Parts | Days | Selection | Search | Updates | Downloads | Help

T: Fachverband Teilchenphysik

T 78: Experimentelle Methoden I

T 78.3: Talk

Monday, March 28, 2011, 17:15–17:30, 30.23: 2-1

p-values for Model Evaluation — •Frederik Beaujean¹, Allen Caldwell¹, Daniel Kollár², and Kevin Kröninger³ — ¹Max Planck Institut für Physik, München — ²CERN, Genf — ³Georg-August-Universität, Göttingen

In the analysis of experimental results it is often necessary to pass a judgment on the validity of a model as a representation of the data. A quantitative procedure to decide whether a model provides a good description of data is often based on a specific test statistic and a p-value summarizing both the data and the statistic’s sampling distribution.

Although there is considerable confusion concerning the meaning of p-values, leading to their misuse, they are nevertheless of practical importance in common data analysis tasks. We motivate the application of p-values using a Bayesian argumentation. We then describe commonly and less commonly known test statistics and how they are used to define p-values. The distribution of these are then extracted for examples modeled on typical new physics searches in high energy physics. We comment on their usefulness for determining goodness-of-fit and highlight some common pitfalls.