Linguistics; language as persuasion; part three... My ignorance about the many ways the statement of a particular statistic can be worded and the mathematical procedures required to get from one of those ways to another is so profound that I can't even begin to explain what bothers me. I'm just going to give you two examples that, as I perceive it, illustrate the problem....
The first comes from Thomas J. Moore's article "The Cholesterol Myth," in the September 1989
Atlantic (pp. 37-70):
"An answer to the broad question of the benefits to life expectancy from lowering cholesterol levels through diet was found by a team of physicians and researchers at Harvard University. Using the risk-factor equations from the Framingham study and the diet results of the MR. FIT trial, the researchers calculated the benefits of a lifelong program of dieting. The team, led by William C. Taylor, published the results in the Annals of Internal Medicine in 1987. For persons without other risk factors, such as smoking or high blood pressure, they concluded 'we calculate a gain in life expectancy of 3 days to 3 months from a lifelong program of cholesterol reduction.' Taylor and his co-authors also said that they might be overstating the benefits of cholesterol reduction, because they had assumed that no other increased risk would arise from limiting cholesterol intake and offset any reduction in the risk of heart disease." [pp. 69-70]
Patients aren't ordinarily given cholesterol statistics in that format; they're told that if they follow a particular diet regime for the rest of their life their risk of dying from heart disease will be reduced by some significant percentage. And most of them, like me, lack the statistical sophistication required to know that that means they might live "3 days to 3 months" longer than if they ate what they wanted to eat for the rest of their lives instead.
I am oversimplifying, yes; I know that. So are the medical professionals who are presenting the patients with the statistics and the proposed diets, and so are the journalists who quote the medical professionals in the media.
[The only place I've been able to find this article as a non-PDF online is at
http://www.chelationtherapyonline.com/technical/p35.htm , where various comments and graphics-doodads have unfortunately been added to the text; however, the entire text is there, and it's easy to tell what was in the original and what has been added.]
My second example is an excerpt from "Both statistics and manner matter to patients," by Victoria Stagg Elliott, on July 9, 2007, at
http://www.ama-assn.org/amednews/2007/07/09/hlsa0709.htm . The section header is "What makes patients say yes?", and then it goes on like this:
"A study in the June 19 Annals of Internal Medicine suggests that the type of statistics a physician uses to explain a drug's ability to prevent disease may affect patient willingness to consent. Researchers randomized 1,754 participants in a general community health survey to be presented one of three hypothetical scenarios about a drug that could prevent heart attacks, and another thousand to scenarios about a medication that reduced the risk of osteoporosis. Here is how the various methods of presenting the numbers compare:
Survey situation 1: If a doctor says --
'For one heart attack to be prevented, 13 patients have to take this drug for five years.'
76% of patients would 'certainly' or 'probably' consent.
55% thought these numbers were understandable.
'The drug may not completely prevent heart attacks. It postpones heart attacks for a while, and all who take it for five years will live about two months longer before having one.'
54% would 'certainly' or 'probably' consent.
47% thought these numbers were understandable.
'The drug may not completely prevent heart attacks. It postpones heart attacks for a while. One of four patients who take the drug for five years will live about eight months longer before having one. Others will have no benefit.'
63% would 'certainly' or 'probably' consent.
51% thought these numbers were understandable." ...
When I teach verbal self-defense, I quote biomedical/epidemiological statistics taken from peer-reviewed journals that I consider to be respectable scientific sources. I would like to be able to word those statistics in the format that is most ethical, and most informative, and least coercive. And I would like to be able to tell, when I read those statistics in the journal articles, whether their presentation
there is in the format that is most ethical, and most informative, and least coercive.