To put it another way, according to 538's polls+ model, we would expect this particular grouping of states to be ~3-4 points more Republican than the national average (I mean, it includes states like Georgia which are a good ~10 points more Republican than the national average, but doesn't include states like New Jersey or Washington to balance it out). So because the result here is Clinton +6, that corresponds to the national margin being about Clinton +9 or 10.
Exactly, and well put.
Does the battleground tracker generally weight each state according to its population, or is it a roughly equal sample size from each state? I could not find a clear answer to this from the description.
I assumed the former, but if I'm wrong, that would obviously make a difference in my calculation above. That being said, the difference would be relatively small, since a simple mean (not weighted by population) of my margins gives 6.4%, close to the 6% margin reported.