By Mark Frauenfelder at 12:13 pm Tue, Mar 8, 2011
(Left: Andrew Sullivan's Map of the Day. Right: Diabetes Belt stretches across Deep South and Appalachia)
Also not eating pickles will save your life. Everyone who eats a pickle at least once in their life will eventually end up dead.
Anybody have a map of education levels? poverty levels?
Detailed information like that requires census information and that’s just a tool of evil big gov’t
guessing tongue is firmly in cheek with the conclusion
I’d better renew my vaccination – my passport is about to expire.
“Anybody have a map of education levels? poverty levels?”
I think that’s kinda the point of this article. “Correlation does not imply causation” and all that.
Now if you will excuse me, I’m going to do my part in fighting global warming by pirating…
The diabetes prevalence map is like a map of where to get good food. See the good food indicator there in New Mexico? That is where you get those awesome green chili scrambled eggs. See the line that runs up the Appalachian Mountains? Best deserts and BBQ pork in the US. Coastal gulf and lower Atlantic? Shrimp and crab cakes going on, with some top spot contending BBQ. I’m not sure about Northern Maine, but I’ll bet it involves some butter dipped lobster.
“I’m not sure about Northern Maine, but I’ll bet it involves some butter dipped lobster.”
That’s Aroostook County, which has countless potato farms. The diabetes likely comes from their habit of having french fries smothered in gravy with many meals. (It’s beautiful country with fantastic people. I miss going up there to work; I live in southern Maine.)
As for the lobster – Maine is a poor state, lobster is a rare treat for those who don’t live on the coast.
Interesting about the gravy fries and lobster. I’m more of a biscuits and gravy man myself, but if done right gravy fries sound pretty tasty.
I don’t think northern Maine has much to do with lobster seeing as it isn’t close to the coast. I suspect it has a lot more to do with poverty in that area and lack of access to medical care.
Obligatory “Correlation does not imply causation”
Fett101, so you’re saying a joke’s subtlety doesn’t necessarily cause it to go over some heads?
Obligatory, DUH, no shit Sherlock. That was the whole point of the post. :)
Well shoot, if you lived in Alabama, why would you want to go anywhere else?
Aren’t passports property of the government?
I live in the heartland, but I guess we can call the deep south the fatland, then.
coloration is not caucasian.
@drog – These results would match up closely with level of education and income, I feel certain.
Well.. I demand a refund, getting the passport was damn expensive, especially the hoops we had to jump through for the children, and I still wound up a diabetic.
Diabetics: the government is coming to take away your passports.
That’s it, I’m measuring my blood glucose before and after getting a passport.
One thing that I think needs mentioning is that too often I see people dropping the “correlation != causation” argument in cases where it may not be the most relevant aspect of the issue to bring up.
Correlated variables still have some uses, even when they’re not known to be causally related. If you know that two variables are very strongly correlated, given information about one variable you can make some guesses about the state of the other.
This is not a very useful approach given this example. But say you have a situation where you are looking at a community where you have information about relative socioeconomic status and incomes, and you want to try to guess something about education – highest completed level of the parents or likelihood of finishing high school or something like that.
Or we can look at the earliest studies of lung cancer and smoking. Smoking was demonstrated to be a very strong risk factor for cancer long before it was proven from a medical or biological point of view.
Back to the education example. There may be some causality at work in a hypothetical, but there are a myriad of other unknown variables that will have different direct and indirect impacts. So no, you can’t assume that socioeconomic status is the main deciding factor in the second variable or vice versa (at least, not without a wealth of other data comparing the relative associations of plausible risk factors). But, let’s say you have data from another comparable, similarly sized community that shows clear, unarguable trends in the relationship between wealth and education. It’s not out of the question to make some assumptions about what that data will look like in your new population, as a tentative estimate before you can collect real data. If you’re presented with an area with lower income and wealth, it’s not out of the question to make comparisons to comparable subgroups in comparable populations and make predictions about what their level of education will be. And this isn’t at an individual level; you can’t walk up to one person and cross-reference their income. It’s a probabilistic statement about the likelihood of a certain outcome for a subset of the population.
If you see someone making a direct causal link based on correlations, definitely cry foul. But taking advantage of demonstrated correlations without making assumptions about causality is not out of the question. It’s not the best practice or the best type of evidence, but if it’s all you have it can suffice. And it definitely requires discretion; this example of passports and diabetes is (I assume intentionally) silly, but that doesn’t mean there aren’t cases where the above approach would be reasonable.
tl;dr The adage “correlation != causation” is overused and often misapplied, in my opinion.
Diabetes and xenophobia, something obese conservative trailer trash have in common!
I think it had less to do with xenophobia, and more to do with lacking the means to travel abroad.
I see your xkcd, and raise you a smbc.
I’m a dual-citizen with two passports. Extra protection!
And if you mash this up with the happiness quotient of the states, the best state to live in is Alaska. Maybe Palin is so wacko is because of where she lives, her health (jogging while 8 months pregnant?) and her insanely high levels of happy.
Ummm, Type 1 or Type 2?
Cuz their is a big difference.
Even more compelling*: Moving to Colorado will prevent you from becoming diabetic.
* where “compelling” involves accepting the premise of the original post.
colorado DOES imply caucasian.
This public service announcement brought to you by Snark Frauenfelder.
The subtly of this joke/post makes me a bit uneasy thinking of all the toolbags who could misunderstand or worse BELIEVE it.
Maybe this would happen less if passports were less sugary.
Why? You pay dearly for the Public Health Service to pump precisely this type of stuff at schoolchildren.
An infinite number of SAS programmers, given an infinite amount of time, will eventually produce a statistical correlation that is actually true.
“An infinite number of SAS programmers, given an infinite amount of time, will eventually produce a statistical correlation that is actually true.”
You don’t have to go anywhere near infinite. An association deemed ‘statistically significant’ usually means alpha=5% ie the association has 5% or less chance of having occurred due to random chance. So for every 100 statistically significant associations, 5 of them *still* happened by chance. This being why hypothesis testing is usually considered a much better idea than fishing through your data looking for statistically significant associations..
While most people are aware that correlation does not imply causation, I think the much cooler thing is the reverse – that causation does not imply correlation. One thing can cause another, yet the two things may remain entirely uncorrelated!
That’s actually really cool, and in hindsight, should have been obvious. Thanks!
I know a guy who has a passport AND diabetes. Clearly he is infected with a passport resistant strain of diabetes. Everybody panic!
I live in Mississippi. I have a passport. I do not have diabetes.
Oh wait … prevent diabetes…I see….
My question is related to the diabetes prevalence. There are certain confounders with all prevalence information including (but not limited to) age, sex and ethnicity. Generally we standardise for these in health care as population variances often have us looking in the wrong place otherwise.
So if this is the raw prevalence, I think it is unlikely you could even draw a correlation, much less infer anything from the correlation…
This is why planes don’t have to carry insulin.
These factors seem quite… confounding
correlation & causation blah blah blah…
yeah, BUT… I tend to lose weight whenever I travel. my work a day life is pretty sedentary, but when I travel I’m dashing about all over the place, maybe even eating healthier food (more fresh food, less convenient junk).
perhaps there is something to this.
Alternative conclusion: Eating passports causes diabetes.
Oh, I hate that I get rewarded with interesting stuff when I read the comments. Now I have to continue reading all the comments on every story. Thanks a lot, guys.
As a diabetic passport-holder, I demand my islet cells be restored immediately.
Breathing air is the number one leading cause of death. If you breath air, you shall eventually die.
If you ignore DC and use obesity as a proxy for diabetes (because it was easier to find at the state level) the correlation is r = 0.77 (r^2 =0.6). That is likely significant, but I used excel since I didn’t have my stats program on hand.
The good news is that you can counter passport-deficiency health risks with birthdays – studies show that the more birthdays you have, the longer you live.
It’s SCIENCE, people.
Uhhhh I don’t want to piss anybody off but. . .
For myself I can only say that I have gotten great pleasure from my passport. . .and a great deal of help from penicillin.
Daniel Patrick Moynihan noted that proximity to the Canadian border improved educational outcomes.
type 2 (TWO!) diabetes. seeing that the diseases’ onset cause is completely different (and since the topic is about the cause), please make sure to differentiate the two (type 1 & 2).
…hence why correlation does not equal causality
So if the Government really wants to help with diabetes then they should just subsidise passports. Problem solved.
Can we overlay (snaky or not) post origin about passport ownership vs diabetes?
I spotted some sympathy for the great unwashed up there somewhere. Look, I live in a diabetic hot spot and these tubs-o-lard would rather spend their money on Hummers and $300/month cable packages than broadening their horizons and traveling abroad. My inlaws bought a LARGE camper (which they had to pay to store elsewhere) while living in a house without central air. The purchase of a central air system (which costs roughly 1/10th of their camper) would have saved them THOUSANDS over the years. The windows in their home are the factory windows that came with the house (which came on the back of a truck) another minor thing they could have bought instead of their GIANT F350 that they purchased at the start of the gas crisis. And THEY are respected pillars of the community.
Point being; these people have major priority issues, not means issues. Do not cry for them. For the most part we’re not talking about coal mining towns with limited options. We live 50 miles from D.C. Cry, instead, for people like my wife who was brought up by these dumb, selfish bastards. She honestly thought that a trip to an amusement park was a luxury while her parents were flash-frying money. I was ACTUALLY poor and I’ve been to Europe.
And, for the record, why in the hell are we actually having a cor v cause discussion. My assumption is that you wouldn’t be here if you didn’t already get that. Don’t you people have jobs or hobbies?
I do have a job! i correlate data, so going to work causes me to correlate!
AND FYI one of them (the aforementioned inlaws) has diabetes. The other simply hasn’t been tested but has the wild mood swings one usually associates with blood sugar issues
Are you folks so dumb you don’t recognize satire when you see it? Yes.
Another think that’s dangerous is Latin. Every single person who grew up speaking it as their first language is dead.
Owning a passport prevents *the diagnosis* of diabetes.
I need to figure out what document prevents the gout, because that’s what I have. A fishing license maybe?
I have a passport, birth certificate, library card, drivers license, a social security card, a diploma from a state university, a turquoise star from eBay and a certificate from CafePress for selling a bunch of t-shirts in 2004. None of them are helping.
Dancentury–watch the fructose. http://www.fourhourworkweek.com/blog/2009/10/05/gout/
Does anyone else think that the reverse is likely more true? Being diabetic makes one less likely to travel abroad. This then reduces the need of passports.
Check out this religiosity map. And this education attainment map.
If you’re a baptist you most likely do not have a passport or higher education and you likely have diabetes?
It also appears that passport ownership prevents poverty:
I hope we make this correlation only in jest and remember that empathy is what is needed: Poverty truly can shrink your world, your life, and your potential.
How about an obesity map vs both?
what about owning 2 passports?
Good example. I’ve long used the alarm clock causing the sun to rise.
But another lesson from this association between passport ownership and diabetes is the powerful influence of the social environment. While we often focus on biomedical factors… pumping gazillions of dollars into research on diabetes drugs, etc…. here is a plain display of the bigger picture… that there are things in the lives of people that help explain both why some people are more likely to have passports and less likely to develop diabetes.
Paying a bit more attention to those upstream influences could help us prevent people from getting into trouble, so we are always having to put almost all our resources into trying to rescue them later.
Submit a tip
The rules you agree to by using this website.
Who will be eaten first?
Jason Weisberger, Publisher
Ken Snider, Sysadmin