As Dalecoy implies, a fixed offset will not work terribly well if the actual inside temperature varies over more than a fairly narrow range, since the actual offset shows some absolute temperature dependance. More seriously, the offset caused by the USB data logger seems to be very variable. It certainly depends on whether (= is only a problem when) the USB cable is connected to a 'live' computer, and seems also to depend, though much less critically, on the 'traffic' passing through the logger.