The student workshop brings two eminent statisticians to discuss topics to broaden your horizons and show you new methods to deal with statistical problems you might come across in your research or career.  The workshop will include coffee breaks as well as a lunch which are included in the workshop registration fee.

For information on the speakers see the Workshop Speakers page.  If you have any questions, don’t hesitate to contact us.  To register and pay for the workshop please go to the Registration and Payment page.

Abstracts for the student workshops are below.

Conditionally Specified Distributions

Barry Arnold

In efforts to visualize bivariate densities, it is often helpful to introspect  on the corresponding conditional densities, which are essentially cross sections of the joint density. This observation leads to the consideration of joint densities that are specified in terms of their conditional densities, so called conditionally specified distributions. Questions of compatibility of families of possible conditional densities are thus of interest. Conditions for uniqueness of conditional specifications must be addressed.

Identification of families of joint densities with conditionals in specified parametric families provides potentially useful extensions of well-known bivariate models. Several examples will be provided. Conditional specification is particularly natural in the Bayesian context of eliciting suitable prior densities in multiparameter models. Conditionally conjugate priors yield posterior densities that are tailor-made for Gibbs sampler simulations. Some discussion of inference for conditionally specified models will be provided. Natural multivariate extensions of conditional specification concepts will also be presented.

Generalized Order Statistics

Udo Kamps

Order statistics and record values appear in many statistical applications and are widely used in statistical modeling and inference. Both models describe random variables arranged in ascending order of magnitude. Generalized order statistics provide a unified approach to a variety of models of ordered random variables with different interpretations, such as common order statistics, sequential order statistics, progressively type II censored order statistics, record values, k-record values and Pfeifer-records. These models can be effectively applied, for example, in reliability theory to model influences of failed components on the remaining system. Well known structural properties of order statistics and record values turn out to be valid for generalized order statistics, too.

A survey of their distribution theory is given, and, for example, results on relations and bounds for moments, structural properties, characterizations of distributions, preservation of aging properties and stochastic orderings as well as results on extreme value theory are shown. In particular, statistical inference based on generalized and sequential order statistics is addressed. Recent and future research directions are discussed.