“How do we know what we know” is an interesting challenge— not all claims to knowledge have equal value. While anecdotal evidence is still a kind of evidence, it’s not the same as reproducible laboratory evidence with a hard outcome, and crystal clear methodology.

So, there’s different kinds of evidence, with the burden of proof entirely upon the claimant. What that means is if you’re making a claim that goes against the established knowledge or wisdom (whatever is accepted as being already proven, that’s already met the burden of proof) — if you’re trying to change that, you have to at least equal that. In other words, not all propositions are equal unless you can balance out with equal weight of new evidence. 

There’s many things that can influence our perceptions of events and even what we might think is good evidence — our background knowledge can influence our perceptions — including prior knowledge, expectations or cultural upbringing. The words we use can influence how we’re interpreted by others, such as something being lost in translation; and naturally, we can’t exhaustively test everything all the time. 

Sample sizes can vary and have an influence on the final results — and replication could also change in different environments — such as the boiling point of oil in space or on another planet. The important thing is that we know about these kinds of limitations and factor them in, while still working towards understanding.