It's a social science; you can't exactly run experiments like they do in natural sciences. There are measurement errors & omitted variables errors you have to deal with in social science.
And yet this shortcoming seems to in no way inhibit the field's broad pronouncements and interjection of themselves into public policy. Emmisaries of far more rigorous and testable scientific fields comport themselves with a great deal of academic probity(e.g. Climatologists whose verifiable observations have amongst the most dire of consequences for us, and yet whom still are cautious in their framing). Meanwhile, economists like Rogoff and Reinhart never missed a chance to peddle their faulty math and data errors as God's Only Truth to every outlet available. Or buddy up to oppressive regimes like Friedman and his University of Chicago ilk. Or to peddle discredited, vaguely eugenicist studies like Levitt and Dubner for the sake of selling books.
This. The real problem with making definitive statements in economics is you can't run double-blind experiments.
Let's say you create some sort of giant monetary stimulus package. If the economy gets worse proponents will say "Whew! Just think how terrible things would have been without our stimulus package," and opponents will say "Things got worse because of the stimulus package!"
There's no way to know who's right. The best you can do is make imperfect models to encapsulate trillions of individual decisions people make. But that's not proof.