There is a huge disconnect between health factors that research suggests are most important, and health factors that get the most media and policy attention. A new RWJF working paper suggests that the press overemphasizes obesity to satisfy reader demands:
I think that the condemnation of fat is about aesthetics more than health concerns. Fat is ugly and so people want others to loose weight. I also think that aesthetics is why people turned in smoking, it stinks.
The methodology involves coding the articles based on thewords present in the text. This leads to the conclusion onpage 19 that
Our analyses suggest that the news media tend todramatize the risks of obesity by using, more than thescience on which they are reporting, words like"epidemic" and "war" and by blurring the lines betweenweight categories, giving an impression that thepopulation is heavier than it is.
However on the next page the authors note that
The 1999 JAMA issue and news reporting on that issueoverwhelmingly represented overwieght and obesity as acrisis, at 70% and 72% respectively. This framing wasless prevalent in the 2003 special issue and newsreporting on that issue, at 40% and 34%respectively. This does not mean that the 2003 articlestended to /counter/ claims that obesity was acrisis. Rather, compared to 1999, they were more likelyto take them for granted so that they did not need to bemade explicit.
What do I think this means? My understanding is that in 1999scientists wrote sentences like "Our treatment targets thecurrent epidemic of obesity...". In 2003 they wrote the samekind of thing in their first draft, but when it came to thestage of squeezing their articles to fit within the pagelimits, they thought "We can drop references to the epidemicbecause we are writing for our colleagues who read the 1999special issue and already know that their is an epidemic ofobesity.". So in 2003 the final version reads "Our treatmentfor obesity ...".
Meanwhile journalists are writing for a lay audience thatis only vaguely aware of the 1999 JAMA special issue onobesity and only vaguely aware that there is an epidemic. Sothe journalist knows that the JAMA authors took out the wordepidemic because /their/ audience didn't need it, and thejournalist correctly puts the missing word back in because/his/ audience needs it fully spelled out.
The methodology codes the actual texts of the articles, soit detects dramatization in the news papers, because thenews paper text contains the word "epidemic" when theoriginal article did not.
The article destroys its own methodology when it claims thatthe authors of the JAMA articles /meant/ "epidemic" but hadno need to actually use the word in their text to conveythis to their audience. The article says that you have toread between the lines but its methodology doesn't codebetween the lines.
"The press/policy overemphasis of obesity is probably small compared to the overemphasis of medical care."
Perhaps a little biased, Dr. Hanson?
Alas, I don't understand how the paper's method is destroyed by the fact that "later special issue in the medical journal leaves off the hype about the obesity epidemic."
I found reading the paper hard going, and I ended up skimming, because I felt I was having to fend off dodgy assumptions and omissions as I read.
On page three the paper skips lightly over the inadequacies of the Body Mass Index(BMI). There are perhaps three things wrong with BMI.
First it is mass divided by height squared. Why squared? The obvious approach is to assume isotropic scaling, in which case mass goes as the cube of the scale factor, and one allows for taller people being heavier by dividing by the cube of the height.
If you admit the assumption of isotropic scaling then the three that enters the equation as an exponent is an exact integer determined by the three dimensional geometry of space. Actually the assumption is crap you need a smaller number, in the region of two, to get sensible results. Whithout the assumption of isotropic scaling one drifts in water much deeper than ones anchor chain. The exponent of two can only hope to be a rough and ready empirical fudge factor. If it should really be 2.2 you are not fully allowing that taller people should be heavier.
This is schoolboy arithmetic, so journalists could take scientists to task for using a bodged formula without thinking about what their fudge factor is doing to the results. Journalists are not picking up on this, and the paper itself seems to be part of the problem. If you want to overcome bias you need to be sharper at picking up on formula-worship.
Second, I remember hearing about research by life assurance companies that your trouser waist size was a much better indication of mortality than BMI. I'm not in the health biz so I would have a hard time tracking down the primary reference. This was no surprise (see problem 3) so I thought "It is all over for BMI" but no, its 2007 and uncritical use of BMI goes unremarked.
Third, if a middle aged sedentary worker swears off beer, starts exercising, going to the gym, or swimming or cycling, pretty much any serious exercise, he will gradually lose his paunch and gain some muscle mass, shoulders, arms, legs, it depends on the exercise. This will aid his life expectancy, show up clearly in his trouser waist size, and cause problems for BMI. It takes a long time to build muscle mass, so one expects a fall in weight. Then again, it is hard to shift a paunch. BMI is not a good index to use as a proxy for health. Even though I'm just a lay person, it was nosurprise to me to hear the life assurance companies had found trouser waist a better proxy for longevity.
The problems with BMI are suitable for treatment in newspapers, so omitting this story strikes me as a large bias thathas to be included in any examination of press treatment of obesity.
I've run out of time, I'll cut to the biggest problem I see in the paper.
On page twenty the authors are noting that later special issue in the medical journal leaves off the hype about the obesity epidemic. It is not that they are backtracking, it is now an established fact, established by popular acclaim, so the jounal can take it for granted. You cannot just count the words, you have to read between the lines, understanding the text in the cultural context in which it was written, and making careful note of the date of publication inorder to accurately identify that context.
That is inconsistant with the quantitative approach that is the core of the article. It pretty much destroys its ownmethodology.
The majority want to confirm their biases about fatness, since fat people are today's acceptable scapegoats and laughingstocks - a suitable fill-in for blacks, jews, and other no-longer-PC punching-bags. Believing that being overweight is extremely unhealthy is seen as a mental rationalization justification for the opprobrium in which fat people are held. Just below the surface, fatness is seen as a moral failure.
If I had to bet, I would suspect that lots of weight-loss cycles and severe dieting are much more unhealthy than carrying some extra fat pounds.