The Improvement of the World Must Be Highly Contextualized!

There are so many great things about this presentation. (Taken from Prof. Rodrik's blog.)

At the end, he says that statisticians object to his project, and (for the first time in his presentation) stumbles when trying to explain why. Let me guess that the objection might be that visually convincing data patterns can be statistically insignificant, and significant relations can look like noise in a picture. But limiting people's access to data because they might misuse it is like opposing electric lighting because some people might stick their finger in the socket.

One of the development profs whose lectures I sat in on last year told us a story. She was working for a famous econometrician, and she asked him how to start analyzing some data they were working on. His answer wasn't some fancy instrumental variable or principal component procedure -- instead, he asked, "Have you graphed it yet?" Statistics give us powerful tools for understanding data once you know what you're looking for, but unless you're a semi-autistic genius, it's hard to look at a table of numbers and find the right question to ask.

Free the world's data sets now!