Tuesday, January 08, 2008

Granite State Predictions

For today's New Hampshire primary predictions, AP Government decided to take a shot at predicting vote percentages for the candidates. So, in addition to ranking the final showings, we're giving out numbers.

Live by the numbers or die by the numbers, we'll put ourselves out there.

For the Democrats:
Obama: 36 – 37%
Clinton: 28 – 32%
Edwards: 18 – 20%
Richardson: 4%

For the Republicans:
McCain: 33 – 37%
Romney: 24 – 26%
Huckabee: 10 – 12%
Paul: 8 – 10%

Obviously, the 35% of Granite State voters who identify as Independents will be the big story today. Will they go for McCain or Obama? But we also wonder if Hillary's unexpectedly emotional appearance yesterday will help or hurt her prospects going forward. By a narrow margin, we decided it will help.

If we are right about the Democrats, then the big news out of the Granite State is that Obama is looking good to go. Inevitability is no longer the Clinton claim. And Edwards must now deliver the goods in either Nevada or South Carolina if he is to hold on until February 5th.

And on the GOP side of the fence, McCain has clearly benefited from history (he won this state in 2000) and the abundance of editorial endorsements that he received. Romney will be under pressure to deliver in South Carolina, where he can expect Huckabee to show strong. And Giuliani, who was actually organized in New Hampshire, is looking weaker by the day.

Morning After Update: Wea Culpa
Okay, then. Let's start with the obvious: we missed the impending Clinton surge. We take some comfort in the fact that we weren't alone.

Clearly the historical pride that Granite State voters take in being independent-minded is not mis-placed. Suffice it to say that New Hampshire has been good to the Clinton family. Hillary Clinton's victory in New Hampshire was not a big margin, but given that nearly every poll in the state had her down by 10 points going into the vote on Tuesday, it is suitably impressive. And she has spent the last few days of her campaign speaking more informally, taking impromptu questions from both the press and the public. She's good at that and should have been doing it more often; last night it paid off. We expect more of that in the coming weeks.

The final Democratic numbers from New Hampshire:
Clinton: 39%
Obama: 37%
Edwards: 17%
Richardson: 5%

Our analysis of the GOP vote last night was more accurate. McCain came in at 37%, which was our upper estimate. We figured Romney in the 20s and he came in at 32%; clearly we didn't give him enough credit. At 11%, Huckabee was just where we expected him to be. Giuliani edged out Paul with 9%.......but that won't be enough to keep Rudy's campaign alive.

Democrat John Edwards is in a tricky spot and has a huge uphill battle ahead of him, but he said something last night that resonates this morning: 99% of the nation has yet to vote in a primary contest. A good portion of those folks (more than 50%) will have an opportunity in the coming weeks, especially on February 5th. And it would seem that in both parties we have a real contest on our hands.

It isn't over yet.

Update II: JBro has a nice description of the polling process in the comments section here ..... worth your time. And I share his view of pollster.com as reliable source of polling info; I use it nearly every week in class. Short answer: polling is typically impressively reliable. Except when it's not.


J.Bro said...

How are you students going to deal with being so massively, massively wrong this morning? Are you going to have a conversation about double-standards and all the women that supposedly broke for Clinton because everyone was so mean after she cried? Will you just fail them all?

sister AE said...

It does make me wonder just how accurate some of the polls are, but I don't have the patience (or consistent interest) to trace their historical accuracies.

Specifically, do most of the polls work by phoning people? If so, where are people phoned? If more people are phoned at home, does that skew the answers by who is home and who is out running errands? If more than one voter is in the home, do they ALL get polled? or just the one who answers the phone?

Thanks for letting me vent. I'll try to take a breath now and get on with my boring afternoon at work, where I'm not even sitting at my regular desk and where I have NEVER been polled for what I think about a political candidate.

J.Bro said...

One of the best polling websites on the market (which happens to be run by the person who taught most of my graduate research methods courses) is pollster.com. It's a poll-aggregator (meaning it has purty graphs that collect lots of polling data), but it also has excellent analysis in the blog section.

What it lacks, though, is a good intro-to-polling FAQ (which I think it needs, since more and more people are going to be asking questions after NH's primary). There's a huge literature about sampling methods and question wording and how to handle non-respondents, but the short version is that polling outfits randomly* choose about 1250 people to call. They ask the opinion of the first person to answer over 18 - although sometimes they will choose another home if that person doesn't describe themselves as a "likely voter". Those results are tabulated, and the sample size is big enough to extrapolate the results to millions and millions of folks. It's shockingly accurate, except when the results are influenced by something that the pollster can't incorporate (such as, for example, thousands of previously-undecided voters all breaking the same direction instead of splitting proportionally).

Ah, it looks like Gallup has a really great FAQ about the whole process:


J.Bro said...

*Randomly here may mean the pollster staying, "find me a random sample with 42 latino women in the $40,000-50,000 income range, 110 first-time voters, 400 voters over 65, etc, etc, etc"

Sharkb said...

This is all quite interesting, I usually go for the candidate with the least annoying voice.