Like Sheldon and several others here, I'm interested in constructing a rigorous view of what constitutes evidence. Though of course, one doesn't have to stray far into the realms of analytic philosophy, before realising that this is another of those famous Hard Questions™.

But, in the interest of providing the regulars with useful discoursive ordnance, so to speak, I'll start the ball rolling with a suggestion. Namely, that evidence can be defined as a suitable body of relevant data, that imposes constraints upon the likely truth-value of postulates under test.

I consider this suggested definition useful as a starting point, because it immediately allows the diligent to deal at source with mere fabrication, in tandem with that central axiom of proper discourse, that assertions do not constitute data for this purpose.

Of course, if anyone can provide me with advances on the above, this will be welcome, especially if those advances fortuitously approach the standard set by Willard Van Ormand Quine in his seminal work Two Dogmas Of Empiricism, which, despite the title, was actually a critique of logical positivism.

I suspect Sheldon and several other regulars here will be most interested in taking this thread further.