crime by the criminal, victim or others. This evidence may be used

subsequently to either implicate or exonerate a person suspected of committing that crime, or just togain further insight into the incident. Over the years, with increasing technological advancement,forensic science has become a key part of criminal investigations worldwide.

But forensic science doesn't just involve identifying traces of evidence - sometimes it isn't obvious justwhat a piece of evidence really is. Other important questions that need to be answered are just how theevidence came to be at the crime scene, where did it originally come from, and who left it there. Thissuggests a natural role for statistics, as these questions can typically only be answered in terms ofprobabilities. So it is not surprising that the primary task of forensic statisticians is to evaluate anyevidence found at a crime scene, so that this evidence can be appropriately presented to a jury incourt. This task obviously carries great responsibility.The advent of DNA profiling in the 1980s brought a big change in the way the legal system viewed

quantitative data. Now a quantitative approach is being requested in many areas, far removed from theoriginal area of DNA profiling. The earlier research and development work is being applied and furtherwork is being done to tackle the increasingly more complex cases which arise in bringing a soundstatistical approach to the assessment of evidence.

What does this career entail? Please note first that forensic statisticians can operate under variousguises. At one end of the scale, there are people employed by forensic science units specifically toanalyse forensic data; at the other end, there are some university lecturers who specialise in carryingout statistical research on forensic matters and act as consultant forensic statisticians whenrequired. The methods of statistical analysis used will usually be similar, no matter where on this scale aforensic statistician is operating.

For an appropriate evaluation of evidence, a comparison of probabilities of the evidence under twodifferent propositions is required. These propositions are usually those put forward by the prosecutionand the defence. There are advanced statistical methods for doing this (for readers who are technicallyinclined, they are based on likelihood ratios or Bayes' factors). Much theoretical work has been done inthe development of these methods. Calculations based on them might sometimes be fairlystraightforward, though it also often turns out that there are non-standard issues to consider.One example of casework that a forensic statistician may be involved with is DNA profiling, which is apowerful method of identification using genetics. Often, the evidence to be evaluated involves human (orsometimes animal) biological material such as blood, semen or vaginal fluid. Considerable work has beendone in statistical and population genetics in assessing the importance of such evidence. Applications,however, are often not restricted to simple cases with one sample of DNA left at the scene of a crimeand one suspect. Complications very often arise, for example because relatives may be involved, or the

suspect may have been identified by a search through a DNA profile database, or the sample found atthe crime scene may be a mixture of body fluids from more than one person. More advanced statisticalmethods are required in such situations.Another role of a forensic statistician relates to sampling problems and determination of sample size. Insome cases, it is necessary to examine a consignment of similar-looking items, and it is often notpractical to examine every item. This may be purely on financial grounds but may be on health groundsalso. The question then arises as to how many items should be examined on a sampling basis. Forexample, the consignment to be examined may be a set of CDs, some of which are thought to containpornographic material. Then it is desirable for the examining officers to examine as few CDs as iscommensurate with a good description of the proportion of the CDs which are illicit. The sample size


7/14

determination is really just a quality control problem; there are UN Guidelines where the problemconcerns drugs.Finally, an important part of being a forensic statistician, as indeed it is for any statistician, is the abilityto communicate results effectively to non-statisticians. Forensic statisticians are often required to attendcourt cases as "expert witnesses". This involves reporting calculated probabilities, or other statisticalmeasures, to the jury, and explaining to them how the calculations were performed. This is a challengein itself, as the jury will typically consist of people who have little knowledge of statistical methods, andis further complicated by the need to choose careful wording (so as not to "lead" the jury into a decision

on guilt or innocence of a defendant).

For more information visit: http://www.rss.org.uk/main.asp?page=2241


8/14

finding out what people want

The importance of market research Market research is, in very general terms, a means for providersof goods and services to keep themselves in touch with the needs and wants of those who buy and use

the goods and services.

It involves the systematic gathering, recording and analysing of information relating to the transfer andsale of goods and services from producer to consumer, together with systematic problem analysis, modelbuilding and fact finding for the purposes of improved decision making and control in marketing goodsand services. A huge range of companies and organisations carry out market research, answeringquestions (among many others) like:

Why have the sales of my breakfast cereal decreased over the last few months? If I launch this new pasta sauce, will anyone buy it? If we build our new swimming pool here, will people be able to get to it easily enough, and will

they actually use it?

How much do people understand about our charity and how can we help them to understandmore? I've got to change one of the ingredients in a drink I make, will my customers notice, and if so

will it affect whether they buy the product?A statistician working in market research can expect to use a multitude of different statistical techniquesin order to solve the numerous challenges that are given by clients and researchers.

What does this career entail? The proportion of time a market research statistician spends actuallydoing statistics depends on the company and the type of work it does, but may be anything up to about80%. For this reason, the career is obviously very satisfying for people who want to continue really usingstatistics in a commercial environment.

As a market research statistician, you will be heavily involved with the research staff who run the

individual projects. You will spend a lot of time working in effect as a consultant for these researchers.You will be involved in writing proposals describing how the market research will be carried out. Theseproposals will cover a number of areas of which the most important from the statistical point of view willbe the overall research methodology and the calculation of sample sizes and related power for relevanttests. You will have to advise about the design of the investigations; for example, there might becomplex rotation plans required if products are being tested or a number of different ideas are beingconsidered in the same piece of research. Once the data are collected and carefully checked, thestatistical analysis itself can begin. The analysis may involve anything from the most simple tests tocomplex multivariate analyses or modelling. Part of the challenge for the statistician is firstly to explainthe analysis and results to the researcher, who may well have no mathematical/statistical background,and then to work with the researcher to present the results in a way that the company itself willunderstand.

Who employs market research statisticians? Some companies and organisations, particularly thelarger ones, have their own market research departments, and there are openings for statisticians inthese. Others use specialist market research companies and agencies to undertake the research forthem; these are major employers of market research statisticians.For more information visit:


9/14

http://www.rss.org.uk/main.asp?page=2244

making sense of biological variation

The importance of biometry Biology is forecast to be the science of the 21st century. The intrinsicvariability of biological systems means that statisticians have the opportunity to play a leading role ininterpreting the massive quantities of molecular data now being generated in the search for the secrets

of life. The history of biometry - the study of measurement in biology goes back over 100 years.Biometrika, now a leading international statistics research journal, was first published in 1901, whilst theInternational Biometric Society, founded in 1947 with R A Fisher as first president, is a majorinternational society bringing together over 6000 members with broad interest in the application ofstatistical and modelling techniques to the biological sciences.

Statisticians working in the biological sciences may be tackling problems in: Genetics - for example, analysing the heritability of a rare disease through genetic pedigrees, or

the patterns of gene expression using microarrays Agriculture - for example, testing the performance of new crop varieties, or the efficacy of

precision application for reducing pesticide inputs

Epidemiology - for example, modelling the spread of citrus disease in an orchard, or estimatingthe prevalence ofE. coliO157 in cattle

Toxicology - for example, developing methods for reducing the number of animals in dose-response studies

Ecology - for example, modelling population change of sea mammals.

Applications overlap with those for a medical statistician (for whom the term biostatistician is often used)and environmental statistician. As in other fields, one of the attractions of working as an appliedstatistician is the opportunity it gives to apply generic methods to a wide variety of different applications.

What does this career entail? You are likely to be supporting a scientific research programme in aresearch organisation, hospital or university department. This may involve contributing to a singleresearch project over several months or years, or being a member of a team of biometricians supportinga range of projects. In the latter case, much of your work may involve short-term consultancy, giving

advice to scientists on the design, analysis, interpretation or presentation of studies. Other work may bemore long-term and lead to joint publication of research findings in the form of academic papers ortechnical reports. Development of statistical software or expert systems is a growing requirement forensuring knowledge transfer to research users. Biological problems may require the development ofinnovative statistical methodology which is suitable for publication in statistics journals. There will beopportunities for attending national and international conferences to present your work and to learn fromthe work of fellow biometricians.

As you grow in experience, you are likely to play a greater role in setting objectives and designingprojects. You may be part of a local committee assessing the feasibility of new studies. You willcontribute to research grant proposals and may help coordinate multidisciplinary projects. Your expertisein a particular area of methodology can bring you into contact with many different fields of application,and open up new opportunities in your career.

For more information visit: http://www.rss.org.uk/main.asp?page=2239


10/14

But company officials question report's accuracyNew York Times

Detroit - A federal study of how vehicles interact in crashes has found that Ford Explorer sport utilityvehicles seem to be especially deadly to the occupants of cars they hit, even compared with othermidsize sport utility vehicles.

The statistical study calculated that four-door Explorers with four-wheel drive killed 10 car drivers forevery 1,000 crashes between Explorers and cars that were reported to police from 1991 through 1997.By contrast, the study said, several other midsize sport utility vehicles, killed 5 to 7 car drivers for every1,000 crashes with cars.

The rate of death for car drivers in collisions with other cars was 0.6% per 1,000 crashes, according tothe study, which was issued by the National Highway Traffic Safety Administration. The analysis was

done by Hans Joksch, a University of Michigan researcher and a leading traffic safety statistician, under acontract from the safety agency, which is part of the Department of Transportation.

Limited numbers of crashes in the database for each model created a fairly wide range of error in thecalculations, though. For the Explorer, for example, there was a 95% chance that the true death rate ofcar drivers was between 7 and 13 per 1,000 crashes. The error range meant that it was statisticallypossible, although unlikely, that one or more of the other midsize sport utilities was deadlier than theExplorer. The study also cautioned that the error ranges themselves may be imprecise.

The Ford Motor Co. dismissed the study as meaningless because of the wide ranges of error. "Errorranges are so large as to be inconclusive," said Ken Zino, a Ford spokesman. The study also did notreview any full-size sport utility vehicles, which might inflict more damage, Zino noted.

Appeared in the Milwaukee Journal Sentinel on Feb. 25, 2001.

For more information visit: http://www.jsonline.com/wheels/peak/feb01/study25022201.asp?format=print


11/14

DID YOU KNOW?Statistics is applied in many fields. Here are a few short stories illustrating some of the numerousapplications of statistics:

Sunlight and skin cancer

The first convincing evidence of a link between exposure to sunlight and skin cancer wasmade by an Australian statistician, Oliver Lancaster. He observed that the rate of skincancer in Australia among Caucasians was strongly correlated with latitude, and hencewith amount of sunlight; the northern states had higher rates than the southern ones(this was well before the hole in the ozone layer!).This observation was only possible by careful collection of data, and comparison of skin

cancer rates.Random surgical operations

Right now, an experiment is being carried out at a Melbourne hospital. Patientswho need a particular operation are wheeled into the operating theatre. Justbefore the surgeon makes the first cut, an envelope is opened which reveals tothe surgical team whether they put valve type A or valve type B into the patient.Whether the patient gets A or B is entirely random; the choice has been made

by a random number generator on acomputer (but it's just like tossing a coin). Why is it done this way? Random allocation of patients totreatments usually sounds very strange to people who have never heard of it. The reason it is doneis to get a fair comparison between the two treatments; in the example, valve A and valve B. Ifsicker patients tend to get allocated to valve A (for example), then comparing the outcome of thetwo groups after the operation would not tell which valve was better. We usually know some of thereasons for better or worse outcomes, and we could try to ensure that the two groups were similaras far as those things go; these might be age, severity of condition, etc. But by randomizing, wemake sure that, on average, both groups will be similar in all ways (even similar with respect tounknown influences on the outcome) except the thing we're really interested in: the twotreatments.

Why doesnt anyone know which valve will be used until the last possible minute? This certainly seems

weird, doesn't it? The whole aim is to make the two treatment groups as similar as possible. Even ifthey're randomized it's still possible for biases to creep in. For example:

The surgeon, anesthetist and other theatre staff might make more careful preparations for theexperimental valve;

The surgeon might change his/her schedule to avoid giving one of the valves to a patient he/shewas concerned about;

etc. ...

In general, it is desirable, if possible, for neither the patient nor the doctor to know which treatment thepatient is allocated. This prevents any bias in the evaluation of the trial; the outcome for the patient ismeasured before anyone knows which group they were in. This is done in drug trials, where the twodrugs can be made to look alike.

Obviously, with an operation, the surgeon has to know which procedure is being carried out!How come theyre experimenting on humans? Isnt that illegal? Of course it would be illegal (andimmoral!) to experiment on humans in this way without their consent. These studies obtain informedconsent from the subjects; but the subjects have to consent to being randomized. That is, they consentto their treatment being chosen by the flip of a coin. The ethical justification for such a study is that wereally don't know whether valve A or valve B is better. If we did, we could not reasonably withhold fromthe patient the better option.

Fire-fighters cause damage?!Statistics of the amount of damage caused in house fires show that the larger the numberof firefighters attending the scene, the worse the damage! What do you think the reasonis?


12/14

The apparent association is due to the omission of some important information. In the example of housefires, the size of the fire needs to be taken into account --- more firefighters are sent to larger fires andthe larger the fires, the worse the damage.In many situations, the explanation for some apparent association cannot be identified easily. Oneexample is the association between smoking and lung cancer. It has been argued that the apparentassociation between the two may be due to some genetic factor that predisposes people both to nicotineaddiction and lung cancer. If this is true, then smoking cannot be blamed for causing cancer. It is onlyafter considerable research, with the aid of statistical methods, that it is now generally accepted thatsmoking is a contributory cause of lung cancer.

What was the question again?What is your response to the following question?

"Lately there has been much talk about euthanasia. Would you support or oppose theintroduction of a law which protects doctors who assist terminally ill patients whochoose to end their own lives?"

The wording of questions is often a disputable aspect of a survey. The above question wasused by the AGB McNair Age Poll and it was reported in The Age of 7 June 1995 that three outof four Australians support the legalisation of euthanasia. However, Dr David Weedon, thefederal president of the Australian Medical Association, objected to the wording. He said: "Youhave asked people whether they support a law which protects doctors. I would think that the result isshowing a vote of confidence in the medical profession, which is unrelated to whether they supporteuthanasia." It is the responsibility of a statistician involved in questionnaire design to ensure thatquestions are not worded in a way that would lead respondents to give biased responses.

German measles and birth defects

The link between rubella (German measles) in pregnant women and birth defects intheir babies was not an easy one to find, because the disease can be relatively mild.After Gregg and others had suggested a link, the Australian statistician OliverLancaster was able to show that at the 1911, 1921 and 1933 Australian censuses therewere peaks in the age distributions of deaf people, corresponding to a cohort born

around 1899, when there was a known epidemic of rubella.

German tanks in WW2

In World War 2, the Allies used statistical methods to estimate German militarystrengths. For example, to determine how many tanks the Germans had in 1943, theAllied Economic Warfare Division in London analysed the serial numbers on capturedGerman tanks. In the simplest form, each serial number gives information -- a serialnumber of, say, 100 means there were at least that many tanks manufactured.

Using similar but more sophisticated statistical methods, the Allies made the estimates shown in thetable below. Allied intelligence agencies were also making estimates based on other information, andthese are shown too. All data are monthly production values.

Date of Statistical Intelligence Germanestimate estimate estimate records-----------------------------------------------------------

June 1940 169 1000 122June 1941 244 1550 271August 1942 327 1550 342In this case the actual numbers became known from the Speer Ministry after the German surrender, sothe true values are known and shown in the table too. The statistical methods gave much better results.

TV phone-in polls are a load of rubbish!

You may have seen TV news programs holding a phone-in poll, in which a simple yes/noquestion is asked, such as: "Should Australia cut off all diplomatic ties with France?"Viewers are invited to dial one number for "yes" and another for "no". These surveys

cannot possibly give a reliable indication of the views of the population in general, since there is nothing


13/14

to stop individuals dialling the number of their preferred answer as many times as they like, and interestgroups can do the same thing. To get reliable information about the views of the general public, there isno substitute for careful, well-designed sample surveys, with random selection of the samples.

Source: Statistical Consulting Centre The University of Melbourne 1994-1999

Most people are familiar with weight-for-height tables. For over half a century, dieters, doctors, andinsurance companies have been using weight-for-height tables to determine whether a person isoverweight or at an ideal weight. Many versions of weight-for-height tables exist, all with differentweight ranges. Some tables have weight ranges based on frame size, some tables separate men fromwomen, and some take a person's age into account.

The Metropolitan Life Insurance Company (MLIC) introduced the first weight-for-height tables. "In 1942,Louis Dublin, a statistician at Metropolitan Life Insurance Company, grouped some four million people

who were insured with Metropolitan Life into categories based on their height, body frame (small,

medium or large) and weight. He discovered that the ones who lived the longest were the ones whomaintained their body weight at the level for average 25-year-olds."- Excerpt from NutriBase Software

The 1942 tables gave "ideal body weights" and were used by MLIC to determine insured persons with thelongest life spans. The tables were revised slightly in 1959 and this time called "desirable body weights."In 1983, they were revised once again and given the name "height and weight tables." Even though theMLIC tables have been revised over the years and called by different names, people still refer to them asthe tables of "ideal body weights." The biggest problem with the MLIC weight-for-height tables is theyare based on a lie. The MLIC tables list different weight ranges for people with small, medium, or largeframes, but the source data had no information about frame size."Some clever fiction writer created frame sizes. I think they [insurers] were embarrassed the weightranges were so wide."

- C. Wayne Callaway, M.D. (member of the 1990 Dietary Guidelines Advisory Committee)"Frame size as used for estimation of lean (fat-free) body mass is subjectively determined in the 1959tables."- Health Implications of Obesity. NIH Consens Statement Online 1985 Feb 11-13; 5(9):1-7

The MLIC tables of 1959 and 1983 are just updated versions of the 1942 tables of ideal body weight.The updated versions are based on the same data from the four million life insurance policies that wereoriginally used for the tables of 1942.

The made up frame sizes arent the only problems with the MLIC weight-for-height tables. Experts havecriticized these tables for several other reasons:

When these tables were first introduced, the people who had life insurance at the time were mostly

middle-aged, middle-classed white men.

1. The tables were designed for adults aged 25 to 59 years, but the data that the tables were basedon didnt include many young adults because young adults couldnt afford life insurance.

2. Insured people tend to be healthier than uninsured people.

3. Not every person was weighed.

4. Some wore shoes and/or clothing and some didnt.

5. The tables do not take into consideration family health history.


14/14

6. The tables do not consider body fat percentage or body fat distribution.

"Credit should be given to the MLIC table creators for attempting to use "frame size" as a way tocompensate for the differences between peoples body shapes and skeletal muscle mass. In theory,

elbow-width or wrist-width does correlate fairly well with muscle and bone mass. But in practice, thedefinition of frame size is too difficult for people to use, so virtually nobody uses frame size as intended.

Instead, people subjectively choose their own categories."- Steven B. Halls, MD

For more information view: http://www.realitypress.com/forum/viewtopic.php?t=35http://www.halls.md/ideal-weight/met.htmFor better ideal body weight calculation methods, visit this page: http://www.halls.md/ideal-weight/body.htm

Documents

Statistics Articles