First of all in this case I don't think there is a correlation between government spending and life expectancy. A graph with 3 selected countries and two bars

First of all in this case I don't think there is a correlation between government spending and life expectancy. A graph with 3 selected countries and two bars is not a correlation. It is not scientific either. I'll quote from http://en.wikipedia.org/wiki/Correlation http://en.wikipedia.org/wiki/Correlation : "establishing a correlation between two variables is not a sufficient condition to establish a causal relationship". Do you think this is wrong? You cannot infer causations simply from correlating numbers. Numbers can give you ideas and you can then build a hypothesis that you can test. I am not saying every correlation is a coincidence, I am saying that there are limits to what you can conclude based on correlations. Have you read about falsification and Karl Popper?