Alex Hoffman pointed me to this widely-circulated map comparing the fifty states in something called the Human Development Index:
As Alex points out, the coding of the map is kind of goofy: the states with the three lowest values are Louisiana at .801, West Virginia at .800, and Mississippi at .799, but their color scheme makes Mississippi stand out as the only yellow state in a sea of green.
But I'm concerned about more than that. Is Alaska really so developed as all that? And whassup with D.C., which, according to the table, is #4, behind only Connecticut, Massachusetts, and New Jersey? I know about gentrification and all that, but can D.C. really be #4 on any Human Development Index worth its name?
Time to look behind the numbers.
The report gives the following information:
The HDI combines three basic dimensions:
* Life expectancy at birth, as an index of population health and longevity
* Knowledge and education, as measured by the adult literacy rate (with two-thirds weighting) and the combined primary, secondary, and tertiary gross enrollment ratio (with one-third weighting).
* Standard of living, as measured by the natural logarithm of gross domestic product (GDP) per capita at purchasing power parity (PPP) in United States dollars.
OK, I think I see what's going on. The 50 states don't vary much by life expectancy, literacy, and school enrollment. Sure, Hawaiians live a few years longer than Mississippians, and there are some differences in who stays in school, but by far the biggest differences between states, from these measures, are in GDP. The average income in Connecticut is twice that of Mississippi.
To check out the relation between HDI and income, I loaded in the tabulated HDI numbers and plotted them vs. some state income numbers (excluding D.C., unfortunately) that I happened to already have on my computer:
Interesting. The pattern in strong but nonlinear. Let's try plotting the ranks:
The pattern seems pretty clear, with most of the states falling right on the 45-degree line. The correlation between the two rankings is 86%. I'm actually surprised the correlation isn't higher--and I'm surprised the first scatterplot above is so nonlinear--but, then again, I'm using state income rather than GDP, so maybe there's something going on there.
(In response to some mathematically-inclined readers: No, the log transformation is not what's doing this, at least not if you're logging income as is stated in the report. Logging stretches out the lower end of the scale a bit but does not change the overall pattern of the plot. The income values don't have enough dynamic range for the log transformation to have much effect.)
Or maybe more is going on with those other components than I realize. If anyone's interested in following up on this, I suggest looking into South Carolina and Kentucky, which are so close in average income and so far apart on the HDI (see the top scatterplot above).
Searching around, I found the aforementioned numbers that range from 0.799 to 0.962, and another set that range from 3.58 to 6.37. The rankings of these two sets are identical; unfortunately I couldn't find the exact formula for either of these. But I did find this report which gives some formulas and also, in its Appendix B, the actual numbers used for the 50 states in preliminary calculations. These rankings do not exactly agree with the ones shown in the map--in the preliminary data, Massachusetts, not Connecticut, is #1, D.C. is in the lower half, and Alaska is second-to-last.
So I'm not completely sure what's happening here. But if you go by the maps that everybody's linking to (having appeared in Catherine Rampell's New York Times blog), you're pretty much just mapping state income and giving it a fancy transformation and a fancy new name.
P.S. To clarify a point made by several people in the discussion: I'm not commenting one way or another on the Human Development Index as used for international comparisons. What I'm criticizing is the application to the U.S. and the implication that the widely-distributed map is doing much more than ranking state incomes with a wacky color scheme.
P.P.S. More here, here, here, and here. (That last link explains, among other things, that "percent" means "divided by 100.")
45 comments
>> I'm surprised the first scatterplot above is so nonlinear
It is likely non-linear because of the Ln they are doing on the value. Even after adjusting first for PPP, the natural log of the value is likely to create that kind of curve.
I've worked with HDI data before, and calculated HDI values for some entities (largely "territories") for whom data isn't collected by UNDP.
If you'd like to know how the score is calculated, here is the report:
http://hdr.undp.org/en/media/HDR_20072008_Tech_Note_1.pdf
It should indeed be a value between 0 and 1, so I have no idea where those other numbers are coming from.
* Standard of living, as measured by the natural logarithm of gross domestic product (GDP) per capita at purchasing power parity (PPP) in United States dollars.
"I'm surprised the first scatterplot above is so nonlinear"
Why are you surprised? The report says HDI is calculated based off of the log of GDP. Did you even read your own article?
"fancy new name" - this is a neat case study in the danger of over-interpreting a model, or confusing a model with reality.
The first graph looks pretty much like a logarithmic curve to me, as surely should be expected since they're using the natural log of per capita GDP...
Three comments:
1. Of course, rank transforming two variables will turn any monotonic relationship into a linear one... (see Spearman's rho)
2. The index was developed to look at worldwide disparities, which it may do very well, as countries vary on all components, even if states within the US only vary on one. This is not a failing of the model, as one could make the argument that we are well off enough that the only room for improvement in our human development is through increased wealth.
3. The subscales are standardized to ensure that they go from 0 to 1, then the result is averaged. I assume this scaling was done relative to the world population not the us one. Standardizing by subtracting the mean and dividing by the standard deviation instead of 0-1 scaling would ensure that each component contributes equally.
Great post, Nate quality with a nice explanation and conclusion.
Ian-
Good points for the world, but doesn't your post prove Andrew's point - namely that these measures suck within the U.S. where they are being used.
HDI is not terribly well suited to analysis of well-developed countries. Most economists admit that as much as two-thirds (if not more) of the variation is explained by differences in income.
My opinion is that HDI is largely intended to provide a gradient of which developing nations are "ahead" in development. Comparing US states may make us Northerners feel good, but it's largely an exercise in income permutation.
The problem is that they are applying a statistic that is meant to compare nations around the world -- with the consequent variability that that would entail -- to regions within a single country.
It wasn't designed for that, and therefore doesn't really work well for that.
Rampell linked to a mapscroll post (http://mapscroll.blogspot.com/2009/05/human-development-index-by-state.html)that ranks countries and states, including a sequence that says “Romania, Malaysia, Montenegro, Serbia, Saint Lucia, Kentucky, Belarus, Tennesee, Oklahoma, Alabama, Macedonia, Albania.... “
Seeing the Bluegrass state in that remarkably Balkan neighborhood was kind of stunning. Is that, too, simply an effect of income?
Hey, cool - it's the map from my blog! And it's, um, getting debunked. Well, pretty cool anyways!
I can't speak to the statistics in any technical sense; but if you go to the Measure of America maps, the eight states with the very low HDI scores actually seem to be the biggest outliers when it comes to their "health index" rather than their "income index." Indeed, Tennesse is represented as having a fairly average income, and Alabama and Kentucky are only in the second lowest grouping; but all eight of those states plus South Carolina - and only those nine states - are in the lowest grouping for the health index.
Meanwhile, here's what the Measure of America's FAQ page says about their metric, which differs from the conventional HDI metric (the terms of which this map represents): "The modified American Human Development Index measures the same three basic dimensions as the standard HD Index - health, knowledge, and standard of living - but it uses different indicators to better reflect the U.S. context and to maximize use of available data. Health is measured in the modified American HD Index by life expectancy. Knowledge is measured by a combination of educational attainment and school enrollment. Standard of living is measured using median earnings. All data are from official 2005 U.S. government sources."
Now what I don't know is whether the numbers that were listed on Wikipedia, from which the map was derived, are actually listed as such somewhere in the report, i.e., whether the MoA translated their "American" HDI into the conventional form; or whether someone else took the liberty of doing so and listed them on Wikipedia as such. What clearly CAN'T be the case is that the map's numbers are just translated directly from the MoA American HDI numbers, because (as you can see from this document (pdf)) the MoA's numbers show different rations between various states' HDIs. For example, the MoA has Mississippi as a lone significant outlier, whereas the map's numbers have MS, WV, and LA all closely clustered.
What may have been the case is that someone took the raw data from the MoA and fed it into the conventional HDI algorithms. In which case, who the hell knows if it's reliable or not.
"Is Alaska really so developed as all that?"
Try to focus on objective, well-grounded analysis in lieu of arrogant generalizations in the form of sneering rhetorical questions, please. That sentence does nothing to inform; in fact, all it accomplishes is reminding me of Sean Quinn's blithering.
Way to be incredibly oversensitive and fixate on something that is a valid criticism of the way the data is presented J III.
It seems like to me the graph looks like some sort of sigmoid function. Surprised that no one picked up on that.
black people...
I love this site, so I'm a little disappointed in this post.
The previous commenters have made a couple of points for me; HDI is, of course, better suited for broad international comparisons, especially with highly developed countries. And the MoA study produces a second score for HDI that may well be better for comparing states, but just confuses the matters when comparing to other countries.
First, discussing the colour scheme, it's obvious that the map was drawn with the same colours as the Wikipedia article on HDI. The broad break for Mississippi is because it has an HDI below 0.800, the long-standing cut off for high development nations.
Secondly, the range of incomes doesn't contribute as much to the difference in HDIs as the other factors: if you take the income for top-ranked Connecticut as $38,000 and for bottom-ranked Mississippi as $18,000 (approximating, those cute plots are hard to get exact values from) and apply the transformation used to calculate the income portion of HDI (using the UN's formula), CT has a score of 0.991 and MS of 0.867, a difference of .125. The overall HDI is an average of the income, education and health scores, and the difference between these states is 0.163; therefore, the income score difference must be contributing less to the overall score than health and education.
Thirdly, the reason these HDI values correlate so highly with income is because the US is fairly internally consistent with regards to economic, health and education systems. Better educated people tend to have higher incomes and wealthier Americans have better access to health care.
Compare this with Cuba, Saudi Arabia and Malaysia, three countries with similar HDIs (between 0.855 and 0.823; in that SC-KY gap). Cuba is poor, but prioritizes educational and health outcomes. The Saudis are stinking rich, but do poorly with regards to education, specifically for women. Malaysia is industrializing in a path similar to the US. There's no correlation between income and HDI for these three countries.
Finally, to close with two wonky notes, I suspect that the data aren't directly comparable to UN data for two additional reasons. The UN data uses purchasing price parity for incomes, and I don't think this study did (not that I'd pay to find out). Also, one component of the educational scale is gross enrolment, which includes tertiary education. The US has high internal mobility for college, so all of the Deep South students studying at the Ivy League are boosting New England state scores.
I wouldn't expect the latter few points to jump out at someone who isn't an HDI nerd, but surely you could have looked at the Wikipedia article (the first Google link) to see the colour scheme, and the calculation technique for with the income data.
@ Susan Weston: Kentucky is a fair bit wealthier than Serbia, but the Balkans have high life expectancy and very high educational scores, like most former Eastern Bloc nations. Income difference is $25K-11K, but the Serbs live to 75.3 versus 71.6 and have 96.4% literacy, versus 88%.
@Kevin,
I have always hated filtering essentially continuous data through step functions. The colour scheme sucks because it jumps discontinuously at 0.8. Yes it is the same scheme as wikipedia, but it still sucks. As jpg files allow 256 separate values of red, green, and blue, one can easily devise an almost continuous colour map.
I think you mean the correlation is .86, not 86%.
He probably meant to say that the regression explains 86% of the data, mixed that up with the correlation and ended up with that.
Standard of living, as measured by the natural logarithm of gross domestic product (GDP) per capita at purchasing power parity (PPP) in United States dollars.What I'm criticizing is the application to the U.S. and the implication that the widely-distributed map is doing much more than ranking state incomes with a wacky color scheme.What's needed here is the state cost of living, to use to index GDP. That gives somewhat different state rank (Miss is still near the tail), and may or may not completely alter the HDI pattern. (NY, for instance, would be high income, high expense, high development and Miss would be the other way around.)
Missouri COL page, which uses the ACCRA Cost of Living Index ($).
max
['HTH.']
Mississippi lags behind because people who grew up there left the state (like me). The only educated people in the state are either doctors or lawyers and there are few professional opportunities for businesspeople or engineers. I don't see things changing any time soon because the state is damned conservative. People may be friendly but they are not open minded. Only 10% of white people in the whole state voted for Obama.
I'm not suprised that the first plot is so nonlinear (and, no I don't think its related to logarithms)... I would attribute it to the Law of Diminishing Returns. You can only improve the literacy rate by so much and you can only extend the human lifespan by so much. All 400 billionaires in the country could be the sole inhabitants of a theoretical state abbreviated 'BN', which you'd find to the very far right of CT on the plot, but would be marginally (if at all) higher on the HDI index.
Nate -
This rarely happens, but here, I disagree with your analysis.
Your interpretation is that it's essentially an epiphenomenon of income, because between state absolute variance is greatest for income, but I'm not convinced.
An obvious test would be to eliminate income from the calculation, and see how the states are differentiated when you do that.
Local per capita income is stongly related to local health statistics and local average education level. In my view, although there is much positive feedback, the latter two are more horse, and income is more cart. When you're healthy and educated, it's easier to be economically productive.
But the bottom line is that there's probably a network of indirect but causal relationships between income, education, and health.
This would suggest to me that, rather than arguing that one is driving the entire thing, you should be able to look at any one of the three parameters and see approximately the same pattern.
I also think it still makes sense to look at all three, even if they are correlated with each other. There are odd places where local conditions may lead to good health in the absence of good education, or the like.
(Alaska is plenty developed. Because the other 49 states dump money to Alaska. Seriously, I don't like Sara Palin either, but it's a state with a fairly decent standard of living.)
The number of things wrong with this map just boggles the imagination.
Map Scroll uses this Wikipedia map to compare US states to nations. That's an interesting excercise but it should come with a huge disclaimer. The US HDI is not at all comparable to the World HDI. For example the income component of the US HDI is based on median earnings but the income component of the World HDI is based on GDP per capita. In addition the education component of the US HDI is based on an aggregate index of educational attainement (as well as educational enrollment) but the education component of the World HDI is based on adult literacy (as well as educational enrollment).
Furthermore it's not at all clear that the scaling of the health component is compatible or that the weights for identical components are the same. It appears that Chachy wasted a lot of time comparing apples to oranges.
On a different note, where did Wikepedia get those numbers? The rankings are indeed the same as the American HDI but the scale is completely different. The American HDI runs from 3.58 to 6.37. It looks to me like someone rescaled it so that the average of the state US HDI scores is 0.95, or the to be the same as the US score in the World HDI. Just another reason why you can't trust anything posted in Wikepedia.
Also, look at the methodology for the US HDI linked to in this blog entry. It dates from 2005 and the data used is completely different from the 2008 US HDI. The 2005 US HDI uses personal income per capita instead of median earnings. And instead of an aggregate index of educational attainment it uses high school graduation rates and a measure of elementary school quality. And the health index uses mean age at death instead of average life expectancy.
In other words the 2008 US HDI isn't even comparable to the 2005 US HDI. In particular note that Alaska jumped from 49th on the 2005 US HDI to 17th on the 2008 US HDI, mostly because of the health component. Although the average life expectancy in Alaska is supposedly 79.5, the average age at death is an appalling 63.4. Other states that benefited from the change in methodology are California and Hawaii.
And in case anyone thinks that Wikepedia is the best thing since sliced bread check out this horrible entry:
http://en.wikipedia.org/wiki
/List_of_U.S._states_by_GDP_per_
capita_(nominal)
The figures listed there are not GDP per capita at all. They are per capita personal incomes. To get the actual state GDP per capitas you'll have to go to the BEA website, but it only releases real GDP per capita, not nominal. Furthermore, note the map that's posted there. It doesn't correspond to any of the data listed there.
Wikepedia is the world's leading source of missinformation.
Interesting how the wealthy states are blue and the poor states red. This phenomenon is caused by the fact that Democrats are willing to invest more in public goods, causing their living standards to rise.
Also interesting is that poor people vote more democratic than rich people. This shows you the pitfalls of using aggregate data.
Without seeing the actual data I can't be certain but for what it's worth; Aren't there a small group of states (UT, ND, AZ, VT & HI) whose HDMI significantly exceeds that predicted by average income and another group (TN. WY, NV & FL) in which HDMI is less than would be predicted by income. Since the other components of HDMI are measure of health and education, It suggests that the outliers are telling us something about quality of life
John said...
black people... What? Aren't you going to give links to dubious internet articles offering them as "proof" when in truth they actually speak towards disproving your misconceptions and revealing your thinly hidden racial prejudices?
P.S. *cough*West Virginia.*cough*
According to this paper, "it can be be shown that the component statistics of the HDI are highly correlated with one another. The paper demonstrates that an implication of this correlation is that a wide range of index weights produce indices that are statistically identical to the HDI. Indexes with only two of the three HDI components are also very highly correlated with the HDI. These results can be interpreted two ways: either the HDI is robust to a wide range of index weights, or it is largely redundant." In other words, the HDI is nearly always highly correlated with any of the individual components, although as noted by a previous poster, outliers like Saudi Arabia and Cuba do exist.
Professor Vic -
Thank you for supporting my comment above, either intentionally or unintentionally.
Mark A. Sadowsky -
I think it's more like a plums to nectarines comparison.
It's highly reasonable to compare health, education, and income statistics across US states.
The results obtained are intuitively correct.
Another thing that would be interesting to look at would be crime rates.
These results support the broad conclusion that either having better access to health care and education helps people become more economically productive, and/or that when localities are richer, they usually choose to provide better population access to health care and education.
Either way, standard right wing ideology is challenged.
As a non-statistical outlier, I may represent the illogical hordes that get more of a gestalt notion of outcomes than a provable hypothesis of relationships, but nevertheless....
That first plot, the one that gets all the static about being logarithmic or in some other way not pristine enough, makes total sense to me, as is. Having lived and worked for considerable periods in several parts of the country, I'm not at all surprised that the 'fall-off' at the left side is populated by the identifiably marginal economic areas. Most of the country is about the same (with generally minor variances in income or gross product), as represented by the essentially homogeneous 40 states or so across the top of the plot, and the remaining trailers slide precipitously off into systemic disrepute as described by their literacy and poverty rates.
Why is this such as issue? Does every line have to be straight to be useful?
Looks like a 2nd order phase transition. There is a similar looking pattern if you look at Government Spending vs. log(GDP) (and since GDP has increased linearly in log space vs. time, you get a similar "phase transition" in time, and it happens in the 1930s.)
I'm not sure what that means as a phase transition, but it looks like you reach a certain critical value of GDP, your HDI goes to nearly 1.
@harold,
My main criticism was Map Scroll's attempt to use mysteriously rescaled US HDI to compare it with World HDI data that is in fact noncomparable. The rank orderings of the US data is correct but the numbers are all wrong, and it's not clear who rescaled it or how they did it. It's just not fair to compare Mississippi to Turkey using incompatible data, paramaters and a mysterious scaling. (And it's not clear who it is unfair to, from what I've seen of Istanbul, Ataturk, and Jackson and Tupelo, MS, Turkey is leagues ahead of Mississippi).
I actually like the 2008 criteria better than the 2005 criteria. Median earnings is probably a more equitable income measurement than per capita personal income; life expectancy is probably a better current measurement of health than average age at death; and the education attainment index is probably more appropriate for the US than the world at large (where literacy is perhaps sufficient).
I have no problem with challenging right wing ideology, but I am concerned when people carelessly post maps and statistics of mysterious Wikipedian origin.
P.S. My last name ends in an "i."
P.P.S. To be truthful, what got my curiousity about the source of the numbers was the US HDI use of per capita personal income rather than GDP per capita. The very Blue state of Delaware, my home state, has long led the nation in GDP per capita by a large amount. Usually when I see things using per capita income rather than GDP per capita I get suspicious that it is the result of some right wing bias.
Please see here for a clarification of some of the confusion. The data on which the map are based are incorrect. The American Human Development Project has its own maps which present their data on a unique scale - one which is not compatible with the scale used by the UN.
Mark A. Sadowski -
Apologies for the mis-spelling. Hard to justify when the correct spelling was right there.
I had a few more things to say, but I'm running late.
There is no inherent reason why one can't compare individual states to countries using the HDI as long as one uses identical components. In fact, the United Nations Development Program, the folks who annually calculate the HDI figures across the world, actually did a similar comparison in roughly their 1995 issue of their report. They calculated the HDI for just whites and just blacks in the U.S. The white only population came in at #1 in the world while blacks in the U.S. came in at #40, roughly the same as Costa Rica and Panama. African Americans scored significantly lower than whites in all three of the primary components of the HDI.
86% = 86/100 = 43/50 = .86. They're all the same number. I realize that it's customary to express correlation as a decimal (and for this reason the use of a percentage may be confusing to some), but this criticism seems a bit punctilious.
Hardly the first map to circulate the internet that turned out to largely be GDP. Remember the purported "average IQ" map that circulated after the 2004 election?
T J Hairball -
I'm kind of glad you made that comment.
I had commented above that I don't think that this map is "just" GDP (or whatever income measure), and that if GDP were left out and valid measures of health and/or education ranked, you would get about the same results.
I made the point above that all of these factors are inter-related in causal and feedback relationships.
For example, I personally think that if a society starts by doing whatever it can to make people healthier and more educated, the healthier and more educated people will be more economically productive. Of course, it's also likely to be true that wealthier societies are even more likely again to provide health care and education. And so on.
What I had meant to comment yesterday is that the dominance of GDP is partly an artifact of scales and baselines chosen. It's true that GDP in dollars is more variable across states than education and life expectancy in total years.
However, if you measured something like education in years beyond a baseline point greater than zero, or life expectancy in years (or months) beyond a baseline like whatever the life expectancy is in the state with the lowest, you'd see stronger effects of those variables.
The one debate I see is that these factors are so closely correlated, and that weird outliers that are high in one but low in others, like Saudi Arabia or Cuba, are so rare, that some might argue that it is redundant to measure more than one of them. (That's essentially what you and Nate are saying - GDP tells the entire story.)
I actually think it makes most sense to look at them all, especially as the US itself has been an outlier (high in per capita GDP but not so great in health and education), although our outlier status may be "corrected" by a relative drop in per capita GDP the next time numbers are released.
I posted the following on Chachy's site. I thought you all might be interested:
Chachy,
Actually, on reflection, as a concept, your site rocks!
I fiddled today with the problems associated with constructing an UN HDI consistent index for the US satets today and I observed the following:
1) The income component is easy. Just take the real US GDP per capita statistics from BEA and multiply them by the US GDP deflator for 2005 (approximately 1.1306).
2) The health component is similarly easy. The CDC has the state level statistics but so does the US HDI website. They are essentially consistent (small deviations).
3) T.J. Hair is right. The education component is a bear.
a) First of all, the literacy rate. The last time the Census collected state level literacy statistics is 1970. The last time they collected it on a national level is 1979. Why? Because literacy was essentially universal by the 1970's in the United States. For the UN HDI, any nation that has literacy rates above 99% or that does not collect such stats is alloted a score of 0.99 for that component. If one looks at the 1970 state level stats you will observe that the lowest literacy rate was for Louisiana or 97.2%. Based on even the lowest rate of decreases in the rate of literacy it is clear that by 1990 all the states in the United States probably would have had a literacy rate of 99% or higher by the UN's low standards.
b) It turns out that the combined educational enrollment rate may be the thornier problem. The American HDI website lists such data but it is not consistent with the data reported in the UN HDI report. It is lower by a factor of 0.93. I suspect that the problem is not with the numerator (total enrollment) but with the numerator (population in relevant age group). In any case I rescaled the data by a factor of 1.075.
The results? Missippi scores at 0.901. Connecticut scores at 0.969. Hawaii scored the highest at 0.974.
What did I learn from this exercise? The biggest differences in HDI by the UN standards occurred because of the differences in longevity. Twenty of the US states max out on income. All perform well by the low educational standards of the UN index. In longevity however the US does not perform very well.
On reflection, what I think we need is not an index that is consistent with the UN as it is designed for comparisons between developing nations or even an index based on US standards. What I think is needed is an index that is designed to compare the US states to the EU states and the more advanced states of East Asia etc.
I would recommend GDP per hour worked for the income component (OECD) as a measure of productivity, life expectancy for the health component (widely available), and a measure of educational attainment similar to the US HDI index. This would show how well the different US states fare compared to the rest of the advanced world in terms of things that really differentiate at this level of human development attainment.
I honestly think this is a good project for someone so inclined (maybe even myself).
As much of a fan of the HDI as I am, when I decided to do a state comparison version, I didn't use it. I unfortunately don't have my analysis anymore, but I used rankings of economy, crime, environment, education and health. Basically it revealed a development disparity along the 40th parallel.
that map's representation is not necessarily bad. It's just poor decision on the author part or just lazy on his part to use the decimal's system. More divisions or choosing divisions to reflect quintiles would be an easy fix.
Are these plots using mean or median income by state? If the former, how much of the work in, say Nevada being to the right of the curve, is being done by a few casino and hotel owners at the top of the distribution?
@ Mark:
There are no 2005 and 2008 US HDIs... What happened here is that we had a background paper on our website, woith a very preliminary version of the methodology used to calculate the American HDI, and this paper was circulated and people assumed the methodology described in it was actually used.
The "2008 version" (which uses 2005 data, by the way, since it was the latest year for which mortality data was available from the CDC) is the final version, and is the methodology that was actually used. And the major changes (median earnings, life expectancy, educational attainment) were made precisely by the reasons you mentioned in your post.
For those interested in the full methodology, we will be posting it on our site (www.measureofamerica.org) in a day or so.
- Eduardo Martins, co-author of “The Measure of America: American Human Development Report 2008-2009″, and Statistics Director, American Human Development Project.
PS. Your posts on Chachy's blog are right on the mark, I just added one post there pretty much confirming the results from your exercise in trying to create a UN HDI for US states.
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店經紀,
酒店打工經紀,
制服酒店工作,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
酒店經紀,
菲
梵,
Post a Comment