One of my PHD mates suggested me to read the entire number of the Journal of Economic Perspectives. This post just to thank Giuseppe and to recommend to everybody a very interesting reading. I was literally fascinated by the Prof. Edward E. Leamer article "Tantalus on the Road of Asymptopia", a really farsighted and entertaining paper, flawless from the econometric point of view. Just one of the funny parts:

... We have been taught that experimental randomization of the treatment eliminates the requirement to include additive controls in the equation because the correlation between the controls and the treatment is zero by design and regression estimates with or without the controls are unbiased, indeed identical. That’s true in Asymptopia, but it’s not true here in the Land of the Finite Sample where correlation is an everpresent fact of life and where issues of sensitivity of conclusions to assumptions can arise even with randomized treatments if the correlations between the randomized treatment and the additive confounders, by chance, are high enough...

The whole number can be downloaded for free by clicking here

Print Friendly, PDF & Email