In view of the recent report by Connaughton et al., we analyze continuous time-tagged event (TTE) data of Fermi-gamma-ray burst monitor (GBM) around the time of the gravitational-wave event GW 150914. We find that after proper accounting for low-count statistics, the GBM transient event at 0.4 s after GW 150914 is likely not due to an astrophysical source, but consistent with a background fluctuation, removing the tension between the INTEGRAL/ACS non-detection and GBM. Additionally, reanalysis of other short GRBs shows that without proper statistical modeling the fluence of faint events is over-predicted, as verified for some joint GBM-ACS detections of short GRBs. We detail the statistical procedure to correct these biases. As a result, faint short GRBs, verified by ACS detections, with significances in the broadband light curve even smaller than that of the GBM-GW150914 event are recovered as proper non-zero source, while the GBM-GW150914 event is consistent with zero fluence.