Abstract
The statistical models used to derive the results of experimental analyses are of incredible scientific value and are essential information for analysis preservation and reuse. In this paper, we make the scientific case for systematically publishing the full statistical models and discuss the technical developments that make this practical. By means of a variety of physics cases - including parton distribution functions, Higgs boson measurements, effective field theory interpretations, direct searches for new physics, heavy flavor physics, direct dark matter detection, world averages, and beyond the Standard Model global fits - we illustrate how detailed information on the statistical modelling can enhance the short- and long-term impact of experimental results.
Highlights
In 2000, Fred James and Louis Lyons convened an unusual meeting at CERN [1] between physicists and a few statisticians to discuss the somewhat dry topic of confidence limits
The discussion has been continued within the Reinterpretation Forum [8], with the current status and updated recommendations presented in Ref. [6]. This paper takes these decade-long efforts to what we argue is the logical conclusion: if we wish to maximize the scientific impact of particle physics experiments, decades into the future, we should make the publication of full statistical models, together with the data to convert them into likelihood functions, standard practice
This sets the stage for the main task of this paper, which is to give physics cases that highlight the potential for substantial scientific progress if full statistical models were publicly available and to indicate where progress is currently impeded because they are not yet routinely published
Summary
In 2000, Fred James and Louis Lyons (with considerable help from Yves Perrin) convened an unusual meeting at CERN [1] between physicists and a few statisticians to discuss the somewhat dry topic of confidence limits. The section includes a discussion of what statistical information and products should be published, and serves as a high-level introduction to Section 3, which discusses technical considerations pertaining to the publication of statistical models as well as descriptions of existing statistical infrastructure and tools This sets the stage for the main task of this paper, which is to give physics cases that highlight the potential for substantial scientific progress if full statistical models were publicly available and to indicate where progress is currently impeded because they are not yet routinely published.
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call