It's predictable - every primary season people obsess about national polls thinking they prove something. In 2007 I wrote something Nate Silver cited repeatedly, but I can see it is forgotten. This diary is shorter than the previous ones I have written on the subject - but here is the data.
In the last 7 competitive primaries, the leader in September has lost the nomination 5 times. In a sixth case, 1984, the leader went on to lose the lead only to regain it later in the process.
I know some people have trouble with this data. But there are two lessons two learn from a study of polling history:
1. At this point in the cycle voter preferences are just not very solid. We know from exit polls of Iowa and New Hampshire that many voters do not decide until the last week or month.
2. The national race often differs from the race in Iowa and New Hampshire for a very understandable reason: candidates are focusing time and attention there. Not a day goes by in New Hampshire or Iowa without a candidate visit. Television ads are running, ground organizations are contacting voters.
So the difference between national and Iowa and New Hampshire polling isn't a surprise.
The point is that Sanders is going to win or Clinton is going to win. If you look at GOP primaries the leader in September has won 4 of the last 5.
The point is that voters outside of New Hampshire and Iowa have not focused on the race yet.