The Seductiveness of Hans Rosling
Hans Rosling is an extremely fine presenter of data. His visualizations using Gapminder are excellent and very effective – sometimes perhaps seductive.
In his TED talk “The best stats you have ever seen” (2006) he shows a visualization of the percentage of the world population as a function of income per person per day. He maintains that the income gap has been decreasing and is disappearing. This depends on his definition of gap. If he means the dip/relative minimum in the curve he is right. But if gap means income inequality between the poor and the rich then he not right. In fact income inequality has been increasing in recent years.
Hans Rosling exhorts all of us to use the enormous amount of data that exists for the benefit of all. He says:
“We need really to see them. We need to get them into graphic formats, where you can instantly understand them. Now, statisticians don’t like it, because they say that this will not show the reality; we have to have statistical, analytical methods.”
When Rosling says “instantly understand” I take him to mean “intuitively understand”. He is on the verge of seducing us into accepting that the relationship/correlation between the variables he visualizes implies causation.
But then he seems to feel uncomfortable with this and says:
“Many people say data is bad. There is an uncertainty margin, but…. the differences (in the data I use) are much bigger than the weakness of the data.”
This is of course an application of statistical thinking and he finally escapes by the skin of his teeth from giving the impression that he thinks that correlation implies causation by saying:
“But this is hypothesis-generating.”
The visualizations that can be made with Gapminder are extremely fine and if you are not on your guard you can easily be seduced by them. The same applies to the equally fine visualizations made with Tableau.