Bad Math Strikes Again- Part 1- Fat, Lazy Americans

Today has been wrought with bad math in important subjects where people should not be misled so I'm going to do another double down day and post on both, starting with the one that's made me less angry. This less aggravating topic comes from an article on msn.com found here.

The article is discussing the obesity problem in America and breaks down what are considered the 10 most obese cities/metro areas in the nation. Their goal is apparently to create a hall-of-shame for these struggling metro areas. It also flaunts the government's lofty goal, set in 2010, to reduce the national obesity rate from 1/3 of all Americans down to just 15%. How does the government plan to accomplish this goal? The only effort listed in the article is a list of 70 ways to prevent and control childhood obesity published in 2010. More importantly, since when is it the government's job to control the weight of its citizens? And by extension what we eat, how much, and when we eat it?

Here are some numbers for you to chew on:

City	Obesity rate	Health care costs	Resident population	Diabetes rate	Poverty rate
Reading, PA	32.7%	$190.2	88,000	10%	35%
Kennewick-Pasco-Richland, WA	33.2%	$116.5	No data	No data	No data
Topeka, KS	33.3%	$109.8	127,473	No data	No data
Lakeland-Winter Haven, Florida	33.5%	$279.3	No data	No data	No data
Charleston, West Virginia	33.8%	$146.9	51,000	17%	No data
Beaumont-Port Arthur, Texas	33.8%	$182.8	No data	No data	No data
Rockford, Illinois	35.5%	$179.4	152,000	10%	23%
Huntington-Ashland, West Virginia-Kentucky-Ohio	36%	$146.9	No data	20%	No data
Binghamton, New York	37.6%	$131.5	45,000	No data	27.8%
McAllen-Edinburg-Mission, Texas	38.8%	$410.9	No data	No data	No data
National Rates (gathered independently from the article)	35.7%	?	311 million	8.5%	15.1%

Health care costs are all represented in millions of dollars. This data, except the final row, comes directly from the article.

There are a few important things to note in this table:
1) all but three of the obesity rates are actually under the national figure (which I obtained from the CDC website)
2) all of the cities with data on the resident population are relatively small populations which makes this number more easily skewed in one direction or the other
3) only Lakeland-Winter Haven and McAllen-Eidenburg-Mission have health care costs above $200 million and, in the case of Lakeland-Winter Haven, this may be attributed to other circumstances such as a much larger elderly population. McAllen-Edinburg-Mission's health care costs are so much higher than any of the other cities with similar obesity rates that it is likely that this city also has some other condition that is effecting the health of its people.
4) the metro area with the second highest obesity rate has the third lowest of the 10 areas as far as health care costs, probably due to the extremely small population

Other than the numbers being rather inconclusive, I have several other issues with this article, and the panic over the "epidemic" of obesity in general.

My first issue is the most generalized and concerns how we judge one person to be obese and another person not. I'm talking about the BMI scale. Now, there are actually many different BMI graphs and calculators circulating out there but the concept is that you calculate between your weight, height, and in some cases gender and age, and are given a label such as under-weight, normal, over-weight, and obese. The scale does not take into account any other important factors that can affect weight such as ethnicity, and muscle tone/mass, it just assigns you a label and you become another statistic.

It would be one thing if the BMI labels were fairly liberal to account for the inaccuracy of the scale. The problem with the BMI is that, looking at the various graphs, there seem to be some rather radical determinations. In one graph I found, if you are 6' tall and weigh 135 lbs you are considered normal but if you are 5' tall and 130 lbs you are considered over-weight. In another graph, if you're 6'3" and 200 lbs you're bordering on obese and if you're 4'11" and over 120lbs you are over-weight. It's bad enough that the media idolizes women who look as if, and sometimes do, starve themselves in the name of beauty. Now the medical professions is labeling people as "over-weight" when they are otherwise healthy, active individuals.

The next issue I have with the article regards the health care costs, specifically the estimated savings for each metro area if they drop down to just a 15% obesity rate. Notes 3 and 4 on my table begin to touch on this but I would like to expound on it further by saying that there is no way to accurately predict the amount of health care costs that go specifically to treating complications from obesity. Doctor's don't bill an illness as "obesity" (except maybe in the case of stomach staples). The human body is a complex organism and a myriad of factors contribute to health conditions including genetics, environment, age, gender, ethnicity, and yes, weight. But the article's continued use of diabetes as a "chronic disease associated with obesity" obscures the fact that there are two different types of diabetes and they can affect thin and average people as well. The same goes for heart disease and high cholesterol. True, these illness are seen in greater prevalence among those who are on the heavier side, but they are not exclusive to those who are over-weight and their health care costs should not be treated as such. Would reducing the obesity rate result in a decrease of health care expenses? Highly likely. But the numbers are way too far into conjecture to be reported in a news article with any accuracy.

Finally, as you will note, there are many fields in the table with missing data. I understand that news writers have a limited amount of column space, but the gaps leave many questions unanswered. Were the facts left out to obscure data that contradicts with the overall trend or was it just left out to preserve space? The concluding paragraph of the article makes me concerned:

The Gallup-Healthways Well-Being Index results are based on telephone interviews throughout 2011, with a random sampling of 353,492 adults living in the U.S. Health-care costs were based on the National Institute of Health’s estimate of $1,429 per person, per year, in additional health-care costs for people considered obese, compared to those of non-obese individuals.

The sample size is very telling to me. It is 0.1% of the national population that they are drawing on to make the determination that these ten metro areas are the most obese out of the entire nation. When divided around the country, the number of people polled for these particular locations must have been very small (probably less than 100 in many cases). They list no margin of error in the article or how many people were polled in each metro area so I'm going to have to call bull-shit.

Overall, this article gets a 3 on my FT Scale for both people who read it and take it at face value and the people who conducted the survey if they think there is a strong statistical significance in their data.