Logo Leibniz Universität Hannover
Logo: Institut für Risiko und Zuverlässigkeit
Logo Leibniz Universität Hannover
Logo: Institut für Risiko und Zuverlässigkeit
  • Zielgruppen
  • Suche

Gastvortrag: Prof. Frank Coolen

Nonparametric predictive inference for reproducibility of hypothesis tests, Nov. 9, 2018, 13:00, Room 116

Prof. Frank Coolen

Department of Mathematical Sciences

Durham University, UK

Friday, November 9th, 2018, 13:00 - 15:00, Room 116, Institute for Risk and Reliability, LUH

Nonparametric predictive inference for reproducibility of hypothesis tests


Nonparametric predictive inference (NPI) is a frequentist statistics method based on few assumptions, with uncertainty quantified by imprecise probabilities. After a brief introduction to NPI we will apply it to investigate reproducibility of several basic statistical hypothesis tests, considering the question whether or not the final test result wrt rejection of the null hypothesis would be the same if the test were repeated. This is an important practical issue which has received relatively little attention, and about which there is a lot of confusion. We end the presentation with a brief discussion of related research challenges.


Frank Coolen is Professor of Statistics at the Department of Mathematical Sciences at Durham University, UK. He completed MSc (`Mathematical Engineer', 1990) and PhD studies (1994) at Eindhoven University of Technology, The Netherlands. His MSc project focussed on reliability assessment for a critical heat exchanger in a chemical plant and involved the use of expert judgements. His PhD thesis is on imprecise probabilistic statistical methods for the use of expert opinions. In 1994 he joined Durham University as Lecturer in Statistics, followed by promotions to Reader (2000) and Professor (2005). He has worked on a wide variety of topics, with main contributions including the development of Nonparametric Predictive Inference (NPI) and the introduction of the survival signature for system reliability. NPI is a frequentist statistical method based on relatively few modelling assumptions, enabled by the use of imprecise probabilities. Events of interest are explicitly formulated in terms of one or more future observable random quantities. The survival signature is a summary of the system structure function which simplifies uncertainty quantification for system reliability for larger systems with multiple types of components. In addition to these ongoing research topics, current topics of his research interests include development of imprecise statistical methods for accelerated life testing scenarios and NPI methods to assess reproducibility of statistical tests. Frank Coolen is member of the editorial boards of several international journals, including Journal of Risk and Reliability and Journal of Statistical Theory and Practice.