I'll try to answer this with three points about the scientific method and how "certain" we are of the truth in our theories. Keep in mind that scientists are overly dogmatic about pet theories but we should aspire to transparency about how wrong we might be and distrust everything until the evidence, be it scant or ample, is verified.
First, you can gather quite a lot of insight by listening to Richard Feynman's analogy between discovering the laws of nature and learning the rules of chess through observation of a fraction of the board. In particular, there's the part where he talks about a bishop changing it's colour despite ample observations of this never happening. His overall point is that we're never truly sure but we are always inadvertedly gathering evidence that the theory is right.
Secondly, you should read Isaac Asimov's essay The Relativity of Wrong. His point is that while a theory might be "wrong", sometimes they're very wrong ("the Earth is flat") but sometimes less wrong ("the Earth is a sphere"). In some cases, you can quanitify this. For a contemporary example, cosmologists have settled on $\lambda$CDM as the right model of the Universe. The point isn't that $\lambda$CDM is necessarily the whole story but that, if it isn't, then the evidence we've gathered already implies that the whole story can't be much different.
Finally, let's think back to the superluminal neutrino fanfare. It made big news, with the media painting a picture that made it look like the scientific community needed to revolutionize special relativity (SR). But a lot of scientists responded skeptically, even by offering to eat their shorts. So why the skepticism? Surely that flies against the scientific mantra of doubting authority?
Not quite. There were good reasons to doubt the result and anyone who dismissed those results should've defended their position. It was quickly pointed out that, if neutrinos travelled faster than light, we'd detect supernovae early. Also, I think Glashow and others pointed out that we'd see something like Cerenkov radiation from the neutrinos.
But more importantly, SR is, to me, a theory that is close to being "certain". It was and still is tried and tested extensively and it forms the basis of other theories that are themselves successful. So the odds of SR being "wrong" are outrageously small. We have inadvertedly tested it bazillions of times and it's worked perfectly. And the amount by which it can be wrong is very small. At the time, it could've been like the first time a pawn was queened into a bishop, but, to roll out the cliche, extraordinary claims require extraordinary evidence.