Anecdotal Evidence

[robvalue]

An untestable verbal account.

It's poor evidence because it's subject to all sorts of corruption, and there is no way to discover whether the events happened or not without further evidence. Even if they did, the accuracy cannot be determined. There's plenty of good explanations why several people, even large groups of people, would report the same apparently extraordinary event. And they're all more likely than "it was something magic", in my opinion.

If anyone believes that something as-yet undemonstrated to even be possible has happened based on anecdotal evidence alone, they have made a serious error. It may be a good starting point for investigation. It cannot be the investigation. You're granting these witnesses the power to correctly identify and categorise an unknown phenomenon.

Attempts to try to reduce all evidence down to the level of anecdotes is the last ditch tactic of someone desperately resorting to tu quoque fallacies. Attempts to conflate the mundane (Jack killed him) with the extraordinary (a ghost killed him) are along the same lines. Your evidence doesn't get any better by talking about it, nor by talking about other evidence. If someone has seen your evidence and is unconvinced, you need to try something else to convince them; not to try to tell them that they should be convinced.

As a last point: compare what happens when you hear an anecdote that you would like to be true, versus one you wouldn't like to be true. Do you treat both equally? Would you concede an (extraordinary) point you don't want to be true on the basis of piles of anecdotal evidence?