Yesterday I talked about how off the pollsters were in Iowa last week. Did they redeem themselves in New Hampshire? By and large, yes.
Of the five major polls I was watching, all five got the top two correct. One of the outlets — the Marist Institute poll — was almost perfect, predicting a 13-point victory for Kerry over Dean, showing a virtual tie between Clark and Edwards for third place, and predicting a fifth-place finish for Lieberman with 9% — all of which happened exactly.
The controversial Zogby poll showed Dean a lot closer than he actually was. Zogby had Dean trailing by three points on Monday and seven points on Tuesday. Zogby also appears to have underestimated Edwards’ support a bit, predicting Edwards was tied with Lieberman for fifth place.
Gallup came awfully close to getting New Hampshire exactly right. It showed Kerry with an 11-point lead (which was just two points off), Dean in second with 25% (only one point off), and Clark in third with 13% (which was exactly right). The only flaw was, like Zogby, Gallup showed Edwards tied with Lieberman for fifth.
American Research Group was also pretty close, showing Kerry with a 10-point lead, Edwards and Clark about tied, and Lieberman trailing the field. Kerry did a little better than expected, and Edwards a little worse, but on whole, it was a solid result for ARG.
The Boston Globe/WBZ poll did the worst, but even it did pretty well. This poll showed Kerry with a 17-point lead, which turned out to be a little too big. The Globe poll’s biggest mistake was understating Clark’s support — it showed him with 8% when he finished third with 13%.
Overall, I’d give the pollsters an A-. Of course, the two polling outfits that did the worst in Iowa — Survey USA and Research 2000 — were hardly polling in New Hampshire at all. They’ll have a chance to redeem themselves (or perhaps look even worse) next Tuesday.