Friday, June 12, 2009

Econometrics

Josh - I've been cleaning out my inbox and found this e-mail from you. I kept it because I like the way you frame the argument so I'll preserve it here for posterity.

Under the current norm it is not in the interest of authors to send their data sets to strangers who ask for them. Their work has been verified under peer review, and published, which means that their arguments are sound and the data supports their conclusions. An additional reviewer can add no marginal credibility to the work, but can tear it apart if he discovers different results in the process of replication.

However, if that norm is adjusted to where Levy believes the field is moving, that eventually only replicable work will be trusted, then it will be in the interest of authors to provide their data sets to others. Failure to do so will bring the author's credibility into question.

Compared with the totality of knowledge which is continually utilized in the evolution of a dynamic civilization, the difference between the knowledge that the wisest and that which the most ignorant individual can deliberately employ is comparatively insignificant. ~Fredrich Hayek in The Constitution of Liberty