Data-Driven Background Modeling for Precision Studies of the Higgs Boson and Searches for New Physics with the CMS Experiment

The Large Hadron Collider (LHC) at CERN has produced an unprecedented amount of high-energy proton-proton collisions during the years 2016 to 2018. Data recorded and certified by the Compact Muon Solenoid (CMS) experiment from this time period amount to $137 \, \mathrm{fb}^{-1}$ and are analyzed in this thesis. Collision data are used to perform precision tests of the Standard Model (SM) of particle physics and to search for new phenomena beyond this model. The presented precision measurement focuses on the SM Higgs boson decaying into a pair of tau leptons. Moreover, all presented searches for new particles target their decay into leptons. Proton-proton collisions, however, lead to abundant particle formation governed by the strong interaction. Particle misidentifications in collisions with such numerously produced particles are unavoidable and pose a background in measurement and search regions alike that needs to be estimated. Modeling particle misidentification precisely by means of simulation turns out to be difficult. Thus, dedicated data-driven methods are developed to assess the amount of misidentification entering measurement and search regions. In this thesis, such data-driven methods are presented and their application to several analyses is demonstrated. For the SM Higgs boson decay into two tau leptons, a measurement of signal strength modifiers, quantifying the signal size relative to the respective SM expectation in up to twelve kinematic regions, is presented. Results of more inclusive signal strengths yield $0.81^{+0.17}_{-0.16}$ and $0.67^{+0.20}_{-0.18}$ for Higgs boson production via vector boson fusion and gluon-gluon fusion, respectively. Furthermore, two searches for additional Higgs bosons, decaying into a pair of tau leptons, are presented in this thesis. No excess over the SM expectation is observed and model-independent upper limits on the production cross section times branching ratio to tau pairs are set in both cases. In addition, a search for third-generation leptoquarks in a di-tau final state is presented. All these analyses adopt the data-driven background estimates shown in this thesis to evaluate contributions from misidentified leptons. Lastly, a search for pair-produced light top squarks is reviewed in the context of applying a newly developed data-driven background estimation technique.