I use metaphysics in a much more specific meaning than its vague modern-day meaning of “beyond the physical world”(God, soul, free will, etc.), and rather that a metaphysic is any relation. If you look at the history of the concept of metaphysics one of its subdisciplines is Natural Philosophy, which became what we now know of as science. All fields of rational study, as well as studies beyond the physical sciences or what would be traditionally considered science, would qualify as metaphysics in that it explains relationships between physical/objective events, as the relationships themselves are not physical/objective but rather all-encompassing universals beyond those specific objects themselves. That would allow “soft sciences”, though not as rigid as physical sciences in their ability to disconfirm, to be qualified as in the same ballpark of metaphysics as physical sciences are, that is relations and explanations of observable events(just human behavior, which is harder to make disconfirmations from especially with such vague theories).
A scientist will, and should, of course, go with the theory that provides the most truth content and least amount of false content, or have the best verisimilitude, in the given field or topic that she is looking to find explanations. Why a scientist would choose to stick with a theory lacking verisimilitude is simply a lack of open-mindedness on her part, and essentially, if you look at it evolutionarily, she, and her ideas, will become subject to a sort of intellectual natural selection, since her theories, lacking truth content, will die out.
This is also how this kind of epistemology can apply not only to science, but to normal life and even non-human life. All living creatures make predictions, that is to make theories, and in the case of animals often the failure of their theories/expectations means their death. In the case of humans and human abstraction, our expectations can often be pursued without our death or even negative consequence, and often we tend to operate on a certain theory/expectation until it fails us. An example might be that my expectation when trying to get a taxi is to lift my hand when a taxi is coming by, but if my expectation is repeatedly not met, that is disconfirmed, I will then form a new theory/expectation in order to accomplish my goal. Some expectations can be biologically built-in, some are cultural or “common sense”, etc.
In the case of science, we make a theory to explain observed phenomena, and from that theory eventually new problems will arise. The theory, as it exists, is then disconfirmed, and so is either modified or an entirely new theory is formed.
The general formula for this process is given like so:
Problem 1 -> Tentative Theory -> Error Elimination(resolving internal contradictions) -> Problem 2
No explanation is possibly absolute, as new problems arise from new theories. The theory that is preferred is the one that is the most specific in terms of its predictions and has the least falsity.
The reason all of this is deductive rather than inductive is that it’s all about prediction rather than making general claims. Inductive science would operate by making observations and generalizing them into “laws” or absolutes, many of which then fail since the future by no means has any necessity to reflect the past. Popper avoids any claims of absolutes, and says rather that the goal is to make predictions rather than absolutes, and so when the events come to pass we can deduce whether the prediction/theory is false or not; hence the goal, and only clear qualification, of falsification, and a best current answer or best current prediction. It is still deductive, since we’re not making inductive absolute claims, and instead are making future-predictions that we can then deduce to be false or not-yet-false.