(no subject)

[theweaselking]

Holding constant levels of GDP per capita and the public health spending share of GDP, increases in private health spending are associated with higher--not lower--infant mortality:

via james_nicoll

Comments

[catblade.livejournal.com]

If you drill down, it appears that the "study" is flawed. One of the big factors in infant mortality is HIV in South Africa...

Also, R squared values should usually be above .95 to be taken seriously.

:\

[theweaselking.livejournal.com]

HIV epidemics are not unique to South Africa.

[catblade.livejournal.com]

One of the extreme points is South Africa. If that is causing part of the huge skew, it is confounding the results.

They don't have enough evidence. Go look at the scatter thoughts if you don't believe me. If I had ever turned in a Chem E assignment where I tried to prove something with less than a .95 r squared value, I would have flunked it. There is a visual trend, but it could also be all noise.

93% of all statistics are made up. ;)

[ice-hesitant.livejournal.com]

Acceptable R-squared ratings vary by discipline. Economics isn't chem. Sociology is very not chem. There's simply too much noise and non-replicatability at that level of abstraction.

[catblade.livejournal.com]

I have not only taken chemistry classes, but other classes as well and am well versed in more than the pure sciences. Every time the field requires a .95 or better for a "positive" correlation. Ask any statistician. Otherwise, it is still a guess and unproven. From what I read in this article, it appears to be impossible to effectively isolate the "private" and "public" from the noise and they include extrema points in their results, which most social papers that I have read throw out.

[glitteringlynx.livejournal.com]

That right there would explain why the USA is one of the 10 countries with the highest infant mortality rates in the world. And for the most part, infant mortality rates are how you ascertain a "developed" nation from an "undeveloped" one. Stupid thing? I learned that back in 1998-1999 and it STILL HOLDS TRUE.

[james-nicoll.livejournal.com]

USA is one of the 10 countries with the highest infant mortality rates in the world.

I believe if you check, that would be one of the highest infant mortality rates *for a developed nation*. On a global scale, it's nowhere near as bad as Sierra Leone or Angola.

http://en.wikipedia.org/wiki/List_of_countries_by_infant_mortality_rate

[glitteringlynx.livejournal.com]

See, I sseem to recall the table having included Bangladesh, but I even just checked info for the 90's and in 1995 it about 7 per 1000, so either I'm remembering incorrectly, or perhaps the chart was showing the US compared to various countries.

Anyway, good to know it's lower than I had thought.