It's sort of like the weather, you can make some good SWAGs (Scientific Wild Arse Guess) but your margin for error curtails accuracy and long term prediciton isn't possible, except in very broad sweeps which will occassionally be wrong anyway. I would argue, that as most people are reasonable, the simple existence of multiple major political factions and idealogies is proof that you can't compare these things very well. You can show strong indicators, but if there is still strong debate on something it's because you can't prove it.



By any means that produces sufficient accuracy to legitimately talk about differences, if that does not exist at this time, then you don't do it. I can want to know chemical composition of the moon's surface is, but that doesn't mean that if all I have for data is is it's mass and distance I can't twiddle with those numbers and get a result, then justify it's validity by pointing out there was no other method around. And this isn't a ludicrous example, knowing those two things, Earth's own makeup, and the general laws of physics and chemistry I could produce a pretty good guess, but probably not enough to justify selling it to an aluminum mining company who was thinking of buying a different moon or asteroid instead. Weird example, but it fits the model.



Well, there's an old quip about how you'll know you've found the right theory when the math is smooth and elegant... unfortunetly there are dozens of systems where that simply isn't true, like the weather, or ecology, or humans behavior. I don't expect it to be perfect, I just expect people not to try to use it for things it's insufficient to discuss. As another absurd example, the HDI is not well suited to predict what type of vehicle will be purchased this year, but could predict within a reasonable margin, what the average amount of money spent on vehicles would be, of course in that case, using the HDI is redundant where PPP or GDPpc will probably give a truer value, because while SOL might range fairly close to HDI, car purchases are not likely to be heavily weighted by EI or LE on anything approaching the degree of money.



I'm leery of saying we know they are accurate to 1%, just because the methods of gatheirng the data are not standard nor are they techincally gathering the same data on EI - comparing different degrees, different languages, etc. When I said earlier that these are tricky to compare that was sort of Isaac-speak for 'impossible, futile, unlikely to be comparable any meaningful and accurate way'. Also, I'd say 2.7% is a pretty big inaccuracy for comparing the various western nations since they all scored within that range of each other, also while margin for error is roughly bell-curvey, with most samples likely to hold near the middle, the whole point is that it is only accurate to that range, and really the fringe values are still highly likely. Remember, the margin isn't 2.7 in a case like that, but plus or minus 2.7, meaning 5.4, and inside that the nature of guassian curve sis that most results will lie inside a few points in the middle, which still exceeds the size of the 'western bracket' if I may coin a term. I'd be willing to bet if you averaged all the top countries, then plotted all 30 or so of them, say one dot for each country and each year for the last decade, you'd probably find yourself staring at bell curve. Which would indicate what? That the measurements to that level of accuracy are basically random on account of the error.

That's always a good way to check if a measurement really is accurate to a certain level, plot a chunk and see if it's distribution follows that when it shouldn't if accurate. If you plot a bunch of LE's in a high bracket, and there distribution isn't follwoing the obvious (slow upward trend from year to year) but start conforming to a bell curve or flat random walk, then you're trying to squeeze more accuracy out of something than is actually there to be had. I'd want to run the numbers for say the top 25 countries for the last decade but I do have a pretty good eyeball for these things and it looks like a hazy gaussian curve will result.



Well, ironically I think when one decides to use these factors, instead of weighting PPP/GDPpc as more, you might get more accurate results using less. It is (as it should be) a log (though the coefficient is definetly debatable and additional terms are likely warranted) but because life expectancy and education are so tied to wealth, you might not even need it in there except as a small correction factor. Because I'd bet that earnings vs education or LE probably follow a log curve themselves and likely you can get a more accurate curve that way. Sort of like our atmosphere and gravity thing, escape velocity becomes more important that total mass for retention, so short of just using cash per capita by itself, education and LE take on the forms of escape velocity, likely making inclusion of wealth a bit redundant as a primary factor.



Well, another source of error, we know genetics plays a pretty sizaeable role in LE, China for instance probably will have a skewed LE down the road because of the higher percentage of males who live shorter and do so by a pretty notewrothy percent of the avg LE. Since genes (race) is a fairly noteable factor in LE (or so I'm told) and those are hardly even vaguely evenly distributed even in the US let alone anywhere else, you get another noteworthy skew where someone might score higher, and whilke that higher score is legit, it doesn't really show up as a relevant justification for political policies, except maybe the ethnic cleansing and genocide sort.



I'm a big believer in all major policy and expenditure being subject to analysis to find the loose curve of diminsihing returns. Obviously, spending 300 million to vaccinate the country and prevent an estimated 1 million deaths is a good deal. SPending 3 billion to save 10,000 lives? That's trickier, and the justification is gonna be humanitarian and a quick 'well, the research values involved will give us additional hidden returns too' spend 30 billion to save 10 lives? Not a chance, not even if it were 10 little kiddies and I knew that 30 billion was the exact amount of money to be spent buying X-boxes. Anyway, that's what luxury taxes are for, in the capitalist model you can encourage chairity and discourage frivolous purchases by making the one deductible and the other a bit more expensivee, just a bit though, punitive luxury taxes can hurt more than help when too high.

I think an example is education, contrary to what I may have implied, I'm a big believer in it, but I don't like this emphasis on ever shrinking classes. Half the size of a class, double your cost, and you don't double your results. Some subjects, you don't get virtually any improvementm whereas a math class, smaller tends to be much better dollar for dollar than say, a history lecture. These are pretty standardized and should be subject to meaningful and fairly accurate comparison. You should be able to extract the ideal class size for a given course, then the school, with their relatively small numbers, can make a choice (with handy advice from above compiled by statistcians) about what their cut off size should be to break classes into an additional section. If 27 turns out to be the ideal HS bio class size, and they have 115 students needing to take it, then they can decide to have 4 or 5 classes based on their circumstances, i.e. Mr. Johnson is a real good bio teacher, we'll slot him for 32 students, and give the others 28 each. Or their teachers are maybe a bit below the norm, so they make 5 classes of 23 instead, etc...

I tend to feel all gov't spending should be subject to that sort of diminishing return analysis and come budget time congress shouldn't be weighing in on funding much more than to say 'well, we've got X and Y, each asking for Z and we have 2Z, looks like we'll get a a bit better return from X, but there's a 3% margin of error, so while 1.01 Z and .99Z seem the ideal funding levels, let's go with 1.02 and .98 instead' rephrased in politcal bickering speak of course. Not to imply putting a limit on their powers, but to pressure them to conform to that, because that gives us best results. Nothing the government does is frivilous (or is supposed to be) so every thing on the budget is important and we should be comparing as best we can, maybe we will get better results moving 10% funding from military to schools, maybe we'd be better shifting 10% of higher ed funding to military in the form of GI Bills. Maybe too hevay a luxury tax on Xboxes and such increases stress and lowers LE, all these things, in countries with hundreds of millions of people, should be subjectable to analysis - but not by social scientists, it may be a real science, but I don't consider most of it's practioners to be real scientists, most are mathemetically inept and many have a bad habit of pushing their own desires into their results, way more than other fields. You'd be better off nationalizing healthcare, eliminating medical insurance, and then re-employing all the actuaries to study it. Ironically that's probably the best way to actually reduce government spending by annexing private business.