* Correlation is not causation.
* Poor people may often remain poor because they are not rational. Charity may not be a rational activity. So perhaps what they were measuring was not selfishness, but rationality.
* Being born in a rich family is correlated with a better education, so the point made in the article's last paragraph may be incorrect.
* I'm (currently) poor, but I don't think much of most organised charitable activities, or to giving money to homeless people. I think most charities are an utter waste, and giving money to homeless people only gets them buying alcohol and cigarettes. I'll happily give my sandwich to a homeless person on the street though. My suggested figure for "% of salary to donate to charity" is 0%, and will remain 0% for the rest of my life. However, once rich (assuming I get there), I'd be quite happy to devote years of my life to helping people directly (which may involve leveraging financial resources, but not donating money to charities).
* Average figures tell us little. What was the distribution? Did 90% of rich people suggest 0% and 10% of them suggested 50%? How did they get to the 2.1% figure?
* What kind of social pressure was there when asking people how much they would donate (of their credits, or of their salary)? Social pressure is a huge influence. Here on HN, an open-minded, thoughtful environment, I explain my views on charities above (even though I know I'll get some passionate replies). Sitting in a circle of people I don't know, I might still express these views (because I'm like that..) but most people probably would be wiser and keep those views quiet.
* What kind of social pressure was there before? What kind of social pressure did they expect after? Did people give an answer to please the examiner?
A lot of holes in this article, I think. Surprising, coming from a quality publication like the Economist.