As regular readers of this blog may know I am critical of the cult of statistical significance in the human sciences. In the absence of solidly established background theory, it has allowed data-mining to create the illusion of "breakthrough results." A terrific article in the New Yorker highlights a lot of important issues in how science works in practice. Most notably, the article shows the signifiance (sic!) of journals' lack interest in lack of replication. The article focuses on psychology and biomedical science, but the problem shows up in economics, too. This piece by Robert Goldfarb is also thought provoking.
Recent Comments