of 47 /47
Lies, Damned Lies and Statistics Introduction 8 Questions To Ask of Statistics 5 Statistical Concepts
• Author

stacie
• Category

## Documents

• view

44

3

TAGS:

• #### important social issues

Embed Size (px)

description

Lies, Damned Lies and Statistics. Introduction 8 Questions To Ask of Statistics 5 Statistical Concepts. Polls & Surveys. Polls and surveys are simply attempts to discover what people think about a given topic. These have limited value at the best of times. - PowerPoint PPT Presentation

### Transcript of Lies, Damned Lies and Statistics

Polls & Surveys

Toronto Sun, March 28, 2007. Battle with blubber. The situtation has reached epidemic proportions as obesity rates among children have risen almost threefold between 1978 and 2004.14Example 7:When my mother-in-law sits around the houseLets do some math.From the AHN Media Corp. story: Committee chair Rob Merrifield said he was shocked by the study, which said 26 per cent of Canadians between the ages of two and 17 are overweight or obese. In 1978, overweight children only accounted for 15 per cent.Using the percent change formula: Subtract old value from new value and divide by new value, multiply by 100. 1978: 15% of children overweight2007: 26% of children overweight((26-15)/15)X100 = 73.333That s a 73% increase.From the Toronto Sun story:The situation has reached epidemic proportions as obesity rates among children have risen almost threefold between 1978 and 2004.A threefold increase would be 200%Where is the discrepancy coming from?

15Example 7:When my mother-in-law (cont.)Some more analysis:Overall, the report found the proportion of children aged 2-5 who were overweight or obese has stayed virtually the same from 1978 to 2004. However, the proportion of overweight children aged 6-11 doubled while the rate was up 14 per cent to 29 per cent for children aged 12-17.Ages 6-11 increased by 100%Ages 12-17 increased by 29%What happened to the other 79%?

Obesity Scandal, Health Care News, May 1, 200516Example 7:When my mother-in-law (cont.)In March 2004, Tommy Thompson, then Secretary of the U.S. Department of Health and Human Services, joined representatives of the Centers for Disease Control (CDC) and National Institutes of Health at a joint news conference warning, "Americans need to understand that overweight and obesity are literally killing us."At the news conference, the CDC released a study concluding obesity is now the second leading cause, behind tobacco, of preventable, premature death in America. The report attributed some 400,000 deaths per year to obesity. Almost immediately, the study came under heavy criticism. The May 2004 issue of Science magazine fired the first volley, reporting that CDC scientists who had cast doubt upon the reliability of the 400,000 figure (one called it "loosey-goosey") were ignored."I am worried that the scientific credibility of CDC likely could be damaged by the manner in which this paper and valid, credible, and repeated scientific questions about its methodology have been handled," Terry Pechacek, associate director for science in the CDC's Office on Smoking and Health, told the Wall Street Journal last year."I would never clear this paper if I had been given the opportunity to provide a formal review," said Pechacek.The Journal conducted its own review of CDC documents and reported in November 2004 that the 400,000 figure was inflated by approximately 20 percent because of a statistical error. Obesity Scandal, Health Care News, May 1, 200517Example 7:When my mother-in-law (cont.)About those definitionsBeginning in 1942, the Metropolitan Life Insurance Company developed height and weight tables for its insureds, taking into account gender and frame size. The tables, widely used to identify "desirable" body weight, were revised upwards in 1959 and 1983.

The federal government adopted the BMI in the 1990s as a guideline to help doctors determine when to address medically their patients' overweight or obese status. Initially, the BMI tables used by the federal government labeled men as overweight if they scored 28 or above, and women at 27 and above. In 1998, the National Institutes of Health lowered the overweight score to 25 for both men and women.

Under the 1999 Metropolitan Life tables, a 5'3" woman with a large frame and a weight of 151 pounds was not considered overweight. Under the revised BMI, however, she has a BMI score of 27, solidly in the overweight column.Obesity Scandal, Health Care News, May 1, 200518Example 7:When my mother-in-law (cont.)This decision [to redefine overweight and obese] was made by a National Institutes of Health obesity panel chaired by Xavier Pi-Sunyer, one of the most influential obesity researchers in the country. Over the years, Pi-Sunyer has received support from virtually every leading weight-loss company, including Novartis, Sanofi-Aventis, Ortho-McNeil, Wyeth-Ayerst, Knoll, Weight Watchers, and Roche. He has served on the advisory boards of Wyeth-Ayerst, Knoll, Abbott, Johnson & Johnson, and McNeil Nutritionals.He once headed up the Weight Watchers Foundation and is currently a board member of that organization. Pi-Sunyer gave the "obesity overview" presentation on behalf of Knoll, maker of the weight-loss drug Meridia, at a 1996 FDA advisory panel hearing on the drug. He has also been paid to sign his name to ghost-written journal articles used to promote the dangerous weight-loss combination known as "fen-phen."

Obesity Scandal, Health Care News, May 1, 200519Example 7:When my mother-in-law (cont.)Who Was Behind the Redefinition of "Obese"

Case Western Reserve University professor Paul Ernsberger describes how financially conflicted researchers control the governments pronouncements on obesity:"Medical beliefs about obesity are shaped by expert panels that are highly selective in the data they consider. Experts included on government consensus panels have been disproportionately drawn from the ranks of diet clinic directors, which might explain the congruence between panel recommendations and the economic interests of the diet industry. In short, economic factors encourage a systematic exaggeration of the health risks of obesity." Many of Americas most influential obesity experts receive significant financial support from the \$46 billion weight-loss industry. These experts help drive obesity hype by churning out a steady stream of studies, alarmist public pronouncements, and treatment guidelines.

Obesity Scandal, Health Care News, May 1, 200520Warning: Pictures of Overweight & Obese People21Warning: Pictures of Overweight & Obese PeopleObese

22Warning: Pictures of Overweight & Obese PeopleObese

23Warning: Pictures of Overweight & Obese PeopleObese

24Warning: Pictures of Overweight & Obese PeopleOverweight

25Warning: Pictures of Overweight & Obese PeopleOverweight

26Example 8:Are there enough cemeteries?When doing his thesis in 1995, a graduate student grabbed media attention with his statistics that the number of children killed by gunfire had doubled each year since 1950. This statistic was picked up an re-printed in various publications.Following is a chart showing what these figures would mean.27Number of children killed by firearms doubled each year since 19501950195119521953195419551956195719581959248163264128256512102419601961196219631964196519661967196819692,0484,0968,19216,38432,76865,53665.536131,072262,144524,28819701971197219731974197519761977197819791.049 mil2.097 mil4.2 mil8.4 mil16.8 mil33.5 mil67.1 mil134.2 mil268.4 mil1536.9 mil19801981198219831984198519861987198819891.1 bil2.2 bil4.3 bil8.6 bil 217.2 bil34.4 bil68.7 bil137.4 bil274.9 bil549.8 bil19901991199219931994199519961997199819991.1 tril2.2 tril4.4 tril8.8 tril17.6 tril35.2 trilLarger than the population of the United States in that year.Larger than the population of the entire world.28Cemeteries (cont.)The student had misread a 1994 report by the Children's Defense Fund that found the number of American children killed each year by guns had doubled since 1950 not doubled every year since 1950. In other words, it had increased 100%, not 17.6 trillion %.Even this statistic isnt as alarming as it might appear at first since the population has increased 73% since 1950, meaning that in 44 years there has been a 27% increase in the number of children killed by guns not 100%29Eight Questions to Ask of StatisticsWhere did the data come from?Have the data been peer-reviewed?How were the data collected?Are the comparisons appropriate?Are the numbers in context?Are the definitions valid?Has there been a change in circumstance?Are the questions neutral?

301: Are you looking at a study, or a report on a study?A study, and a report about a study are not the same thing.If it is a report, does it cite the original study so you can find it? Best of all is when a web site links directly to the orignal study.A study may say something very different from what the report on the study is claiming. This discrepancy is often the result of conscious manipulation, but it is also commonly the result of poor and lazy reporting.