PHYSTAT informal review: Spurious signal method

by Aritra Banerjee (UMN), Oliver Rieger (Nikhef National institute for subatomic physics (NL))

Wednesday 4 Feb 2026, 15:30 → 17:00 Europe/Zurich

This is a PHYSTAT Informal Review event. Today, Oliver Rieger (physicist) and Aritra Banerjee (statistician) will review the topic "Spurious Signal method". The spurious signal method introduced by the ATLAS experiment quantifies potential biases in background modeling by fitting a signal-plus-background model to signal-free simulated or control data and treating any extracted “signal” as a systematic uncertainty on the background shape.

Agenda:

• 3.30 pm Opening:
• 3.30 pm Physicists Presentation (20'+10')
• 4 pm Statisticians Presentation (25'+10')
• 4.35 pm General Discussion and Closing (25')

 

PHYSTAT informal reviews: In this virtual format, a Tandem consisting of a physicist and a statistician will review a statistical method introduced by one of the parties or a general critical analysis topic from the Physicist's and Statistician's perspectives. The virtual events comprise: two 20+10 min. complementary presentations followed by ~30 minutes of general discussion.

 

Abstract: In many high-energy physics analyses, backgrounds are modeled using functional forms whose true underlying shapes are not fully known. Imperfect modeling can lead to artificial signal-like contributions, referred to as spurious signals, which bias the extracted signal yields and affect statistical interpretations. Spurious signal studies aim to quantify this bias and define appropriate systematic uncertainties on the parameter of interest. The first talk introduces spurious signals and reviews the main techniques for estimating them, including a toy example.
The spurious signal approach, however, naturally suffers from model misspecification. The second talk shows, via simulation studies, how even small deviations from the true background model can lead to false discoveries  in likelihood-based approaches. Furthermore, a novel methodology is introduced that enables valid inference on the signal intensity without requiring a perfect description of the background shape. The robustness of the proposed method to different background models is demonstrated through the analysis of toy datasets.

Literature:  https://cds.cern.ch/record/2743717/files/ATL-PHYS-PUB-2020-028.pdf?version=10

Aritra Banerjee.pdf

GMT20260204-144028_Recording_1920x1130.mp4

InformalReview_SpuriousSignal_ExpView_v0.pdf