BTW, Nate’s model is very very sensitive to Iowa. Changing it from one value-state to another seems to impact quite a bit.