Thursday, June 01, 2006

Gapminder: all sociological correlations for all countries

Scared by the noise? Go to the individual version of this Gapminder article, browse the articles one by one, and avoid the extraterrestrial ones. Sorry for inconvenience.

You should look at a new tool bought by Google, the Gapminder. It shows the countries as disks in a two-dimensional plane parameterized by
  • GDP per capita
  • children per woman

You can also choose the year - and see the map. Concerning the pair of quantities above, there used to be no correlation decades ago, but as you approach 2006, things converge to a rather precise negative power law (inverse proportionality, roughly speaking).

There are actually 14 additional quantities that you can choose as either of the two axes - enough room for the web sociologists to play.

Hard-working women won't make your nation rich

For example, you will find out that there is no visible correlation between GDP per capita (in international dollars) and the percentage of women in labor force, something that the feminists would like you not to know. If there is a small correlation, it is negative.

They call it polution; we call it life

On the other hand, there is an almost perfect linear (increasing) relation between the GDP per capita and the CO2 emissions per capita - something that the Kyoto supporters would like to hide from you.

Small errata

Whenever I say "linear" in this article, I mean "linear" when you draw it on the logarithmic scale, which really describes a power law relation between the actual quantities. (The script allows you to draw it in both ways.) Whenever I say "GDP", I really mean GNP. But I guess that it won't make much difference.

Some generally expected correlations

The child mortality is an almost accurate linear decreasing function of GDP per capita.

Contraceptive use amongst adult women grows with GDP per capita, but the correlation coefficient is not strong. There is no correlation between the economic growth and the GDP per capita.

The number of internet users (much like phone users) per 1000 people is an almost exactly linear increasing function of GDP per capita, and life expectancy vs. GDP per capita is similar.

The military budget as a percentage of the budget is uncorrelated with the GDP per capita. The nations with a majority of girls outside schools are poor.

Countries with roughly 1 physician per 1000 people are poor, but for those with at least 2 physicians per 1000 people, there is no correlation with the GDP per capita.

There is no correlation between the total population and GDP per capita. GDP per capita grows with the percentage of urban population but the deviation from a linear relation is large.

You can play with the remaining 100+ of possible graphs, and also draw them for different years.

No comments:

Post a Comment