CNN pollsters hard at work
So, the media's freaking out because they say the polls were terribly wrong in New Hampshire. But, really, they're paying about as much attention to the numbers now as they did before the primary.
I highly recommend reading, "New Hampshire: So What Happened?," by Mark Blumenthal, Editor/Publisher of Pollster.com. It's the best article I've come across on the subject.
If you look at the last polls before New Hampshire, and look at the actual results, and you take into account the "margin of error" that the polls always mention and everyone always ignores, the numbers for all of the candidates in each party were on target.
With the single exception of Hillary Clinton.
And I think that can be easily explained by the fact that Diebold machines were used for the voting and the Clintons hacked into the system.
Well, I'm kidding about the Clintons hacking the system. Although, some bloggers are shouting "Fraud!" And one blogger claims "Those Diebold op-scan machines are the exact same ones that were hacked in the HBO documentary, Hacking Democracy." And, now that I think about it, Hillary did compare herself to Lyndon Johnson the other day, though I still can't figure out why. She said that Martin Luther King Jr. may have had the dream, but it was President Johnson that brought about the change with the Voting Rights Act. Which is about the stupidest remark I've heard in the entire campaign. Why a woman who voted for the Iraq War would want to compare herself to LBJ, while comparing Obama to Dr. King is really beyond me, as much as I think LBJ was a better president than he's given credit for. But now that I've spoken poorly of Mrs. Clinton and planted seeds of doubt about her victory, what about the real story?
The question is: Why were her numbers so far off? Some last-minute swing? The choking-up thing? Women feeling sorry for her? Actually, several of the 3-day polls, if you look at them more closely, showed Hillary gaining significant ground the last day or so. But they were 3-day polls, so the numbers didn't reflect that momentum much.
Jeff brought up the problem of cell phones not being counted in polls, and that is a concern in general. I would argue, though, that by excluding cell phones, which are used more by younger people, the numbers - if the were really impacted - would have been better for Obama in the end, as many of his supporters would've been ignored before the primary.
New Hampshire had a huge amount of voters who didn't make up their minds until the last day - 17% for the Democrats and 19% for the Republicans. And 21% more of the Democrats didn't make up their minds until the last 3 days. So, 38% of the Democratic voters didn't know until the end. Hard to poll that.
I crunched some numbers, and McCain got 38,000 votes from people who made their minds up in the last day or last 3-days. Obama lost by 7,500 votes. A lot of Independents were torn between McCain and Obama, and I think Barack lost some votes that way, because people thought he was far enough ahead that it didn't matter. But again, his numbers were actually on target.
So, something last-minute got missed in Hillary's numbers. As time goes on, I'm sure more thorough explanations will be forthcoming.
Pollsters certainly aren't perfect, but a lot of the blame lies on the way the media covers the polls, and on us for not paying more attention ourselves. I'm not that worried about the polling. It is what it is. You wanna bet on horse racing using only speed ratings? Go ahead. But those numbers are only part of the picture.
What happened in New Hampshire is probably a good thing, because pollsters will have to be even more careful and attentive with their polling, reporters and pundits will actually have to look more closely at the numbers, and the rest of us will have to take these things with a bigger grain of salt. That's never bad.
Thursday, January 10, 2008
CNN pollsters hard at work