The expected problem is if the snow falls during the school day and the commute home from school, things can get incredibly snarled and kids don't get home until really late. Attempts to maneuver around this by having a half day or 1 or 2 our early release (2 hour release and half day being functionally equivalent in many cases) tend to work very poorly. Any kids in x-day tend to have parents who will fail to pick them up at the early release time (x-day will be canceled in the event of early release), thus requiring staff to hang around, not just until the usual time, but the end of usual x-day, and you're still up against the snarl problem. In really bad situations, staff wind up spending the night at the school with the kids, which can work, but in the event of a medical situation and/or power outage can get very sketchy.
If the edge of a storm is near a school district, and one or more surrounding towns will be hit hard by that storm, then a school district needs to take into consideration the ability of its employees to get to and from school safely -- and whether they will have kids at home all day due to a snow day in their home district. In that event, cascading cancellations by personnel rapidly overwhelm the ability of substitutes to cover -- and substitutes also live in surrounding towns and may be subject to the same dynamic.
Thus, the correct way to predict whether a school will close for a snow day is as follows:
Obvs, if there is a high probability of a storm in town, then school will be canceled.
But school will also be canceled in the event that there is a high probability of a storm in towns where school staff live AND/OR towns where school staff live have already canceled.
There is another factor, which is how many snow days have already been used. If few or no snow days have been used AND we are already well past January 1, THEN it is a lot easier to call a snow day if one or the other of those is not true (hard to snow day in November; hard to snow day in December when you've already missed a week due to weather).
Finally, there is a trust factor. When school has been canceled in advance of the weather becoming visible (that is, trusting in the forecast) AND THE SNOW WAS CATASTROPHICALLY BAD then everyone will nod and go, okay, these people know what they are doing, and they will be much more willing to close next time. When school has been canceled in advance of the weather and it is a bust, then everyone will shake their heads and go, okay, these fools don't know what they are doing. Imma make up my own mind next time.
It should be apparent from this discussion that the following are not actually important in canceling school.
(1) What schools did in previous decades, when you were a kid. Or when your kids were young.
(2) What schools do in other states.
(3) What the weather model says.
Yes, I said it. What the weather model says is very secondary to other factors. Which is why I'm a lot better at predicting school closure than I am at predicting the weather, and my husband is the reverse. He's been expecting a bust up here [ETA: Apparently his definition of a bust includes up to 3 inches.] since I first started paying attention yesterday. We shall see what happens -- it might still mess up the roads before end-of-school day.
ETA: Went for a walk finally around 10 a.m. It is windy and thus feels wicked cold. It's in the 20s, so it shouldn't feel that bad, but the snow is currently sideways which is a Sign. Oh, and it is snowing now. It'll be difficult to measure accumulation unless the wind stops because we'll wind up with bare ground and drifts.