Doing research requires exploring a tree of different ideas and then, upon failures, tracking back to some extent, up to giving up on the whole project and changing topics (or even quitting your PhD). You might err on both sides: change approach too much, give up too early, or too late. (See for instance this answer or Half good and some not good results in a research paper?). Most importantly, though, often you won't know that you'll actually succeed until you did, and things might look bleak until then.
Does some degree of stubbornness help being a researcher (I couldn't extract an answer so easily from 2)? Lacking that, do you know any metaheuristics to approach this decision?
EDIT: I read this idea off the mention of "stubbornness and self-delusion" in this rant - and I've observed this trait in at least some researchers.
EDIT 2: an answer suggested that I talk about persistence instead. And probably that's the right compromise and what you actually should have. But I prefer the more provocative phrasing, also because I've anecdotical experiences of stubbornness as a "professional risk" of the profession.