You've touched on another one of the ingredients to the misunderstandings here.
Science works by assuming "equal a priori probabilities" that is, when entertaining hypotheses, one should evince no bias in advance.
Now this is useful when one is trying to keep one's ear to the ground in order to not throw away finely-grained, but actually significant ("signifying" some effect), experimental effects.
But the problem is that this over time is taken by the scientist as some sort of necessary *indication* that "the Universe ITSELF" is like that.
But correlation is not causation; and absence of evidence is not (sufficient) evidence of absence -- merely an indication.
It may hold STRONGER in the natural world, if the assumption of fixed natural laws is true: the more rigorously a law is demonstrated, and over a wider variety of conditions, the more one feels ("gains assurance") it is likely to be true; but if dealing with an explicitly non-corporeal entity, moreover one possessed of independent will and purpose, one can never guarantee when or why it will act --
at least not by experment; such is only going to happen by personal acqauintance.