People find this interesting correlation between religion and poverty. Results show that the poorer the people are, the more likely that they think religion is an important issue.
Apart from the finding, I am also quite interested at how multi-dimensions is used. In this 2D graph, with 2 axises (contributing to the POSITION variable), it is fair enough just to indicate the religion dimension and the poverty dimension. However, utilizing the SIZE and COLOR variable, the graph also shows the information of different religions and the population of each country.
However, there are two problems remained. 1) The use of COLOR is biased. The colors for ‘Muslim’ and ‘Others’ are clearly different from the colors for ‘Catholic’ and ‘Other Christian’, which results in a biased view of the graph. 2) The use of SIZE obscures part of the graph, making some countries not viewable. For example, the countries on the upper left are barely recognizable.
There is a point from Tufte’s Graph Theory: The number of information-carrying (variable) dimensions depicted should not exceed the number of dimensions in the data. This is quite a controversial point. And by practice, it is also controversial. Shall we stick to the two dimensions or develop more like using SIZE and COLOR?