Interesting: The Comeback That Never Was

I realize the media narrative will be all about Hillary's incredible comeback in New Hampshire tonight. But there's something worth checking out with the history of RCP polls. Yes, all of the well-known national pollsters had it terribly wrong. But look at the local results over the last week.

The results were there all along, it just didn't generate any buzz.

                                                                Obama      Hillary

Suffolk/WHDH       01/05 - 01/06 500 LV              35           34

Concord Monitor   01/04 - 01/05 400 LV              34           33

CNN/WMUR/UNH    01/04 - 01/05 359 LV               33          33

Suffolk/WHDH      01/04 - 01/05 500 LV                33         35

Reuters/CSpan/Zogby 01/03 - 01/05 844 LV          30         31

Mason-Dixon 01/02 - 01/04 600 LV                      33          31

CNN/WMUR/UNH 12/27 - 12/30 521 LV                30          34

Comments

[Jim Addison]

Is it just me, or do all these polls have something in common? That being the latest of them ended on Saturday the 5th, only two days after Obama's big Iowa victory. One quoted poll ended December 30, well before the Iowa Caucuses.

So, many polls taken without accounting for the full Iowa bounce for Obama showed Hillary close or ahead slightly (when she had enjoyed a commanding lead through November, double digits in most polls). Why are later polls excluded? Because they ended up being wrong?

There is plenty of room to criticize the polls, but to say they predicted the result based on their OLD data is a big stretch.

[Dan Riehl]

No, the top one ended on the 6th. Rasmussen tracking also had it at one point the day of the election. All I'm pointing out is that the polls people thought were the outliers were actually far more correct. Their methodologies might help to explain what happened with the modeling with the others, if that's what it was.