Assorted Links

by Chris Auld

  1. My colleague David Giles points out that the usual test statistics in linear regression follow textbook distributions assuming normal errors when the errors are not normal but come from the class of elliptically symmetric distributions (which includes Student and logistic). I didn’t know that!
  2. Has everyone noticed that Stata 12 has been released?
  3. A failed attempt to mock regression analysis in the social sciences.
  4. Rejection letters from journal editors are somewhat more polite, but I’d prefer this.