J. Berger
Issues of multiplicity in testing are increasingly being encountered in a wide range of disciplines, as the growing complexity of data allows for consideration of a multitude of possible hypotheses (e.g., does gene xyz affect condition abc); failure to properly adjust for multiplicities is a major cause of the increasing lack of reproducibility in science. Bayesian adjustment for multiplicity is enormously powerful, in that it occurs through the prior probabilities assigned to models/hypotheses. It is, hence, independent of the error structure of the data, the main obstacle to adjustment for multiplicity in classical statistics.
Not all assignments of prior probabilities adjust for multiplicity, and assignments in huge model spaces typically require a mix of subjective assignment and appropriate hierarchical modeling. These issues will be reviewed through a variety of examples.
Palabras clave / Keywords: multiple testing, reproducibility, prior probabilities
Programado
Sesión GT08-1b Inferencia Bayesiana-1 (Parte 2): New Insights on the Role of the Bayesian Thinking in Model Selection Problems (BAYES-1b). Organizador: Gonzalo García-Donato
29 de mayo de 2018 12:20
Sala 3
