ar X iv : g r - qc / 0 60 81 14 v 1 2 8 A ug 2 00 6 When is Enough Good Enough in Gravitational Wave Source Modeling ?

A typical approach to developing an analysis algorithm for analyzing gravitational wave data is to assume a particular waveform and use its characteristics to formulate a detection criteria. Once a detection has been made, the algorithm uses those same characteristics to tease out parameter estimates from a given data set. While an obvious starting point, such an approach is initiated by assuming a single, correct model for the waveform regardless of the signal strength, observation length, noise, etc. This paper introduces the method of Bayesian model selection as a way to select the most plausible waveform model from a set of models given the data and prior information. The discussion is done in the scientific context for the proposed Laser Interferometer Space Antenna.