If y'all get a chance, a graph of Iowa vs. NH error would be intriguing as it would show who just got lucky by under/overestimating their favorite candidate consistently, who is actually good at predicting (in the context of two races of course), and who got Iowa wrong and then put too much stock in those results as race-changing.
Eh...2 states is a bit of small sample size. We could do it but I doubt it would reveal anything significant.
Obviously. No one can predict well with any degree of significance! This is for bragging rights which is even more important