States of Happiness

Ladies and gentleman, your latest state-by-state quantification of human feeling:



This map is based on a new study that finds correlations between subjectively reported happiness and certain objective factors like air quality, cost of living, and climate:

The new research published in the elite journal Science on 17th December 2009 is by Professor Andrew Oswald of the UK’s University of Warwick and Stephen Wu of Hamilton College in the US. It provides the first external validation of people’s self-reported levels of happiness. “We would like to think this is a breakthrough. It provides an justification for the use of subjective well-being surveys in the design of government policies, and will be of value to future economic and clinical researchers across a variety of fields in science and social science” said Professor Oswald.

The researchers examined a 2005- 2008 Behavioral Risk Factor Surveillance System random sample of 1.3 million United States citizens in which life-satisfaction in each U.S. state was measured. This provided a league table of happiness by US State reproduced below. The researchers decided to use the data to try to resolve one of the most significant issues facing economists and clinical scientists carrying out research into human well-being.

That issue: whether subjective reports of well-being (like those portrayed here) can be trusted. Seems that they can.

The study used subjective reports of well-being, but then checked those reports against a number of other variables for each state, including "precipitation; temperature; wind speed; sunshine; coastal land; inland water; public land; National Parks; hazardous waste sites; environmental ‘greenness’; commuting time; violent crime; air quality; student-teacher ratio; local taxes; local spending on education and highways; [and] cost of living." It turned out that the objective factors which would be expected to correlate with subjective happiness - nice climate, affordability, short commutes and all that - actually do correlate to the reported happiness of those 1.3 million surveyees. According to Professor Andrew Oswald, the lead author of the study:

“The state-by-state pattern is of interest in itself. But it also matters scientifically. We wanted to study whether people's feelings of satisfaction with their own lives are reliable, that is, whether they match up to reality -- of sunshine hours, congestion, air quality, etc -- in their own state. And they do match. When human beings give you an answer on a numerical scale about how satisfied they are with their lives, you should pay attention.

People’s happiness answers are true, you might say. This suggests that life-satisfaction survey data might be tremendously useful for governments to use in the design of economic and social policies,” said Oswald.

The happiest state is Louisiana (!), followed by Hawaii, Florida, Tennessee, and Arizona. The South does well in general, and the Northeast and Rust Belt not so much, which is interesting: happiness levels seem to be in strikingly inverse proportion to levels of economic and social development. The unhappiest state, it thrills me to report, is New York, followed by Connecticut and New Jersey - a trifecta for the tri-state!

The rest of the unhappiest quintile of states form a Bleak Belt from southern New England to the Great Lakes, with California thrown in for good measure. California can't blame it on the climate, of course, so their other factors must have been really brutal. On the other hand, Montana and Maine managed to sneak into the top tier despite their godforsaken climes.

Via the NY Times.

Support for US Health Care Reform in Three Dimensions

A nice map repetition from a New York Times op-ed shows support for health care reform along axes of income and age:



The accompanying opinion piece is by Nate Silver, Andrew Gelman, and Daniel Lee. Say the authors:

Using a statistical method called multilevel regression and post-stratification, we ... mapped opinion on health care, breaking down voters by age, family income and state. We’re used to thinking about red states and blue states, but the geographic variation is dwarfed by the demographic patterns: younger, lower-income Americans strongly support increased government spending on health care, while elderly and well-off Americans are much less supportive of the idea. But in general, senators seems to be less interested in what their constituents, old and young, rich and poor, might think about health care, and more interested in how they feel about President Obama.

This may actually be good news for the Democrats. Although the Annenberg surveys had shown health care subsidies to be quite popular — they had 67 percent support nationally in 2000 and 73 percent support in 2004 — that was back when they were a mere abstraction, and before voters might have been considering how to pay for them. Nowadays, President Obama enjoys higher approval ratings — in the low to mid-50s, according to most polls — than do the Democrats’ health care reform plans, which are mired in the mid-40s in most surveys. Conditions being what they are, Democrats would rather have a referendum on the president than one on the health care bill itself.

Support for Obama seems to be driving attitudes about health care reform to some extent, and not the other way around. Of course, a lot of what this has to do with is trust. As Machiavelli said, reform is hard: vested interests who benefit from the status quo will oppose it at every turn, and they tend to be well-organized, while support for reform tends to be diffuse and shallow. Whether you're going to support a large intervention in a system that, for all its shortcomings, is our system - the one most of us have grown accustomed to - will depend in large part on whether you trust the folks who are doing the reforming. And of course, if you're already a beneficiary of guaranteed government-provided health care, like everyone in the US over the age of 65, you really don't have much incentive to support reform; unless, that is, you aren't entirely self-interested, and actually care about, for instance, the ability of young people to acquire health care when they're in their 20s and don't have access to the kind of job stability that's necessary to acquire employer-based health insurance; or who can't get health care in the free market because of a pre-existing condition like asthma; or are just too poor to afford quality health care.

But of course the maps show a geographical dimension too. Support tends to be lower in Republicaan-leaning regions like the Plains and the Utah-Idaho-Wyoming triad of conservative Western states; it tends to be higher in the Northeast and Great Lakes states. (By the way, when and why did Wisconsin become more liberal than Minnesota?) What will be interesting to me is to see how support shifts once a bill is actually passed. My guess is that support will increase across the board, once there are a bunch of headlines about Obama signing "historic leegislation" and all that. On the other hand, I wouldn't put it past Congress to end up with such a watered-down bill, with so many sops to the health insurance and health care industries, that it just pisses everyone off.

The Spread of Swine Flu: Blame it on Louisiana

Nate Silver has a map based on data from Google Flu Trends that shows the timeline of the spread of swine flu around the US:



Google Flu Trends works by applying the Google Panopticon to searches that correlate with CDC data on actual flu cases, and has the benefit of being immediately responsive to trends in outbreaks of influenza (CDC data tends to lag by a week or two).

Says Silver:

This map is fascinating on a number of levels. Although the initial outbreak of H1N1 back in April was centered on Texas, California, New York, Illinois and South Carolina, the place where the flu first hit critical mass several months later was in Louisiana. It then slowly radiated its way outward to most of the neighboring states -- Maine finally hit the 5,000-point threshold just last week. There also appear to be other points from which the flu spread -- a less prominent 'epicenter', for instance, centered in Minnesota and the Dakotas. And somehow, there came to be quite a lot of flu at various points in both Alaska and Hawaii -- Hawaii's peak actually came way back in June and July, well before the one in the Deep South.

Here's something I don't begin to understand: everyone kept saying there'd be a second wave of swine flu in the fall, because the slu likes colder temperatures. Sure enough that second wave came to pass - but it looks like it actually erupted in one of the warmest regions of the country at the height of summer. That makes the opposite of sense to me.

Anyhoo, here's some good news, according to Nate: "the flu is pretty much on the decline in all states except Northern New England." Though if you're looking for a reason to feel glum, you should be informed that more people have died in the US from swine flu than died in the attacks of September 11, and most of them were fairly young.

Meanwhile, I see that Google is going global (or at least semi-global) with their flu map:



Bad times for cold places.

Where the Uninsured Are

NPR has a map of the uninsured by congressional district and by state.



On the face of it, it looks like the usual, albeit paradoxical, story: areas that vote more Democratic, and which support a broader social safety net, have less need of one, since fewer people are uninsured in those areas; whereas Republican-leaning areas, where support is presumably greater for the status quo (the maintaining of which seems to be the Republican approach to health care), tend to have more uninsured. Unfortunately, this map doesn't do a good job of letting you see urban congressional districts, so the appearance of the map could be rather unrepresentative of the country as a whole, and especially of Democratic-leaning areas (many of which are in cities).

However, you can also see uninsured numbers by state, which reveals that of the 26 states (counting DC as a state for wishful thinking purposes) where the uninsured are less than 15%, 21 were won by Obama in 2008. And of the 13 states where the uninsured are more than 20%, 10 were won by McCain. (McCain won 7 of the 12 15-20% states.) That's a rather striking correlation, no?

Meanwhile, Nate Silver uses math n' stuff to create a map that projects support for the public option for every congressional district:



Based on a few polls in certain states and districts, Nate created a regression analysis to project what support across every district in the US would likely be, based on a few variables, including poverty rate and Obama's vote share in the district. He found that:

-- The public option is estimated to have plurality support in 291 of the 435 Congressional Districts nationwide, or almost exactly two-thirds.
-- The public option is estimated to have plurality support in 235 of 257 Democratic-held districts.
-- The public option is estimated to have plurality support in 34 of 52 Blue Dog - held districts, and has overall popularity of 51 percent in these districts versus 39 percent opposed.

By implication, the public option was favored in 56 of the 178 Republican-held districts. Nate breaks out the projected support numbers for every district in his post.

Thanks to Matt Osborne for that one.

Fun With Epidemiology!

This game lets you respond to outbreaks of disease!



Deliver face masks, develop vaccines, and watch verite videos of indeterminately Teutonic scientists and panicky masses as you try to slow the spread of mean-looking red dots across a Risk-like map of the world. Note with equanimity the catastrophic consequences of your misallocations of resources as millions die, and consider the fundamental absurdity of a universe in which such picayune decision-making can lead to such widescale suffering and death. Fun for all ages!

The Unhealthy Behaviour Axis

A new map from Gallup and AHIP (and a continuation of their study of well-being across the states, covered here before), measures states by healthy behaviour:



Says Gallup:

The midyear results from the AHIP State and Congressional District Resource for Well-Being, a product of the Gallup-Healthways Well-Being Index, find the nation as a whole dropping substantively on the Healthy Behavior Sub-Index, from 63.7 in 2008 to 62.6 in the first half of 2009. The Healthy Behavior Sub-Index is one of six sub-indexes that make up the Gallup-Healthways Well-Being Index, and asks Americans four questions: do you smoke; did you eat healthy all day yesterday; in the last seven days, on how many days did you exercise for 30 minutes or more; and in the last seven days, on how many days did you have five or more servings of fruits and vegetables. The Healthy Behavior Sub-Index scores for the nation and for each state are calculated based on a scale from 0 to 100, where 100 would be a perfect score.

Healthy Behavior scores in most states are trending down in the first half of 2009 compared with 2008, though many have not decreased by a statistically significant degree. Mississippi, whose score ranks among the bottom 10, is the only state to record a statistically significant increase in its healthy behavior score thus far in 2009.

The healthiest states, in order, are Vermont, Hawaii, Montana, California, New Mexico, New Hanpshire, Maine, Delaware, Idaho, Wyoming, and Oregon. The least healthy is Kentucky, followed by Arkansas, West Virginia, Indiana, Ohio, Alabama, Tennessee, Oklahoma, Mississippi, Illinois, and Louisiana.

This is sort of a weird map. On the one hand, there is a very clear nexus of unhealthy states - all of the 'higher range' states are contiguous, in fact, with 'mid-range' states mostly forming a periphery around that core. But the weird thing is that the group of unhelthy states, despite its contiguity, transcends just about every other cultural and geographical distinction youcould try to make: North/South; warm-weather/cold-weather; urban/rural; manufacturing/service/agricultural economy; liberal/conservative; Obama/McCain; large/small minority population... If you break down these states by any intuitive metric, they seem to form no pattern at all, yet they create as tight a spatial clustering as you'll find on any map of the states. Is it a coincidence, or is there some hidden variable here that would explain the pattern?

The map does vaguely remind me of the personality type maps from Richard Florida. In particular, there are a few personality traits which seem to notably predominate both in the South and in the Midwest, in roughly the same areas as the "unhealthy behaviour" states in the map above: people in those regions tends to be extroverted, conscientious, and not very open to experience. Do those traits correlate with smoking, eating junk food, and not exercising? Don't see any reason why they should, but who knows.

Via M. Yglesias.

The Tight Fist of the Invisible Hand: The 'Free Market' for Health Insurance in the US

The Center for American Progress has an interactive map showing that the 'free market' for health care in the US, um, blows:



You can click on states for details; for instance, in Illinois, 69% are insured by the top two companies, Blue Cross Blue Shield (47%) and WellPoint (22%). Says CAP:

Today many Americans have few choices when it comes to health insurance. This is because many insurance markets are dominated by only a handful of firms, even though there are over 1,000 private health insurance carriers in the United States. This concentration limits employers’ and families’ health insurance options as well as the care they receive.

In many states small insurers compete against one another in the individual market to insure only low-risk, healthy individuals. They refuse to insure Americans with pre-existing conditions [ed.: like me!] such as high blood pressure, asthma, cancer, or diabetes and those who have ever taken certain prescription drugs—and they create barriers to needed care for those who are insured.

The map shows that in many states insurance markets are dominated by only one or two insurance carriers. In at least 21 states, one carrier controls more than half the market. More than half of the market is controlled by two carriers in at least 39 states. In 2007, a survey conducted by the American Medical Association found that in more than 95 percent of insurance markets, a single commercial carrier controlled at least 30 percent of the insurance market.

Fortunately, though, health insurance companies are entirely benign institutions that seek to promote the common good, and would never think of seeking to profit exorbitantly off of the health needs and suffering of the people who support their businesses.

Oh, wait:

Where markets are dominated by only a few firms, health insurers revenues are growing faster than health inflation as insurers maximize rates they charge employers and families and create barriers to care.

Listen, free marketeers: the insight that competition breeds innovation is wonderful - but it is not the end of economic analysis. It seems ridiculous to have to point this out, but it's not the case that efficiency and quality will be maximized for every single conceivable good by leaving it to the whims of the marketplace. National defense is not like that, education is not like that, and in a sane world - or in Europe - it would be manifestly obvious that health care is not like that.

Some More Stuff From the American Human Development Project

The American Human Development Project has some other tools beyond their Human Development Index that help to make sense of the geography of development in the US, including maps that break their data down by congressional district (pdf). Here's a map of overall human development by CD:



And here's their map for life expectancy:



There are also maps for income and education level. And, too, they have something called a Common Good Forecaster, which lets you set up conditional scenarios for counties around the country to see how changes in educational attainment would affect things like like expectancy, median income, and community involvement. For instance, if everyone in Cook County Illinois, where Chicago is located, got a high school degree, the average income would rise from $36,171 to $37,364, and the murder rate would drop from 12.5 per 100,000 to 11.3.

And then there is the well-o-meter. In the past I've expressed doubt about the idea that you could measure "human development" at the level of the individual, as the concept of development would seem to be the sort of thing that pertains to a community or society as a whole. But the AHDP more or less lets you calculate just such a measure ("your own human development level"). Answer a series of questions, like "Have all four of your grandparents lived past the age of 80?" and "Are you happy?"; then they spit out a number. I got a 5.69 - slightly below Maryland and well below Asian males, but slightly better than the Midwest.



Now if only I could raise taxes to fund my health care and education, I could really get my human development on...

Swine Flu: It Still Exists!

Remember how last week we were all going to die because of swine flu? And but now it looks like it just gives you a case of the sniffles? The media does seem to have a difficult time calibrating its coverage of these sorts of issues, doesn't it? Well, fortunately there are people still following the spread of the flu who actually know what they're talking about - people like Dr. Henry Niman, who made this google map of the flu and who has now set up shop with a new and fancier map.



It has numbers of suspected, confirmed and fatal cases for every country, and the data seems to be a bit ahead of other sources. E.g., who knew New Zealand had 174 cases? You can zoom in to see the spread of the virus within countries, too, like yea:



There are other maps as well, like this one of the spread of swine flu in the US by county. Blue counties have had 1-5 cases; greens have had 6-15; yellows have had 16-40; and reds have had more than 40:



Hmm. It seems to be much more widespread in the US now than it is in Mexico. Maybe we ought to close the border and keep the dirty Americans from infecting their neighbors to the south.

Mood Map of the United States

Via the Southern Political Report, this map, based on data from a telephone survey performed by the Centers for Disease Control, shows percentages of residents who reported more than 14 days of emotional discomfort, including "stress, depression and problems with emotion" over the prior month.



Appalachia has problems. Kentucky is the saddest state, with 14.4% reporting extended periods of mental discomfort, and a depression belt stretches right through the entire Upland South region from West Virginia to Oklahoma. This corresponds to a region with some of the highest rates of neuroticism, as well as some of the lowest rates of overall well-being.

At the other end of the spectrum, the most content states are Wisconsin, Iowa, Nebraska, and - happiest of all, for utterly inexplicable reasons - Hawaii. All of those states had fewer than 8% report sustained mental distress.

What strikes me about this map is that urban and suburban counties generally seem to rate consistently toward the middle of the mood spectrum throughout different regions of the country (albeit with a few exceptions, like Los Angeles, Detroit and Tampa). But rural areas vary dramatically - the happiest counties are rural areas of the Upper Midwest, and the saddest counties are, again, rural areas of the Upland South. So what accounts for the discrepancy? Well, broadly speaking, National Geographic says this:

Previous studies have linked regional income and education levels to well-being. And in general, people with higher incomes and college degrees report fewer instances of prolonged depression or stress, said study author Matthew Zack, a medical epidemiologist with the CDC.

But that's probably not the whole story, he added.

For example, communities with low [frequent mental distress] levels may have above-average support structures for residents—subsidized health clinics, for example, or job-retraining programs.

"There may be different influences in different communities," Zack said. "Once we find out what the most important ones are, we may be able to develop programs to reduce the levels of mental distress."

So are such support structures much more prevalent in places like the rural Midwest than in Appalachia? And if so, why? And what accounts for some of the intra-regional differences, like the fact that people in the Appalachian counties of Tennessee and North Carolina seem to be happier than their counterparts in Kentucky and Virginia? And why is it that Oklahoma is not just very sad, but appears to taint neighboring regions in other states? Lots of intriguing questions here.

This study will be published in the June 2009 issue of the American Journal of Preventive Medicine.

More Swine Flu Maps

Search Engine Land has links to some more maps of the porcine influenza. You know how google has a flu map for the US based on web search activity? Well, they've added an experimental flu map for Mexico, based on the same principles.



A few bits seem to be missing in this map of Mexico, but you can make out some trends: Oaxaca, Morelos, the Distrito Federal, and Quintana Roo (the best-named Mexican state) are showing the most "flu activity," as per Google's algorithms. One wonders, though, if their method holds up when there's so much media attention on the flu: couldn't all the news stories influence rates of Google searches?

Meanwhile, Tech Crunch looks at searches for "swine flu" across the 50 states. It's not as precise as Google Flu. As Tech Crunch notes, "this method is less likely to be predictive of the actual spread of the disease because it just measures raw searches." So think if it as one interesting data point, and nothing to hang your epidemiological hat on; and what that one data point shows is that the top ten states for swine flu searches are: Texas, Indiana, New York, Vermont, New Mexico, Kansas, Illinois, Ohio, Arizona, and California. Again, this goes into the "for what it's worth category;" but it's interesting that several of those states have already had reported or suspected cases. That's probably just a reflection of media attention on reported cases, but if, say, Vermont and New Mexico pop up next in CDC reports, this map will have been prescient.

HealthMap is tracking swine flu by posting news reports and even some anecdotal reports from around the world.



The redder the marker, the more serious the news: "Swine Flu: Zambia Takes Steps" is a light yellow, "Fort Worth Schools Close for Swine Flu" is a rather more ominous crimson. (HealthMap also follows news of outbreaks of all sorts.)

Another map shows reported cases as well as the travel paths of infected individuals, according to news reports. And this heat map, at UMapper, gives a good picture of areas where the virus is currently spreading, with higher concentrations in white, shading towards purple at the peripheries.



Of course, if you like your swine flu maps to have the imprimatur of big media officialdom, you can check out the New York Times' characteristically high-quality maps of the outbreak in both the States and throughout the world.



And the BBC's map has a slider that lets you see the spread of the disease in pseudo-animation. Meanwhile, remember that google map of swine flu I posted a couple of days ago? Turns out it was put together by Dr. Henry Niman, an expert on viruses at the University of Pittsburgh. So that's good! Niman talks about his map in a video interview here.

Well, I guess I've gone and contributed, in my tiny way, to the media frenzy over swine flu. But I couldn't help it - this thing is just so eminently mappable.

Swine Flu: The Inevitable Google Map

It was only a matter of time before this showed up.



A Google map of the spread of the global health scare du jour. Pink markers are suspected cases; purple markers are confirmed or probable; deaths have no dot in the middle of the marker; yellow markers are negative cases. Click on markers for copious details on individual cases.

That someone would make such a map may have been inevitable, but nonetheless: note how cool it is. The map format, combined with (theoretically) near-time updates on reported cases and free 'n easy public access make this an ideal medium for monitoring the spread of an epidemic like swine flu.

Mapping the Conficker Worm

Via PCMag's Security Watch blog, the Conficker Working Group has infection maps of the Conficker worm, which is some sort of centipedal beast that feasts on the souls of computers or something.



The map is based on infection rates as of April 1st. This caveat is given:

While the maps appear very detailed, the mapping process itself is somewhat inaccurate. Each area of color is a spot that must be placed and then a range of color applied based off the density of data in that spot. So, the greater the scale of the map there is an increased bleed effect of the applied dots and distribution. Areas can appear far more infected than they are in actuality. So we present the maps as something to see within those limitations and built-in levels of inaccuracy.

Still, some patterns are evident. As the Security Watch blog post notes, the infection rate seems to be much denser in Italy than in France.



The blog also gives rates of infection for some countries; the winner is Vietnam, with a rate above 13%, followed by Brazil and the Phillipines at 11%. The US is at 4.7%, and Italy has the highest infection rate in Europe at 3.6%. (It's not clear to me, though, whether these numbers represent the percent of all infected systems that reside in a given country, or the percent of systems in the given country that are infected.)

Google Flu

No, it's not a mad scheme to embed the blueprint of Google's corporate ambitions in the genome of every living human (though don't think they haven't thought of it). It is, rather, an effort to use Google's alarmingly Everestian stockpile of data to track the spread of the flu bug.




Says Google:

We've found that certain search terms are good indicators of flu activity. Google Flu Trends uses aggregated Google search data to estimate flu activity in your state up to two weeks faster than traditional flu surveillance systems.

We have found a close relationship between how many people search for flu-related topics and how many people actually have flu symptoms. Of course, not every person who searches for "flu" is actually sick, but a pattern emerges when all the flu-related search queries from each state and region are added together.

During the 2007-2008 flu season, an early version of Google Flu Trends was used to share results each week with the Epidemiology and Prevention Branch of the Influenza Division at CDC. Across each of the nine surveillance regions of the United States, we were able to accurately estimate current flu levels one to two weeks faster than published CDC reports.

So by using the use of search terms as a proxy for people's deepest wishes and concerns, and by extension as a proxy for their physical state, Google is providing a public service - letting us know where and when the flu is spreading. Faster than the Centers for Disease Control itself. Nothing nefarious about that.

Right?