Whither the polls?
One of the biggest stories of the NH primary is the failure of the polls to have accurately predicted the results. It’s not that it’s so unusual for polls to be incorrect, but for so many to be incorrect in exactly the same way—predicting about an 8% Obama victory and getting a result of Clinton by 3%, off by a factor of approximately 11%—is highly unusual.
There have been a host of attempts to explain; see this, this, and this, for example.
My own reflection is that Hillary’s emphasis on reaching the women of NH probably paid off, but perhaps late in the game. Her ads in NH were a media blitz on the female voters of the state, simplistic but relentless appeals in female voices only (young and old), stating how Hillary would help family and children and education. Just how she would be doing this as president was not specified, but apparently it did not need to be. “She’s a woman, we’re women, she understands”—that was the continual message, repeated ad nauseam on every radio station.
It didn’t reach the women I know, who uniformly seemed to like Obama and are disappointed in the primary results. And from my small and decidedly unrepresentative sample (I’m turning in my pollster badge), I still think Hillary is in a bit of trouble, because I know far too many loyal lifelong liberal Democrats who simply cannot stand her.
It was unfortunetly predictable the the “Bradley effect” thing would be brought up. Anything to “prove” the “structurally racist” nature of the US.
Obama’s Bolsters
Obama’s bolsters lied to the pollsters and confounded the media spin.
For they were not sure in there own hearts whether they wanted to win.
It was only in the back of their minds what they’d think of his spiritual guide,
an evasive zealot who seemed too much like a Dr. Jekyll or a Mr. Hyde!
raving lunatic clip (The End of the Line)
http://www.youtube.com/watch?v=HfNEfEBYIZs&eurl=http://sweetness-light.com/archive/a-message-of-hate-from-obamas-pastor-via-youtube
xcript here
http://sweetness-light.com/archive/a-message-of-hate-from-obamas-pastor-via-youtube
rational but evasive clip (Hannity interview)
http://www.youtube.com/watch?v=e8YmQUqj15g&feature=related
from yesterday’s Sweetness and Light:
“Though Wright and Obama do not often talk one-on-one often, the senator does check with his pastor before making any bold political moves.”
http://sweetness-light.com/archive/obamas-mentor-gives-farrakhan-his-award
“Obama’s connection to Wright first drew controversy in a February 2007 Rolling Stone article which described a speech in which Wright eloquently and forcefully spoke about racism against African-Americans.[4] Citing the article and fears that further controversy would harm the church, Obama scrapped plans of having Wright introduce him at his Presidential announcement”
http://en.wikipedia.org/wiki/Jeremiah_Wright#Relationship_with_Barack_Obama
Some will vote for a woman because
1) it is time for a woman in the White House
and
2) the Presidency will benefit from a woman’s touch
and
3) it is past time to deal a defeat to sexist men and sexist Conservatives.
Some will vote for Hillary because
A) they are part of a rational minority which is the only hope for America
and
B) it’s Hillary’s turn, and therefore she deserves it.
Polls are a statistic model of what the election will be – as such they are *not* the way one detects discrepancies. Is it unusual for them to be off that much in this way? Yes. Is it unheard of? No – in fact it isn’t even close to being something worth studying.
There are all sorts of reason – bad sampling is *really* easy even when done 100% honest. You will run into people not wanting to answer one way yet when in a private anonymous setting voting that way. And then of course you have all the other problems with intentional skewing (loaded questions and such).
Really, we are getting close to the point where many people are wanting the polls to elect somebody, not the actual election. If it wasn’t Diebold it would be the hand counts (the electronic machine was done because of complaints about hand counting). If you can count the actuall whole population (and we do in this case) then they trump the statistical models as the models are only meant to approximate them. And “margin of error” doesn’t mean what people think it means either – it is the margin of error from the full population that your sample represents which isn’t necessarily the full population of what you are studying (for instance, you could – though highly unlikely – randomly select 10,000 people from all over the US and have all them make under 10k a year and be on govt assistance – you will *not* get a worthwhile number out of that even though you were not stacking the books intentionally).
Polls?
The only ones that matter are in the polling place…
I’m fascinated by the psychology of those doubting the polls. They reject the actual event, which produces hard data in terms of vote count, and prefer to believe the flaky data generated by some guy with a clipboard asking a trifling percentage of the electorate chosen by God knows what method?
Perhaps neo could shed light on this. To me it seems like believing the fix was in to change the weather because yesterday’s forecast was not borne out.
I should clarify. Above “polls” referred to the election, not to pollsters’ results.
Sorry for the ambiguity.
i think i said somewhere that I doubt the accuracy of polls.. to be clear, i meant the opinion or the pre-election polls.. not the actual votes on voting day.