Title: On Stein operators in testing the fit to parametric families of distributions
Authors: Bruno Ebner - Karlsruhe Institute of Technology (Germany) [presenting]
Abstract: Characterisations of families of distribution are widely used as a basis for proposing tests of fit to parametric families of distributions. These families include classical problems like testing normality, exponentiality or Poissonity. We review the utility of Stein operators as well as Stein type characterising identities in goodness-of-fit testing and propose new ways to approach the problem. We show that these classes of tests are theoretically well understood and present themselves as serious competitors to other existing tests.