Assigning confidence to inspiral gravitational wave candidates with Bayesian model selection

Bayesian model selection provides a powerful and mathematically transparent framework to tackle hypothesis testing, such as detection tests of gravitational waves emitted during the coalescence of binary systems using ground-based laser interferometers. Although its implementation is computationally intensive, we have developed an efficient probabilistic algorithm based on a technique known as nested sampling that makes Bayesian model selection applicable to follow-up studies of candidate signals produced by on-going searches of inspiralling compact binaries. We discuss the performance of this approach, in terms of false alarm rate and detection probability of restricted second post-Newtonian inspiral waveforms from non-spinning compact objects in binary systems. The results confirm that this approach is a viable tool for detection tests in current searches for gravitational wave signals.