Correlation, Causation and County Maps

lemurcat12 wrote: »

I had a thought about this over the weekend and keep forgetting to post it here.

One reason I am skeptical about the focus on individual differences between or within counties being the big explanation--or maybe not the key focus--is that they overstate the difference. While CO looks good compared to MS, it looks horrible compared to CO (and some other states) 20 years ago. The colors on the map create the appearance of a big difference between a 24% and 28% rate, when the dramatic story is the change over time.

http://stateofobesity.org/adult-obesity/ -- current state map with trends

http://obesity.procon.org/view.resource.php?resourceID=006026 -- map with illustration of change over time

Data visualization is an interest of mine (and a facet of my job). Colors are one of the worst ways to depict changes in magnitude. Humans aren't able to make the fine distinctions of intensity or color gradations needed.

One thing I'm learning from looking at these maps is that whomever put them together is not educated in proper data visualization techniques. Data comparisons between the maps, even judging the data on a single map is much too difficult for many of them.

stealthq wrote: »

lemurcat12 wrote: »

I had a thought about this over the weekend and keep forgetting to post it here.

One reason I am skeptical about the focus on individual differences between or within counties being the big explanation--or maybe not the key focus--is that they overstate the difference. While CO looks good compared to MS, it looks horrible compared to CO (and some other states) 20 years ago. The colors on the map create the appearance of a big difference between a 24% and 28% rate, when the dramatic story is the change over time.

http://stateofobesity.org/adult-obesity/ -- current state map with trends

http://obesity.procon.org/view.resource.php?resourceID=006026 -- map with illustration of change over time

Data visualization is an interest of mine (and a facet of my job). Colors are one of the worst ways to depict changes in magnitude. Humans aren't able to make the fine distinctions of intensity or color gradations needed.

One thing I'm learning from looking at these maps is that whomever put them together is not educated in proper data visualization techniques. Data comparisons between the maps, even judging the data on a single map is much too difficult for many of them.

That's very interesting -- I'd love to read more about data visualization and how it's done well. Do you have any thoughts about better ways to convey these?

tomteboda wrote: »

Kimberly_Harper wrote: »

Holy crap that's a big change in obesity percentage in a matter of 10 years!

The CDC also redefined obesity in that period (1998 to be precise). I don't think they updated their old map or statistics to reflect the new definition, however.

I think the maps are comparing like to like -- the second set specifically say they are using BMI>30 for the obesity definition, even for the earlier ones. (The change was 28/27 to 25 for overweight.)

tomteboda wrote: »

Kimberly_Harper wrote: »

Holy crap that's a big change in obesity percentage in a matter of 10 years!

The CDC also redefined obesity in that period (1998 to be precise). I don't think they updated their old map or statistics to reflect the new definition, however.

For what it's worth, the American Association of Clinical Endocrinologists (AACE) and the American College of Endocrinology (ACE) is suggesting yet another change to the definition obesity. This time it is a functional change, but it would make a BMI > 25 "obese", if you have any health issues correlated to obesity.

Obesity-related complications include the following: metabolic syndrome; prediabetes; type 2 diabetes; dyslipidemia; hypertension; nonalcoholic fatty liver disease; polycystic ovary syndrome; sleep apnea; osteoarthritis; gastroesophageal reflux disease; and disability/immobility.

Look for millions more Americans to be labeled obese in the near future if these recommendations are taken up by the CDC.

Hmmm. This piqued my interest. I used to teach research methodology for majors in the sciences, social sciences and humanities, and one of the fundamental things I taught was the shenanigans that can take place with definitions. The CDC is at the same time a political and scientific entity (although all science has a political aspect to it), so I wonder if the change in definition has anything to do with public policy shaping: http://content.healthaffairs.org/content/21/6/142.full

It's an old article, but very interesting as it contextualizes obesity with drinking, drugs, sexuality and tobacco use. Here is a passage specifically on poor/minorities:

"Demonizing users—especially poor people and minority groups who drink, take drugs, or harbor sexually transmitted diseases—has been one of the most powerful spurs to government action in U.S. history. There is nothing quite like the fear of sinister others to overcome the stalemate of American policy making.

Although overweight Americans have faced popular prejudice for more than a century, critiques of gluttony have not translated into demonization. Antiobesity activists do not portray overweight people as dangerous to society—like drug addicts or smokers polluting the air with secondhand toxins. In part this may be because more than half of U.S. adults are overweight, and nearly one in five is obese.15 Still, each of the other cases challenged a commonplace activity or condition. In 1965, for example, an estimated 43 percent of American adults were habitual smokers—a figure that has plummeted with changes in social mores, regulatory efforts, and disapproval bordering on the demonization of smokers.16

One common thread in past demonization episodes is at least latent in the obesity case. Poor people and members of minority groups tend to be more obese than other Americans are.17 Given the historical patterns of other ostensibly private consumption practices, the opportunity for demonization may well be present. But, to date, this has not been taken up by those calling for action against obesity.

Demon industry.

In all four of our comparative cases, activists attack the producers or suppliers. They charge corporate villains with seeking profits by peddling poison. Worse, the greedy industry lures children into destructive habits."

And one of the conclusions is as follows:

"Although social scientists often depict the U.S. government as relatively weak, it has been far more ready than most Western regimes have been to regulate (or prohibit) private behavior. The politics of social control generally feature the seven triggers discussed here. Of course, political history does not permit causal claims, but we believe that these descriptive analogies across time and issue areas offer a useful policy guide. Context also matters: In every example of state intervention, political action becomes possible when a “window of opportunity” opens. Even when all seven triggers are in place, policy efforts may fail—without propitious circumstances, luck, timing, or a political plan primed to go when opportunity strikes."

tomteboda wrote: »

lemurcat12 wrote: »

I think the maps are comparing like to like -- the second set specifically say they are using BMI>30 for the obesity definition, even for the earlier ones. (The change was 28/27 to 25 for overweight.)

I actually don't think they actually do, though, because the graphs changed dramatically overnight as well. I remember discussing this when it happened in my biomedical ethics class at UT-Houston MD Anderson Graduate School of Biomedical Sciences.

I'm talking about this specific set of maps (the ones I linked above), which are labeled "over 30 BMI" back to 1990. I'd be curious if you can find something to suggest they are labeled wrong. The change from 1995 to 2000 seems similar to the change from 1990 to 1995, and I thought the obesity rate remained at 30, but the overweight cutoff changed.

jgnatca wrote: »

A weird map on what takeout Americans order most.


I found this because of an argument that vegetarians live longer healthier lives based on nearly every marker. I figured some parts of the US have more vegetarians than others.

I love this map!