My stance is that whether or not an entity is real is not a meta-level philosophical question, the way it is usually treated in the realism vs. anti-realism debates, but an object-level scientific question. Our scientific theories, interpreted literally, are committed to certain ontologies. MWI is quite clearly committed to the existence of something like a universal wave function, otherwise none of its explanations or purported merits make any sense. I also think (like David Wallace) that basic quantum theory, interpreted literally and without any ontological or epistemological add-ons, is committed to something like a universal wave function.

So I think that our best scientific theories should determine our ontology. As for which scientific theory to believe in, I think there are a number of different considerations that go into that -- empirical confirmation, simplicity, concilience with the rest of our theories, feritility (in terms of useful predictions), etc.

My beef with instrumentalism is that it is, in a sense, too philosophical. It treats the reality or unreality of the unobservable entities posited by our best theories as a further question, one not determined by the theories themselves. Even once we have accepted, say, quantum mechanics as true, the instrumentalist says there is a further question about whether to take its claims literally or to treat them as mere calculational tools for making predictions. My take is that there is no further question. Accepting quantum mechanics as true, or believing in quantum mechanics, implies accepting the theory as a guide to reality. If you're unwilling to do that, you need to tell me why. If your standards for determining what really exists are so high that not even our best-confirmed scientific theories can meet them, then I suspect you're working with a concept of "reality" that I, as a pragmatist, have no use for (and can't even really understand).