Sunday, March 9, 2008

YAICHS: Yet Another Income's Correlation to Happiness Study.

Worldwide, Residents of Richer Nations More Satisfied

The above links to a yet another study which explores possible connection between wealth and happiness - by comparing various countries. Like most studies, the statistical analysis is completely flawed and conclusions blatantly wrong - I will address this in a second. The reason I think it is worth seeing is because their entire study is cutely summarized in a single picture:



To anyone educated in statistics this will look like a plot showing strong correlation between variables. However, after looking closely at this plot for a minute or two, I reach a different conclusion. This plot, in fact, appears to be composed mostly of two distinct parts.
First part is countries where incomes are more or less zero. Not surprisingly, happiness is low there - people are dying in wars and starving. But even there, there is a large spread in reported happiness, for basically the same level of income. I would bet this is due to presence or absence of war in a respective country.
Second part of the plot is all the countries where incomes are non-zero, above 2000$ per year or so. There, the variations in reported happiness are so huge, any correlation is nearly meaningless. But if you look even closer, you will see that almost every country that is at the lower end of the happiness spectrum, regardless of income, is either a country where there is a war, social injustice, or despotic oppressive regime. So GDP per capita may look okay, but if a large fraction of population is either left to starve, or is being tortured and imprisoned, or lives in a war zone, you would expect happiness to decline.

So, as is so typical of social studies, the results are misunderstood in this one. And as usual, the pseudo-scientists forgot to estimate what the uncertainty is and what is the statistical significance of their conclusions (it is very small). While the study shows correlation of happiness and income, it doesn't interpret it correctly. To me, it looks like the study demonstrates that happiness is negatively affected when people's right to live is endangered, either through war, or economical or political persecution. That GDP is also correlated with personal freedom and peace, is a secondary affect here. Once someone has bread, warmth and freedom, all correlation between income and happiness almost disappears. The reason we see correlation in the plot is because countries where people are more equal, more free and more peaceful, are usually way more productive.

So what we have here is yet another 'scientific' study that made a classical mistake of confusing correlation with causation. The real cause of rising happiness is less threatening situation in a given country, which correlates with GDP per capita.

No comments:

Post a Comment