Bayes-factor of the ATLAS diphoton excess

TL;DR

This paper calculates Bayes-factors to evaluate the plausibility of the ATLAS diphoton excess as a new resonance, providing a Bayesian perspective on the evolving experimental data.

Contribution

It introduces a Bayesian approach to assess the diphoton excess, avoiding issues with frequentist significance interpretation and updating the plausibility with new data.

Findings

Plausibility increased by about eight with early data.

Plausibility decreased to disfavor the excess after more data.

Bayesian analysis offers a different insight compared to traditional significance tests.

Abstract

We present a calculation of Bayes-factors for the digamma resonance ($\digamma$) versus the SM in light of ATLAS 8 TeV 20.3/fb, 13 TeV 3.2/fb and 13 TeV 15.4/fb data, sidestepping any difficulties in interpreting significances in frequentist statistics. We matched, wherever possible, parameterisations in the ATLAS analysis. We calculated that the plausibility of the $\digamma$ versus the Standard Model increased by about eight in light of the 8 TeV 20.3/fb and 13 TeV 3.2/fb ATLAS data, somewhat justifying interest in $\digamma$ models. All told, however, in light of 15.4/fb data, the $\digamma$ was disfavoured by about 0.7.