As someone who spent 7 years in the education system as a teacher and where everyone (even the Superintendent) was close with everyone else, I can say that the schools (at least in our area) collaborated with EMS and NWS offices to decide what to do.
I'm in the field of Statistics, and there's much to be said about Type I and Type II errors (you can look it up if you want, I won't bore you with the details). The point is that it's impossible to make a perfect forecast -- some will overperform and injure/kill/destroy way more than forecast, and some will underperform and leave most people safely. As a school administrator, you're tasked with keeping as many people safe as possible (granted, some children live in unsafe places during tornadoes but *that is out of the school's control*), so the decision is essentially this: How do you thread the needle where you minimize the student/teacher/staff injuries/deaths while at the same time maximizing the learning days in the classroom? You can put this another way: How many student/teacher/staff injuries/deaths are *potentially* acceptable in order to make room for more learning days in the classroom? Tornadoes, for instance, have a tiny probability of striking any given school, even on the worst severe weather days...but the possibility, that worst case scenario, is what administrators have to weigh. Looked at from a mathematical/probabilistic point of view, schools should probably never be closed -- but if any life in their hands could be put in danger, most administrators today opt for the safest approach and canceling school. I think this is the most prudent approach. No amount of learning is worth a life, in my view.
Also note that in many places, unless the governor has declared a SoE, schools are required to make up the days by law. So in a lot of places, they aren't "losing" days, just shuffling them around.