To a large extent, it's a judgement call. I tend to look for closely spaced decodes, vs a total count, which is why I often post the timestamps of unusual catches. My thinking is this: False decodes are corrupted messages, several of the bits of a message are altered such that the message still passes the checksum. It is essentially a random process, and this means the bogus decodes should be spread over time. DX openings are typically at least a few seconds, if not minutes long, so you should be able to get a few messages in a row, or at least closely spaced, when there is a legitimate opening. The more messages, the more confident I am that the decodes are real. I also consider whether or not there's a path of darkness between me and the station in question. If it is around noon at the DX station, or here, then it is not very likely that it is real (except for the US and Canadian stations that I expect to receive during the daytime).