I'll give my simple reminder that the best approach, from the observer's perspective, is the Occam approach. How many absurd assumptions do you have to make to conclude that Trump has never committed sexual assault? At this point, a ludicrous number.
As a general rule, any one accusation is more likely to be true than false, by simple probability. By simple probability, then, I would default to the notion that E. Jean Carroll is telling the truth, but in terms of legal processes, or even political processes, you can't just stop there. The fact that most assault allegations turn out to be true means that, in the case of assault allegations, the human making the statement is less likely to be lying than telling the truth, but the entity making a statement is still a human, and therefore cannot be treated the way we have mythologized George Washington, or Vulcans. (Minbari get a loophole.*)
So, any one accusation is probably true, but you should require more.
In this case, you have it. Without even going into Carroll's case, there are so many allegations against Trump, combined with his bragging on the grab-'em-by-the-pussy tape, his assertion that women should be treated "like shit," and numerous other public statements...
Here, then, is the calculation.
What is the probability that every allegation is false? The complication is that the allegations are not actually independent. They occur sequentially. That means they have the potential to influence each other. However, there are far more of them for Trump than for any normal politician, and they are backed by numerous public statements, and private statements compatible with exactly the kind of attitude one would expect.
As I have warned you, that "attitude" thing is hard to assess. Would you have expected Bill Cosby to be what he is? You can't always tell the monsters because they don't always present as monsters, and the most insidious monsters are the ones who act civil and polite, and all that. It's just that one of the characteristics of anti-social personality disorder is poor impulse control, which makes that difficult for the monsters who are straight-up sociopaths, like Trump. I doubt Cosby is a sociopath. He's just a horrible person. Not the same thing.
With Trump, though, to assume that he has never committed sexual assault, you would have to discount all of his rhetoric and every allegation. That's a lot to discount, and there is not a single thing in his favor that you have to discount to conclude that he's guilty.
This is yet another example of the statistician's reasoning versus the lawyer's reasoning. If E. Jean Carroll took this into a courtroom, would this count for anything for her? Not a hell of a lot. This is an assessment of Trump's patterns, and for any one accusation, without follow-up investigation, I could not move my Bayesian estimate from the baseline probability of a false accusation. It's just that the probability of Trump not being guilty of sexual assault is in truly absurd territory.
What we don't do in the courts, though, is convict people for having, at some point, committed a crime. It has to be a specific crime. You can't charge him with... well, we know you raped someone, we just don't know whom. Bayesian reasoning versus the law.
Still, the idea that he has never assaulted a woman has gotten to a point of statistical absurdity.
*The spell-checker was OK with "Vulcans," but not, "Minbari." Sorry, J. Michael Straczynksi. Oh, you don't like his name either?! Fine.