The Swedish material is the clearest European demonstration of a general point: witch panics that ran on child testimony were the fastest to grow and the fastest to collapse once anyone examined the testimony carefully.
They grew fast because a child asked a leading question by a frightened adult in authority supplies names, and because a community will believe a child before it believes an accused neighbour. They collapsed because the same testimony, once compared statement against statement, contradicts itself immediately. Blåkulla accounts placed the same people in different places on the same night, and named the dead and the absent.
That is what happened. Investigations ordered from Stockholm in 1676 examined the children's evidence rather than counting it, found it worthless, and the commissions were wound up. The pattern repeats elsewhere with the same shape: the Basque panic of 1609 to 1614 also ran on child testimony and was also ended by an official who tested it rather than tallied it, and Salem's afflicted accusers were mostly young.
The uncomfortable corollary is that nothing about the belief changed in any of these cases. Sweden did not stop executing because it decided witches were impossible. It stopped because a central authority looked closely at the evidence and found it did not stand up, which is a procedural conclusion rather than a philosophical one, and it saved lives immediately.