Entertainment is unusually rich in facts that everyone knows and nobody can source. Production anecdotes, box office records, casting near-misses and on-set incidents circulate for decades in a form that resists checking, because the interesting ones are precisely those with no document behind them. A short method separates what can be established from what merely repeats.
Establish what kind of claim it is
Claims about the work — release date, credits, running time, certification — have documents behind them somewhere: registrations, classification records, trade announcements, contemporary listings. These are checkable, and if a check fails the claim is probably wrong.
Claims about what happened during production are different. They usually rest on someone's recollection, given years later in an interview. The best available evidence may be a single account, and a single account is not corroboration however often it is reprinted. Treating recollection as documentation is the most common error in writing about entertainment.
Find the earliest telling
Work backwards to the earliest version of the story you can locate. Anecdotes accumulate detail as they travel: names become more specific, figures become rounder, dialogue appears that was not in the original account. Comparing the earliest telling with the current one shows how much of the story is testimony and how much is transmission.
Where the earliest telling is an interview, note who was speaking, when, and what they were promoting. None of that makes an account untrue. It sets the appropriate confidence.
Watch for numbers that never had a source
Budget and box office figures are reported selectively and by interested parties. Production budgets are frequently estimates that were never confirmed; gross figures may or may not include re-releases, may be domestic or worldwide, and are usually quoted without adjustment for inflation even when comparing decades. A figure repeated confidently in many places is often one figure, repeated.
Prefer contemporary sources
A review published in the week of release is evidence of how a film was received then. A retrospective written thirty years later is evidence of how it is regarded now, and the two are regularly confused. Films described as flops that were reappraised later — and the reverse — usually acquire that description from retrospective writing rather than from anything recorded at the time.
Record what you could not confirm
The useful discipline is to note the difference between "confirmed", "single source" and "widely repeated, source unknown", and to keep those labels attached when passing anything on. Most of what circulates falls into the third category. Saying so costs nothing and stops the claim from hardening one degree further every time it is repeated.