On the interpretation of the Fermi GBM transient observed in coincidence with LIGO Gravitational Wave Event GW150914

Abstract: The weak transient detected by the Fermi Gamma-ray Burst Monitor (GBM) 0.4 s after GW150914 has generated much speculation regarding its possible association with the black-hole binary merger. Investigation of the GBM data by Connaughton et al. (2016) revealed a source location consistent with GW150914 and a spectrum consistent with a weak, short Gamma-Ray Burst. Greiner et al. (2016) present an alternative technique for fitting background-limited data in the low-count regime, and call into question the spectral analysis and the significance of the detection of GW150914-GBM presented in Connaughton et al. (2016). The spectral analysis of Connaughton et al. (2016) is not subject to the limitations of the low-count regime noted by Greiner et al. (2016). We find Greiner et al. (2016) used an inconsistent source position and did not follow the steps taken in Connaughton et al. (2016) to mitigate the statistical shortcomings of their software when analyzing this weak event. We use the approach of Greiner et al. (2016) to verify that our original spectral analysis is not biased. The detection significance of GW150914-GBM is established empirically, with a False Alarm Rate (FAR) of $\sim 10^{-4}$~Hz. A post-trials False Alarm Probability (FAP) of $2.2 \times 10^{-3}$ ($2.9 \sigma$) of this transient being associated with GW150914 is based on the proximity in time to the GW event of a transient with that FAR. The FAR and the FAP are unaffected by the spectral analysis that is the focus of Greiner et al. (2016).