Obama's negative New Hampshire trendline

CNN released a new poll of the NH Democratic primary: It has Clinton leading by a huge margin against Obama. She has 43% against 20%. Edwards comes in a distant third at 12% and Richardson is fourth at 6%. This is quite a change from CNN's July poll, in which Clinton only had a 9% lead (36% to 27%) and in which Richardson had crossed the double-digit threshold at 11%.

These numbers are similar to most New Hampshire polls we have seen over the past few weeks. In fact, the Pollster.com average of the past few surveys put Clinton at 37% and Obama far behind at 17%. This kind of lead is what we are used to seeing in national primary numbers. And the kind of lead we are prone to dismissing in a heartbeat, for all the classic reasons: national polls are irrelevant, they mostly test name recognition, etc.

But New Hampshire polls in no way fall into that category, and it would be a great mistake to dismiss them. Candidates have been campaigning in New Hampshire in months and most of them have started running ads there. With Obama usually coming in third position in Iowa for now, he cannot count on a boost there and needs to be very strong from the get-go in New Hampshire. He has been in full swing in the state and his name recognition there is high: Only 3% of respondents in the CNN survey said they did not know enough about Obama to form an opinion about him (by contrast, 22% did not know anything about Richardson).

The Obama campaign believed that their candidate was tailor-made for New Hampshire, which is known for backing anti-establishment independent candidates who buck conventional wisdom with stunning wins (Hart over Mondale in 84, McCain over Bush in 2000...). Despite this, despite his name-recognition increase, and despite the ads he has been running, he lost significant ground against Clinton in the CNN poll.

This negative trendline is confirmed by the other polls taken in the state, as compiled by Pollster.com. Obama has not been within 15% of Clinton in New Hampshire in all surveys taken since late August. In the 4 September polls taken by 4 different polling institutes, Clinton's lead is 23, 23, 18 and 19 points. By contrast, Obama ran much closer to Clinton in many surveys taken in June-July: He was tied at 31% in ARG's July numbers (Arg's latest poll has Clinton leading by 20%, with Obama crumbling by 14%) and trailed by low single-digits numerous times. Take a look at Pollster's handy chart:


Obama's bad poll results are no longer easily explained away by Clinton's inherent recognition advantage, not after months of campaign in the state. And even if Clinton had maintained a strong name-ID advantage, this would not explain why Obama is on a stark downward trend. This phenomenon is also observed (to a lesser extent) in South Carolina, where Obama used to sometimes edge out Clinton earlier in the year but where Clinton now usually gets a lead.